QM Lecture Notes PDF

Eyal Buks
Quantum Mechanics - Lecture

Notes
November 4, 2014
Technion
Preface
The dynamics of a quantum system is governed by the celebrated Schrdinger

equation
d
| = H | ,
(0.1)
dt
where i = 1 and = 1.05457266 1034 J s is Plancks h-bar constant.

However, what is the meaning of the symbols | and H? The answers will
be given in the first part of the course (chapters 1-4), which reviews several
physical and mathematical concepts that are needed to formulate the theory
of quantum mechanics. We will learn that | in Eq. (0.1) represents the
ket-vector state of the system and H represents the Hamiltonian operator.
The operator H is directly related to the Hamiltonian function in classical
physics, which will be defined in the first chapter. The ket-vector state and
its physical meaning will be introduced in the second chapter. Chapter 3
reviews the position and momentum operators, whereas chapter 4 discusses
dynamics of quantum systems. The second part of the course (chapters 5-7)
is devoted to some relatively simple quantum systems including a harmonic
oscillator, spin, hydrogen atom and more. In chapter 8 we will study quantum
systems in thermal equilibrium. The third part of the course (chapters 9-13)
is devoted to approximation methods such as perturbation theory, semiclassical and adiabatic approximations. Light and its interaction with matter are
the subjects of chapter 14-15. Finally, systems of identical particles will be
discussed in chapter 16 and open quantum systems in chapter 17. Most of
the material in these lecture notes is based on the textbooks [1, 2, 3, 4, 6, 7].
i
Contents
1.
Hamiltons Formalism of Classical Physics . . . . . . . . . . . . . . . .

1.1 Action and Lagrangian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Principle of Least Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4 Poissons Brackets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.6 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1
2
5
7
8
9
2.
State Vectors and Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.1 Linear Vector Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Diracs notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4 Dual Correspondence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5 Matrix Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.6 Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.6.1 Hermitian Adjoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.6.2 Eigenvalues and Eigenvectors . . . . . . . . . . . . . . . . . . . . . .
2.7 Quantum Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.8 Example - Spin 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.9 Unitary Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.10 Trace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.11 Commutation Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.12 Simultaneous Diagonalization of Commuting Operators . . . . .
2.13 Uncertainty Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.14 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.15 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
15
17
17
18
20
22
22
23
28
29
32
33
34
35
36
38
40
3.
The Position and Momentum Observables . . . . . . . . . . . . . . . .

3.1 The One Dimensional Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.1 Position Representation . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.2 Momentum Representation . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Transformation Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Generalization for 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
49
49
50
54
55
57
58
Contents
3.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.
Quantum Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1 Time Evolution Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Time Independent Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 Example - Spin 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4 Connection to Classical Dynamics . . . . . . . . . . . . . . . . . . . . . . . .
4.5 Symmetric Ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
69
70
71
73
74
76
80
5.
The Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.1 Eigenstates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Coherent States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.4 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
101
102
104
107
115
6.
Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.1 Angular Momentum and Rotation . . . . . . . . . . . . . . . . . . . . . . . .
6.2 General Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Simultaneous Diagonalization of J2 and Jz . . . . . . . . . . . . . . . .
6.4 Example - Spin 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.5 Orbital Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
149
150
152
152
157
157
163
172
7.
Central Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.1 Simultaneous Diagonalization of the Operators H, L2 and Lz
7.2 The Radial Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3 Hydrogen Atom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
203
204
206
208
213
215
8.
Density Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1 Time Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.2 Quantum Statistical Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.4 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
223
227
227
228
239
9.
Time Independent Perturbation Theory . . . . . . . . . . . . . . . . . .

9.1 The Level En . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.1.1 Nondegenerate Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.1.2 Degenerate Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.2 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
293
294
295
297
297
300
Eyal Buks
Quantum Mechanics - Lecture Notes
Contents
9.4 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
10. Time-Dependent Perturbation Theory . . . . . . . . . . . . . . . . . . . .
10.1 Time Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.2 Perturbation Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3 Transition Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.1 The Stationary Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.2 The Near-Resonance Case . . . . . . . . . . . . . . . . . . . . . . . . .
10.3.3 H1 is Separable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
337
337
338
340
341
343
344
344
346
11. WKB Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11.1 WKB Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2 Turning Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.3 Bohr-Sommerfeld Quantization Rule . . . . . . . . . . . . . . . . . . . . . .
11.4 Tunneling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.6 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
351
351
354
358
360
361
361
12. Path Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12.1 Charged Particle in Electromagnetic Field . . . . . . . . . . . . . . . . .
12.2 Classical Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.3 Aharonov-Bohm Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.3.1 Two-slit Interference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.3.2 Gauge Invariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.4 One Dimensional Path Integrals . . . . . . . . . . . . . . . . . . . . . . . . . .
12.4.1 One Dimensional Free Particle . . . . . . . . . . . . . . . . . . . . .
12.4.2 Expansion Around the Classical Path . . . . . . . . . . . . . . .
12.4.3 One Dimensional Harmonic Oscillator . . . . . . . . . . . . . . .
12.5 Semiclassical Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
369
369
373
374
376
377
379
380
381
383
387
388
389
13. Adiabatic Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13.1 Momentary Diagonalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.2 Gauge Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.3 Adiabatic Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.4 The Case of Two Dimensional Hilbert Space . . . . . . . . . . . . . . .
13.5 Transition Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.5.1 The Case of Two Dimensional Hilbert Space . . . . . . . . .
13.6 Slow and Fast Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13.8 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
397
397
399
399
400
402
403
406
409
410
Eyal Buks
Contents
14. The Quantized Electromagnetic Field . . . . . . . . . . . . . . . . . . . . .
14.1 Classical Electromagnetic Cavity . . . . . . . . . . . . . . . . . . . . . . . . .
14.2 Quantum Electromagnetic Cavity . . . . . . . . . . . . . . . . . . . . . . . .
14.3 Periodic Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
415
415
420
422
423
424
15. Light Matter Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.2 Transition Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.2.1 Spontaneous Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.2.2 Stimulated Emission and Absorption . . . . . . . . . . . . . . . .
15.2.3 Selection Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.3 Semiclassical Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
431
431
432
432
433
434
436
438
438
16. Identical Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

16.1 Basis for the Hilbert Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.2 Bosons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.3 Fermions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.4 Changing the Basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.5 Many Particle Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.5.1 One-Particle Observables . . . . . . . . . . . . . . . . . . . . . . . . . .
16.5.2 Two-Particle Observables . . . . . . . . . . . . . . . . . . . . . . . . . .
16.6 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.7 Momentum Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.8 Spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.9 The Electron Gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.10Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16.11Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
445
445
448
449
451
453
453
454
455
458
460
461
463
465
17. Superconductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.1 Macroscopic Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.1.1 Single Particle in Electromagnetic Field . . . . . . . . . . . . .
17.1.2 Drude Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.1.3 The Macroscopic Quantum Model . . . . . . . . . . . . . . . . . .
17.1.4 London Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.2 The Josephson Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.2.1 The First Josephson Relation . . . . . . . . . . . . . . . . . . . . . .
17.2.2 The Second Josephson Relation . . . . . . . . . . . . . . . . . . . .
17.2.3 The Energy of a Josephson Junction . . . . . . . . . . . . . . . .
17.3 RF SQUID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.3.1 Lagrangian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.3.2 Flux Quantum Bit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
483
483
483
485
489
489
493
493
494
495
496
497
499
Eyal Buks
Contents
17.3.3 Qubit Readout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4 BCS Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.1 Phonon Mediated Electron-Electron Interaction . . . . . .
17.4.2 The Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.3 Bogoliubov Transformation . . . . . . . . . . . . . . . . . . . . . . . .
17.4.4 The Energy Gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.5 The Ground State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.4.6 Pairing Wavefunction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.5 The Josephson Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.5.1 The Second Josephson Relation . . . . . . . . . . . . . . . . . . . .
17.5.2 The Energy of a Josephson Junction . . . . . . . . . . . . . . . .
17.5.3 The First Josephson Relation . . . . . . . . . . . . . . . . . . . . . .
17.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17.7 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
500
509
509
512
513
516
518
520
521
522
524
527
528
530
18. Open Quantum Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18.1 Classical Resonator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.2 Quantum Resonator Coupled to Thermal Bath . . . . . . . . . . . . .
18.2.1 The closed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.2.2 Coupling to Thermal Bath . . . . . . . . . . . . . . . . . . . . . . . .
18.2.3 Thermal Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3 Two Level System Coupled to Thermal Bath . . . . . . . . . . . . . . .
18.3.1 The Closed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3.2 Coupling to Thermal Baths . . . . . . . . . . . . . . . . . . . . . . . .
18.3.3 Thermal Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3.4 Correlation Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.3.5 The Bloch Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.4 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18.5 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
539
539
540
540
541
544
547
547
548
552
553
555
555
556
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565
Eyal Buks
1. Hamiltons Formalism of Classical Physics
In this chapter the Hamiltons formalism of classical physics is introduced,

with a special emphasis on the concepts that are needed for quantum mechanics.
1.1 Action and Lagrangian

Consider a classical physical system having N degrees of freedom. The classical state of the system can be described by N independent coordinates qn ,
where n = 1, 2, , N . The vector of coordinates is denoted by
Q = (q1 , q2 , , qN ) .
(1.1)
Consider the case where the vector of coordinates takes the value Q1 at time
t1 and the value Q2 at a later time t2 > t1 , namely
Q (t1 ) = Q1 ,
Q (t2 ) = Q2 .
(1.2)
(1.3)
The action S associated with the evolution of the system from time t1 to
time t2 is defined by
S=
t2
t1
dt L ,
(1.4)
where L is the Lagrangian function of the system. In general, the Lagrangian

is a function of the coordinates Q, the velocities Q and time t, namely

t ,
L = L Q, Q;
(1.5)
where
Q = (q1 , q2 , , qN ) ,
(1.6)
and where overdot denotes time derivative. The time evolution of Q, in turn,
depends on the trajectory taken by the system from point Q1 at time t1
Chapter 1. Hamiltons Formalism of Classical Physics
Q
Q2
Q1
t
t1
t2
Fig. 1.1. A trajectory taken by the system from point Q1 at time t1 to point Q2
at time t2 .
to point Q2 at time t2 (see Fig. 1.1). For a given trajectory the time
dependency is denoted as
Q (t) = Q (t) .
(1.7)
1.2 Principle of Least Action

For any given trajectory Q (t) the action can be evaluated using Eq. (1.4).
Consider a classical system evolving in time from point Q1 at time t1 to point
Q2 at time t2 along the trajectory Q (t). The trajectory Q (t), which is
obtained from the laws of classical physics, has the following unique property
known as the principle of least action:
Proposition 1.2.1 (principle of least action). Among all possible trajectories from point Q1 at time t1 to point Q2 at time t2 the action obtains its
minimal value by the classical trajectory Q (t).
In a weaker version of this principle, the action obtains a local minimum
for the trajectory Q (t). As the following theorem shows, the principle of
least action leads to a set of equations of motion, known as Euler-Lagrange
equations.
Theorem 1.2.1. The classical trajectory Q (t), for which the action obtains
its minimum value, obeys the Euler-Lagrange equations of motion, which are
given by
Eyal Buks
1.2. Principle of Least Action

d L
L
=
,
dt qn
qn
(1.8)
where n = 1, 2, , N.
Proof. Consider another trajectory Q (t) from point Q1 at time t1 to point
Q2 at time t2 (see Fig. 1.2). The difference
Q = Q (t) Q (t) = (q1 , q2 , , qN )
(1.9)
is assumed to be infinitesimally small. To lowest order in Q the change in

the action S is given by
S =
t2
dt L
t1
t2
dt
dt
t1
t2
t1
N
N

L
L
qn +
qn
q
qn
n
n=1
n=1
N
N

L
L d
qn +
qn
qn
qn dt
n=1
n=1
.
(1.10)
Integrating the second term by parts leads to

S =
t2
dt
t1
N

L
d L
qn dt qn
n=1
N

L
n=1
qn
qn
t2
qn
.
t1
(1.11)
The last term vanishes since

Q (t1 ) = Q (t2 ) = 0 .
(1.12)
The principle of least action implies that

S = 0 .
(1.13)
This has to be satisfied for any Q, therefore the following must hold
L
d L
=
.
dt qn
qn
Eyal Buks
(1.14)
Q
Q2
Q1
t
t1
t2
Fig. 1.2. The classical trajectory Q (t) and the trajectory Q (t).
In what follows we will assume for simplicity that the kinetic energy T of
the system can be expressed as a function of the velocities Q only (namely,
it does not explicitly depend on the coordinates Q). The components of the
generalized force Fn , where n = 1, 2, , N , are derived from the potential
energy U of the system as follows
Fn =
U
d U
+
.
qn dt qn
(1.15)
When the potential energy can be expressed as a function of the coordinates

the system is
Q only (namely, when it is independent on the velocities Q),
said to be conservative. For that case, the Lagrangian can be expressed in
terms of T and U as
L=T U .
(1.16)
Example 1.2.1. Consider a point particle having mass m moving in a onedimensional potential U (x). The Lagrangian is given by
mx 2
U (x) .
2
From the Euler-Lagrange equation
L=T U =
d L
L
=
,
dt x
x
one finds that
U
m
x=
.
x
Eyal Buks
(1.17)
(1.18)
(1.19)
1.3. Hamiltonian
1.3 Hamiltonian
The set of Euler-Lagrange equations contains N second order differential
equations. In this section we derive an alternative and equivalent set of equations of motion, known as Hamilton-Jacobi equations, that contains twice the
number of equations, namely 2N, however, of first, instead of second, order.
Definition 1.3.1. The variable canonically conjugate to qn is defined by
L
.
qn
pn =
(1.20)
Definition 1.3.2. The Hamiltonian of a physical system is a function of

the vector of coordinates Q, the vector of canonical conjugate variables P =
(p1 , p2 , , pN ) and time, namely
H = H (Q, P ; t) ,
(1.21)
is defined by
H=
N

n=1
pn qn L ,
(1.22)
where L is the Lagrangian.

Theorem 1.3.1. The classical trajectory satisfies the Hamilton-Jacobi equations of motion, which are given by
H
,
pn
H
p n =
,
qn
qn =
(1.23)
(1.24)
where n = 1, 2, , N.
Proof. The differential of H is given by
dH = d
N

n=1
pn qn dL
N

L
L
L
=
qn dpn + pn dqn qn dqn qn dqn t dt
n=1

d L
dt qn
N

n=1
Eyal Buks
(qn dpn pn dqn )
pn
L
dt .
t
(1.25)

Thus the following holds
H
,
pn
H
pn =
,
qn
L
H
=
.
t
t
qn =
(1.26)
(1.27)
(1.28)
Corollary 1.3.1. The following holds

dH
H
=
.
dt
t
(1.29)
Proof. Using Eqs. (1.23) and (1.24) one finds that

N
dH
=
dt
n=1

H
H
H
H
qn +
p n +
=
.
qn
pn
t
t

(1.30)
=0
The last corollary implies that H is time independent provided that H

does not depend on time explicitly, namely, provided that H/t = 0. This
property is referred to as the law of energy conservation. The theorem below
further emphasizes the relation between the Hamiltonian and the total energy
of the system.
Theorem 1.3.2. Assume that the kinetic energy of a conservative system is
given by

T =
nm qn qm ,
(1.31)
n,m
where nm are constants. Then,the Hamiltonian of the system is given by

H =T +U ,
(1.32)
where T is the kinetic energy of the system and where U is the potential
energy.
Proof. For a conservative system the potential energy is independent on velocities, thus
pl =
L
T
=
,
ql
ql
(1.33)
where L = T U is the Lagrangian. The Hamiltonian is thus given by
Eyal Buks
1.4. Poissons Brackets

H=
=
N

l=1
pl ql L
T
l
ql
ql (T U )
q
qm
n
nm qm
+ qn
ql T + U
ql
ql
l,n,m

=2

n,m
nl
ml
nm qn qm T + U

= T +U .
(1.34)
1.4 Poissons Brackets

Consider two physical quantities F and G that can be expressed as a function
of the vector of coordinates Q, the vector of canonical conjugate variables P
and time t, namely
F = F (Q, P ; t) ,
G = G (Q, P ; t) ,
(1.35)
(1.36)
The Poissons brackets are defined by

N

F G
F G
{F, G} =
,
qn pn pn qn
n=1
(1.37)
The Poissons brackets are employed for writing an equation of motion for a
general physical quantity of interest, as the following theorem shows.
Theorem 1.4.1. Let F be a physical quantity that can be expressed as a
function of the vector of coordinates Q, the vector of canonical conjugate
variables P and time t, and let H be the Hamiltonian. Then, the following
holds
dF
F
= {F, H} +
.
dt
t
(1.38)
Proof. Using Eqs. (1.23) and (1.24) one finds that the time derivative of F
is given by
Eyal Buks

N

dF
F
F
F
=
qn +
pn +
dt
qn
pn
t
n=1

N

F H
F H
F
+
=
q
p
p
q
t
n
n
n
n
n=1
= {F, H} +
F
.
t
(1.39)
Corollary 1.4.1. If F does not explicitly depend on time, namely if F/t =
0, and if {F, H} = 0, then F is a constant of the motion, namely
dF
=0.
dt
(1.40)
1.5 Problems
1. Consider a particle having charge q and mass m in electromagnetic field
characterized by the scalar potential and the vector potential A. The
electric field E and the magnetic field B are given by
E =
1 A
,
c t
(1.41)
and
B=A.
(1.42)
Let r = (x, y, z) be the Cartesian coordinates of the particle.

a) Verify that the Lagrangian of the system can be chosen to be given
by
L=
1 2
q
mr q + A r ,
2
c
(1.43)
by showing that the corresponding Euler-Lagrange equations are

equivalent to Newtons 2nd law (i.e., F = m
r).
b) Show that the Hamilton-Jacobi equations are equivalent to Newtons
2nd law.
c) Gauge transformation The electromagnetic field is invariant under the gauge transformation of the scalar and vector potentials
A A + ,
(1.44)
1

(1.45)
c t
where = (r, t) is an arbitrary smooth and continuous function
of r and t. What effect does this gauge transformation have on the
Lagrangian and Hamiltonian? Is the motion affected?
Eyal Buks
1.6. Solutions
Fig. 1.3. LC resonator.
2. Consider an LC resonator made of a capacitor having capacitance C in

parallel with an inductor having inductance L (see Fig. 1.3). The state
of the system is characterized by the coordinate q , which is the charge
stored by the capacitor.
a) Find the Euler-Lagrange equation of the system.
b) Find the Hamilton-Jacobi equations of the system.
c) Show that {q, p} = 1 .
3. Show that Poisson brackets satisfy the following relations
{qj , qk } = 0 ,
{pj , pk } = 0 ,
{qj , pk } = jk ,
{F, G} = {G, F } ,
{F, F } = 0 ,
{F, K} = 0 if K constant or F depends only on t ,
{E + F, G} = {E, G} + {F, G} ,
{E, F G} = {E, F } G + F {E, G} .
(1.46)
(1.47)
(1.48)
(1.49)
(1.50)
(1.51)
(1.52)
(1.53)
4. Show that the Lagrange equations are coordinate invariant.

5. Consider a point particle having mass m moving in a 3D central potential,
namely a potential V (r) that depends only on the distance
r = x2 + y 2 + z 2 from the origin. Show that the angular momentum
L = r p is a constant of the motion.
1.6 Solutions
1. The Lagrangian of the system (in Gaussian units) is taken to be given
by
1
q
L = mr2 q + A r .
2
c
Eyal Buks
(1.54)

a) The Euler-Lagrange equation for the coordinate x is given by
d L
L
=
,
dt x
x
(1.55)
where
d L
q
= m
x+
dt x
c
L
q
= q
+
x
x
c

Ax
Ay
Az
x
+ y
+ z
,
x
x
x
Ax
Ax
Ax
Ax
+ x
+ y
+ z
t
x
y
z
(1.56)
and
thus
q Ax
m
x = q
x c t
qEx
(1.57)

q
Ay
Ax
Ax Az
,
+
y
c
x
y
z
x

( A)z
( A)y

(r(A))x
or
m
x = qEx +
q
(r B)x .
c
(1.58)
(1.59)
Similar equations are obtained for y and z in the same way. These 3
equations can be written in a vector form as

1
m
r = q E + r B .
(1.60)
c
b) The variable vector canonically conjugate to the coordinates vector
r is given by
p=
L
q
= mr+ A .
r
c
(1.61)
The Hamiltonian is thus given by
Eyal Buks
10
1.6. Solutions
H = p r L

1
q
= r p mr A + q
2
c
1 2
= mr + q
2

2
p qc A
=
+ q .
2m
(1.62)
The Hamilton-Jacobi equation for the coordinate x is given by
x =
H
,
px
(1.63)
px qc Ax
,
m
(1.64)
thus
x =
or
q
px = mx+
Ax .
c
(1.65)
The Hamilton-Jacobi equation for the canonically conjugate variable

px is given by
p x =
H
,
x
(1.66)
where
p x = m
x+
q
c

Ax
Ax
Ax
q Ax
x
+ y
+ z
+
,
x
y
z
c t
(1.67)
and

pz qc Az Az
H
q px qc Ax Ax py qc Ay Ay
=
+
+
q
x
c
m
x
m
x
m
x
x

q
Ax
Ay
Az
=
x
+ y
+ z
q
,
c
x
x
x
x
(1.68)
thus

q Ax q
Ay
Ax
Ax Az
m
x = q
+
y
.
x c t
c
x
y
z
x
(1.69)
The last result is identical to Eq. (1.59).
Eyal Buks
11

c) Clearly, the fields E and B, which are given by Eqs. (1.41) and (1.42)
respectively, are unchanged since

()
=0,
(1.70)
t
t
and
() = 0 .
(1.71)
Thus, even though both L and H are modified, the motion, which
depends on E and B only, is unaffected.
2. The kinetic energy in this case T = Lq2 /2 is the energy stored in the
inductor, and the potential energy U = q 2 /2C is the energy stored in the
capacitor.
a) The Lagrangian is given by
L=T U =
q2
Lq2
.
2
2C
(1.72)
The Euler-Lagrange equation for the coordinate q is given by

d L
L
=
,
dt q
q
(1.73)
q
=0.
C
(1.74)
thus
L
q+
This equation expresses the requirement that the voltage across the
capacitor is the same as the one across the inductor.
b) The canonical conjugate momentum is given by
p=
L
= Lq ,
q
(1.75)
and the Hamiltonian is given by

H = pq L =
p2
q2
+
.
2L 2C
(1.76)
Hamilton-Jacobi equations read

p
L
q
p = ,
C
q =
(1.77)
(1.78)
thus
L
q+
Eyal Buks
q
=0.
C
(1.79)
12
1.6. Solutions
c) Using the definition (1.37) one has
{q, p} =
q p q p
=1.
q p p q
(1.80)
3. All these relations

are easily proven using the definition (1.37).
4. Let L = L Q, Q; t be a Lagrangian of a system, where Q = (q1 , q2 , )

is the vector of coordinates, Q = (q1 , q2 , ) is the vector of velocities, and where overdot denotes time derivative. Consider the coordinates
transformation
xa = xa (q1 , q2 , ..., t) ,
(1.81)
where a = 1, 2, . The following holds

x a =
xa
xa
qb +
,
qb
t
(1.82)
where the summation convention is being used, namely, repeated indices

are summed over. Moreover
L xb
L x b
L
=
+
,
qa
xb qa
x b qa
(1.83)
and
d
dt
L
qa
d
=
dt
L x b
x b qa
(1.84)
As can be seen from Eq. (1.82), one has

x b
xb
=
.
qa
qa
Thus, using Eqs. (1.83) and (1.84) one finds

d L xb
d L
L
=
dt qa
qa
dt x b qa
L xb
L x b
x q
x q
b a b a
d L
L xb
=
dt x b
xb qa

d xb
x b L
+
.
dt qa
qa x b
(1.85)
(1.86)
As can be seen from Eq. (1.82), the second term vanishes since
Eyal Buks
13

2 xb
2 xb
d
x b
=
qc +
=
qa
qa qc
tqa
dt
xb
qa
thus
d
dt
L
qa

L
d L
L xb
=
.
qa
dt x b
xb qa
(1.87)
The last result shows that if the coordinate transformation is reversible,

namely if det (xb /qa ) = 0 then Lagrange equations are coordinate
invariant.
5. The angular momentum L is given by
x
y

z
L = r p = det x y z ,
(1.88)
px py pz
where r = (x, y, z) is the position vector and where p = (px , py , pz ) is the
momentum vector. The Hamiltonian is given by
H=
p2
+ V (r) .
2m
(1.89)
Using
{xi , pj } = ij ,
Lz = xpy ypx ,
(1.90)
(1.91)
one finds that

2

p , Lz = p2x , Lz + p2y , Lz + p2z , Lz

= p2x , xpy p2y , ypx
= 2px py + 2py px
=0,
(1.92)
and
2

r , Lz = x2 , Lz + y 2 , Lz + z 2 , Lz

= y x2 , px + y 2 , py x
=0.
(1.93)
2
2
Thus f r , Lz = 0 for arbitrary smooth function f r , and consequently {H, Lz } = 0. In a similar way one can show that {H, Lx } =
{H, Ly } = 0, and therefore H, L2 = 0.
Eyal Buks
14
2. State Vectors and Operators
In quantum mechanics the state of a physical system is described by a state

vector |, which is a vector in a vector space F, namely
| F .
(2.1)
Here, we have employed the Diracs ket-vector notation | for the state vector, which contains all information about the state of the physical system
under study. The dimensionality of F is finite in some specific cases (notably, spin systems), however, it can also be infinite in many other cases
of interest. The basic mathematical theory dealing with vector spaces having infinite dimensionality was mainly developed by David Hilbert. Under
some conditions, vector spaces having infinite dimensionality have properties
similar to those of their finite dimensionality counterparts. A mathematically
rigorous treatment of such vector spaces having infinite dimensionality, which
are commonly called Hilbert spaces, can be found in textbooks that are devoted to this subject. In this chapter, however, we will only review the main
properties that are useful for quantum mechanics. In some cases, when the
generalization from the case of finite dimensionality to the case of arbitrary
dimensionality is nontrivial, results will be presented without providing a
rigorous proof and even without accurately specifying what are the validity
conditions for these results.
2.1 Linear Vector Space

A linear vector space F is a set {|} of mathematical objects called vectors.
The space is assumed to be closed under vector addition and scalar multiplication. Both, operations (i.e., vector addition and scalar multiplication) are
commutative. That is:
1. | + | = | + | F for every | F and | F
2. c | = | c F for every | F and c C (where C is the set of
complex numbers)
A vector space with an inner product is called an inner product space.
An inner product of the ordered pair | , | F is denoted as |. The
inner product is a function F 2 C that satisfies the following properties:
Chapter 2. State Vectors and Operators

| C ,
(2.2)
| = | ,
(2.3)
(c1 | 1 + c2 | 2 ) = c1 | 1 + c2 | 2 , where c1 , c2 C ,
(2.4)
| R and | 0. Equality holds iff | = 0 .
(2.5)
Note that the asterisk in Eq. (2.3) denotes complex conjugate. Below we list
some important definitions and comments regarding inner product:

The real number | is called the norm of the vector | F.
A normalized vector has a unity norm, namely | = 1.
Every nonzero vector 0 = | F can be normalized using the transformation
|
|
.
|
(2.6)
The vectors | F and | F are said to be orthogonal if | = 0.

A set of vectors {|n }n , where |n F is called a complete orthonormal
basis if
The vectors are all normalized and orthogonal to each other, namely
m |n = nm .
(2.7)
Every | F can be written as a superposition of the basis vectors,

namely

| =
cn |n ,
(2.8)
n
where cn C.
By evaluating the inner product m |, where | is given by Eq. (2.8)
one finds with the help of Eq. (2.7) and property (2.4) of inner products
that
!
"

m | = m
cn |n =
cn m |n = cm .
(2.9)

n
n
=nm
The last result allows rewriting Eq. (2.8) as

| =
cn |n =
|n cn =
|n n | .
n
Eyal Buks
(2.10)
16
2.3. Diracs notation
2.2 Operators
Operators, as the definition below states, are function from F to F:
Definition 2.2.1. An operator A : F F on a vector space maps vectors
onto vectors, namely A | F for every | F.
Some important definitions and comments are listed below:
The operators X : F F and Y : F F are said to be equal, namely
X = Y , if for every | F the following holds
X | = Y | .
(2.11)
Operators can be added, and the addition is both, commutative and associative, namely
X +Y = Y +X ,
X + (Y + Z) = (X + Y ) + Z .
(2.12)
(2.13)
An operator A : F F is said to be linear if

A (c1 | 1 + c2 | 2 ) = c1 A | 1 + c2 A | 2
(2.14)
for every | 1 , | 2 F and c1 , c2 C.

The operators X : F F and Y : F F can be multiplied, where
XY | = X (Y |)
(2.15)
for any | F.
Operator multiplication is associative
X (Y Z) = (XY ) Z = XY Z .
(2.16)
However, in general operator multiplication needs not be commutative

XY = Y X .
(2.17)
2.3 Diracs notation

In Diracs notation the inner product is considered as a multiplication of two
mathematical objects called bra and ket
| = | | .

bra
(2.18)
ket
While the ket-vector | is a vector in F, the bra-vector | represents a

functional that maps any ket-vector | F to the complex number |.
Eyal Buks
17

While the multiplication of a bra-vector on the left and a ket-vector on the
right represents inner product, the outer product is obtained by reversing the
order
A = | | .
(2.19)
The outer product A is clearly an operator since for any | F the object
A | is a ket-vector
A | = (| |) | = | | F .

(2.20)
Moreover, according to property (2.4), A is linear since for every | 1 , | 2

F and c1 , c2 C the following holds
A (c1 | 1 + c2 | 2 ) = | | (c1 | 1 + c2 | 2 )
= | (c1 | 1 + c2 | 2 )
= c1 A | 1 + c2 A | 2 .
With Diracs notation Eq. (2.10) can be rewritten as
!
"

| =
|n n | | .
(2.21)
(2.22)
Since the above identity holds for any | F one concludes that the quantity
in brackets is the identity operator, which is denoted as 1, namely

|n n | .
(2.23)
1=
n
This result, which is called the closure relation, implies that any complete
orthonormal basis can be used to express the identity operator.
2.4 Dual Correspondence

As we have mentioned above, the bra-vector | represents a functional mapping any ket-vector | F to the complex number |. Moreover, since
the inner product is linear [see property (2.4) above], such a mapping is linear,
namely for every | 1 , | 2 F and c1 , c2 C the following holds
| (c1 | 1 + c2 | 2 ) = c1 | 1 + c2 | 2 .
(2.24)
The set of linear functionals from F to C, namely, the set of functionals F : F

C that satisfy
Eyal Buks
18
2.4. Dual Correspondence

F (c1 | 1 + c2 | 2 ) = c1 F (| 1 ) + c2 F (| 2 )
(2.25)
for every | 1 , | 2 F and c1 , c2 C, is called the dual space F . As

the name suggests, there is a dual correspondence (DC) between F and F ,
namely a one to one mapping between these two sets, which are both linear
vector spaces. The duality relation is presented using the notation
| | ,
(2.26)
where | F and | F . What is the dual of the ket-vector | =

c1 | 1 + c2 | 2 , where | 1 , | 2 F and c1 , c2 C? To answer this question
we employ the above mentioned general properties (2.3) and (2.4) of inner
products and consider the quantity | for an arbitrary ket-vector | F
| = |
= (c1 | 1 + c2 | 2 )
= c1 1 | + c2 2 |
= (c1 1 | + c2 2 |) | .
(2.27)
From this result we conclude that the duality relation takes the form
c1 1 | + c2 2 | c1 | 1 + c2 | 2 .
(2.28)
The last relation describes how to map any given ket-vector | F

to its dual F = | : F C, where F F is a linear functional that
maps any ket-vector | F to the complex number |. What is the
inverse mapping? The answer can take a relatively simple form provided that
a complete orthonormal basis exists, and consequently the identity operator
can be expressed as in Eq. (2.23). In that case the dual of a given linear
functional F : F C is the ket-vector |FD F, which is given by

|FD =
(F (|n )) |n .
(2.29)
n
The duality is demonstrated by proving the two claims below:

Claim. | DD = | for any | F, where | DD is the dual of the dual of
|.
Proof. The dual of | is the bra-vector |, whereas the dual of | is found
using Eqs. (2.29) and (2.23), thus
Eyal Buks
19

| DD =
=n |
|n n |
|n n | |
|n |n

=1
= | .
(2.30)
Claim. FDD = F for any F F , where FDD is the dual of the dual of F .
Proof. The dual |FD F of the functional F F is given by Eq. (2.29).
Thus with the help of the duality relation (2.28) one finds that dual FDD F
of |FD is given by

FDD =
F (|n ) n | .
(2.31)
n
Consider an arbitrary ket-vector | F that is written as a superposition

of the complete orthonormal basis vectors, namely

| =
cm |m .
(2.32)
m
Using the above expression for FDD and the linearity property one finds that

FDD | =
cm F (|n ) n |m

n,m
=

n
=F
mn
cn F (|n )
!

n
cn |n
"
= F (|) ,
(2.33)
therefore, FDD = F .
2.5 Matrix Representation

Given a complete orthonormal basis, ket-vectors, bra-vectors and linear operators can be represented using matrices. Such representations are easily
obtained using the closure relation (2.23).
Eyal Buks
20
2.5. Matrix Representation

The inner product between the bra-vector | and the ket-vector | can
be written as
| = | 1 |

=
|n n |
n
|

1 |
= |1 |2 2
.
..
.
(2.34)
Thus, the inner product can be viewed as a product between the row vector

| =
|1 |2 ,
(2.35)
which is the matrix representation of the bra-vector |, and the column
vector
1 |
| =
2 | ,
(2.36)
..
.
which is the matrix representation of the ket-vector |. Obviously, both

representations are basis dependent.
Multiplying the relation | = X | from the right by the basis bra-vector
m | and employing again the closure relation (2.23) yields

m | = m | X | = m | X1 | =
m | X |n n | , (2.37)
n
or in matrix form
1 |
1 | X |1 1 | X |2
1 |
2 | 2 | X |1 2 | X |2 2 |
.
..
..
..
..
.
.
.
.
(2.38)
In view of this expression, the matrix representation of the linear operator

X is given by
1 | X |1 1 | X |2
X=
2 | X |1 2 | X |2 .
(2.39)
..
..
.
.
Alternatively, the last result can be written as
Xnm = n | X |m ,
(2.40)
where Xnm is the element in row n and column m of the matrix representation of the operator X.
Eyal Buks
21

Such matrix representation of linear operators can be useful also for multiplying linear operators. The matrix elements of the product Z = XY are
given by
m | Z |n = m | XY |n = m | X1Y |n =

l
m | X |l l | Y |n .
(2.41)
Similarly, the matrix representation of the outer product | | is given

by
1 |

| | =
2 | |1 |2
..
.
1 | |1 1 | |2
= 2 | |1 2 | |2 .
..
..
.
.
(2.42)
2.6 Observables
Measurable physical variables are represented in quantum mechanics by Hermitian operators.
2.6.1 Hermitian Adjoint
Definition 2.6.1. The Hermitian adjoint of an operator X is denoted as X
and is defined by the following duality relation
| X X | .
(2.43)
Namely, for any ket-vector | F, the dual to the ket-vector X | is the

bra-vector | X .
Definition 2.6.2. An operator is said to be Hermitian if X = X .
Below we prove some simple relations:
Claim. | X | = | X |
Proof. Using the general property (2.3) of inner products one has

| X | = | (X |) = | X | = | X | .
(2.44)
Note that this result implies that if X = X then | X | = | X | .
Eyal Buks
22
2.6. Observables

Claim. X = X
Proof. For any | , | F the following holds

| X | = | X | = | X | = | X | ,
(2.45)

thus X = X.
Claim. (XY ) = Y X
Proof. Applying XY on an arbitrary ket-vector | F and employing the
duality correspondence yield

XY | = X (Y |) | Y X = | Y X ,
(2.46)
thus
(XY ) = Y X .
(2.47)
Claim. If X = | | then X = | |
Proof. By applying X on an arbitrary ket-vector | F and employing the
duality correspondence one finds that
X | = (| |) | = | ( |) ( |) | = | | = | X ,
(2.48)
where X = | |.
2.6.2 Eigenvalues and Eigenvectors
Each operator is characterized by its set of eigenvalues, which is defined
below:
Definition 2.6.3. A number an C is said to be an eigenvalue of an operator A : F F if for some nonzero ket-vector |an F the following
holds
A |an = an |an .
(2.49)
The ket-vector |an is then said to be an eigenvector of the operator A with

an eigenvalue an .
The set of eigenvectors associated with a given eigenvalue of an operator
A is called eigensubspace and is denoted as
Fn = {|an F such that A |an = an |an } .
Eyal Buks
(2.50)
23

Clearly, Fn is closed under vector addition and scalar multiplication, namely
c1 | 1 + c2 | 2 Fn for every | 1 , | 2 Fn and for every c1 , c2 C. Thus,
the set Fn is a subspace of F. The dimensionality of Fn (i.e., the minimum
number of vectors that are needed to span Fn ) is called the level of degeneracy
gn of the eigenvalue an , namely
gn = dim Fn .
(2.51)
As the theorem below shows, the eigenvalues and eigenvectors of a Hermitian operator have some unique properties.
Theorem 2.6.1. The eigenvalues of a Hermitian operator A are real. The
eigenvectors of A corresponding to different eigenvalues are orthogonal.
Proof. Let a1 and a2 be two eigenvalues of A with corresponding eigen vectors
|a1 and |a2
A |a1 = a1 |a1 ,
A |a2 = a2 |a2 .
(2.52)
(2.53)
Multiplying Eq. (2.52) from the left by the bra-vector a2 |, and multiplying
the dual of Eq. (2.53), which since A = A is given by
a2 | A = a2 a2 | ,
(2.54)
from the right by the ket-vector |a1 yield

a2 | A |a1 = a1 a2 |a1 ,
a2 | A |a1 = a2 a2 |a1 .
(2.55)
(2.56)
Thus, we have found that

(a1 a2 ) a2 |a1 = 0 .
(2.57)
The first part of the theorem is proven by employing the last result (2.57)
for the case where |a1 = |a2 . Since |a1 is assumed to be a nonzero ketvector one concludes that a1 = a1 , namely a1 is real. Since this is true for
any eigenvalue of A, one can rewrite Eq. (2.57) as
(a1 a2 ) a2 |a1 = 0 .
(2.58)
The second part of the theorem is proven by considering the case where
a1 = a2 , for which the above result (2.58) can hold only if a2 |a1 = 0.
Namely eigenvectors corresponding to different eigenvalues are orthogonal.
Consider a Hermitian operator A having a set of eigenvalues {an }n . Let
gn be the degree of degeneracy of eigenvalue an , namely gn is the dimension
of the corresponding eigensubspace, which is denoted by Fn . For simplicity, assume that gn is finite for every n. Let {|an,1 , |an,2 , , |an,gn } be
Eyal Buks
24
2.6. Observables
an orthonormal basis of the eigensubspace Fn , namely an,i |an,i = ii .
Constructing such an orthonormal basis for Fn can be done by the so-called
Gram-Schmidt process. Moreover, since eigenvectors of A corresponding to
different eigenvalues are orthogonal, the following holds
an ,i |an,i = nn ii ,
(2.59)
In addition, all the ket-vectors |an,i are eigenvectors of A

A |an,i = an |an,i .
(2.60)
Projectors. Projector operators are useful for expressing the properties of

an observable.
Definition 2.6.4. An Hermitian operator P is called a projector if P 2 = P .
Claim. The only possible eigenvalues of a projector are 0 and 1.
Proof. Assume that |p is an eigenvector of P with an eigenvalue p, namely
P |p = p |p. Applying the operator P on both sides and using the fact that
P 2 = P yield P |p = p2 |p, thus p (1 p) |p = 0, therefore since |p is
assumed to be nonzero, either p = 0 or p = 1.
A projector is said to project any given vector onto the eigensubspace
corresponding to the eigenvalue p = 1.
Let {|an,1 , |an,2 , , |an,gn } be an orthonormal basis of an eigensubspace Fn corresponding to an eigenvalue of an observable A. Such an orthonormal basis can be used to construct a projection Pn onto Fn , which is given
by
Pn =
gn

i=1
|an,i an,i | .
(2.61)
Clearly, Pn is a projector since Pn = Pn and since

Pn2
gn

i,i =1
|an,i an,i |an,i an,i | =

ii
gn

i=1
|an,i an,i | = Pn .
(2.62)
Moreover, it is easy to show using the orthonormality relation (2.59) that the
following holds
Pn Pm = Pm Pn = Pn nm .
(2.63)
For linear vector spaces of finite dimensionality, it can be shown that the
set {|an,i }n,i forms a complete orthonormal basis of eigenvectors of a given
Hermitian operator A. The generalization of this result for the case of arbitrary dimensionality is nontrivial, since generally such a set needs not be
Eyal Buks
25

complete. On the other hand, it can be shown that if a given Hermitian operator A satisfies some conditions (e.g., A needs to be completely continuous)
then completeness is guarantied. For all Hermitian operators of interest for
this course we will assume that all such conditions are satisfied. That is, for
any such Hermitian operator A there exists a set of ket vectors {|an,i }, such
that:
1. The set is orthonormal, namely
an ,i |an,i = nn ii ,
(2.64)
2. The ket-vectors |an,i are eigenvectors, namely

A |an,i = an |an,i ,
(2.65)
where an R.
3. The set is complete, namely closure relation [see also Eq. (2.23)] is satisfied
1=
gn

n
i=1
|an,i an,i | =
Pn ,
(2.66)
where
Pn =
gn

i=1
|an,i an,i |
(2.67)
is the projector onto eigen subspace Fn with the corresponding eigenvalue

an .
The closure relation (2.66) can be used to express the operator A in terms
of the projectors Pn
A = A1 =
gn

n i=1
A |an,i an,i | =

n
an
gn

i=1
|an,i an,i | ,
(2.68)
that is
A=
an Pn .
(2.69)
The last result is very useful when dealing with a function f (A) of the
operator A. The meaning of a function of an operator can be understood in
terms of the Taylor expansion of the function

f (x) =
fm xm ,
(2.70)
m
Eyal Buks
26
2.6. Observables
where
fm =
1 dm f
.
m! dxm
(2.71)
With the help of Eqs. (2.63) and (2.69) one finds that

f (A) =
fm Am
m
fm
fm

n
an Pn
!

"m
am
n Pn
fm am
n Pn ,

f (an )
(2.72)
thus
f (A) =
f (an ) Pn .
(2.73)
Exercise 2.6.1. Express the projector Pn in terms of the operator A and

its set of eigenvalues.
Solution 2.6.1. We seek a function f such that f (A) = Pn . Multiplying
from the right by a basis ket-vector |am,i yields
f (A) |am,i = mn |am,i .
(2.74)
On the other hand

f (A) |am,i = f (am ) |am,i .
(2.75)
Thus we seek a function that satisfy

f (am ) = mn .
(2.76)
The polynomial function

#
f (a) = K
(a am ) ,
(2.77)
m
=n
where K is a constant, satisfies the requirement that f (am ) = 0 for every

m = n. The constant K is chosen such that f (an ) = 1, that is
# a am
f (a) =
,
(2.78)
an am
m
=n
Eyal Buks
27

Thus, the desired expression is given by
Pn =
# A am
.
an am
(2.79)
m
=n
2.7 Quantum Measurement

Consider a measurement of a physical variable denoted as A(c) performed on
a quantum system. The standard textbook description of such a process is
described below. The physical variable A(c) is represented in quantum mechanics by an observable, namely by a Hermitian operator, which is denoted
as A. The correspondence between the variable A(c) and the operator A will
be discussed below in chapter 4. As we have seen above, it is possible to construct a complete orthonormal basis made of eigenvectors of the Hermitian
operator A having the properties given by Eqs. (2.64), (2.65) and (2.66). In
that basis, the vector state | of the system can be expressed as
| = 1 | =
gn

n i=1
an,i | |an,i .
(2.80)
Even when the state vector | is given, quantum mechanics does not generally provide a deterministic answer to the question: what will be the outcome
of the measurement. Instead it predicts that:
1. The possible results of the measurement are the eigenvalues {an } of the
operator A.
2. The probability pn to measure the eigen value an is given by
pn = | Pn | =
gn

i=1
|an,i ||2 .
(2.81)
Note that the state vector | is assumed to be normalized.

3. After a measurement of A with an outcome an the state vector collapses
onto the corresponding eigensubspace and becomes
Pn |
|
.
| Pn |
(2.82)
It is easy to show that the probability to measure something is unity

provided that | is normalized:
!
"

pn =
| Pn | = |
Pn | = 1 .
(2.83)
n
Eyal Buks
28
2.8. Example - Spin 1/2

We also note that a direct consequence of the collapse postulate is that two
subsequent measurements of the same observable performed one immediately
after the other will yield the same result. It is also important to note that the
above standard textbook description of the measurement process is highly
controversial, especially, the collapse postulate. However, a thorough discussion of this issue is beyond the scope of this course.
Quantum mechanics cannot generally predict the outcome of a specific
measurement of an observable A, however it can predict the average, namely
the expectation value, which is denoted as A. The expectation value is easily
calculated with the help of Eq. (2.69)

A =
an pn =
an | Pn | = | A | .
(2.84)
n
2.8 Example - Spin 1/2

Spin is an internal degree of freedom of elementary particles. Electrons, for
example, have spin 1/2. This means, as we will see in chapter 6, that the
state of a spin 1/2 can be described by a state vector | in a vector space
of dimensionality 2. In other words, spin 1/2 is said to be a two-level system.
The spin was first discovered in 1921 by Stern and Gerlach in an experiment
in which the magnetic moment of neutral silver atoms was measured. Silver
atoms have 47 electrons, 46 out of which fill closed shells. It can be shown
that only the electron in the outer shell contributes to the total magnetic
moment of the atom. The force F acting on a magnetic moment moving in
a magnetic field B is given by F = ( B). Thus by applying a nonuniform
magnetic field B and by monitoring the atoms trajectories one can measure
the magnetic moment.
It is important to keep in mind that generally in addition to the spin
contribution to the magnetic moment of an electron, also the orbital motion
of the electron can contribute. For both cases, the magnetic moment is related
to angular momentum by the gyromagnetic ratio. However this ratio takes
different values for these two cases. The orbital gyromagnetic ratio can be
evaluated by considering a simple example of an electron of charge e moving
in a circular orbit or radius r with velocity v. The magnetic moment is given
by
orbital =
AI
,
c
(2.85)
where A = r2 is the area enclosed by the circular orbit and I = ev/ (2r)
is the electrical current carried by the electron, thus
orbital =
Eyal Buks
erv
.
2c
(2.86)
29

This result can be also written as
orbital = B L ,
(2.87)
where L = me vr is the orbital angular momentum, and where

B =
e
2me c
(2.88)
is the Bohrs magneton constant. The proportionality factor B / is the

orbital gyromagnetic ratio. In vector form and for a more general case of
orbital motion (not necessarily circular) the orbital gyromagnetic relation is
given by
orbital =
B
L.
(2.89)
On the other hand, as was first shown by Dirac, the gyromagnetic ratio
for the case of spin angular momentum takes twice this value
spin =
2B
S.
(2.90)
Note that we follow here the convention of using the letter L for orbital
angular momentum and the letter S for spin angular momentum.
The Stern-Gerlach apparatus allows measuring any component of the
magnetic moment vector. Alternatively, in view of relation (2.90), it can be
said that any component of the spin angular momentum S can be measured.
The experiment shows that the only two possible results of such a measurement are +/2 and /2. As we have seen above, one can construct a complete orthonormal basis to the vector space made of eigenvectors of any given
observable. Choosing the observable Sz = S
z for this purpose we construct
a basis made of two vectors {|+;
z , |;
z}. Both vectors are eigenvectors of
Sz

Sz |+;
z = |+;
z ,
(2.91)
2

Sz |;
z = |;
z .
(2.92)
2
In what follow we will use the more compact notation
|+ = |+;
z ,
| = |;
z .
(2.93)
(2.94)
The orthonormality property implied that

+ |+ = | = 1 ,
|+ = 0 .
(2.95)
(2.96)
The closure relation in the present case is expressed as

Eyal Buks
30

|+ +| + | | = 1 .
(2.97)
In this basis any ket-vector | can be written as

| = |+ + | + | | .
(2.98)
The closure relation (2.97) and Eqs. (2.91) and (2.92) yield
Sz =

(|+ +| | |)
2
(2.99)
It is useful to define also the operators S+ and S

S+ = |+ | ,
S = | +| .
(2.100)
(2.101)
In chapter 6 we will see that the x and y components of S are given by

(|+ | + | +|) ,
(2.102)
2

Sy = (i |+ | + i | +|) .
(2.103)
2
All these ket-vectors and operators have matrix representation, which for the
basis {|+;
z , |;
z} is given by

1
|+ =
,
(2.104)
0

0
| =
,
(2.105)
1

01
Sx =
,
(2.106)
2 10

0 i
Sy =
,
(2.107)
2 i 0

1 0
Sz =
,
(2.108)
2 0 1

01

,
(2.109)
S+ =
00

00
S =

.
(2.110)
10
Sx =
Exercise 2.8.1. Given that the state vector of a spin 1/2 is |+;
z calculate
(a) the expectation values Sx and Sz (b) the probability to obtain a value
of +/2 in a measurement of Sx .
Solution 2.8.1. (a) Using the matrix representation one has
Eyal Buks
31

01
1
10
=0,
10
0
2

1 0

1
10
Sz = +| Sz |+ =
= .
0 1
0
2
2
Sx = +| Sx |+ =
(2.111)
(2.112)
(b) First, the eigenvectors of the operator Sx are found by solving the equation Sx | = |, which is done by diagonalization of the matrix representation of Sx . The relation Sx | = | for the two eigenvectors is written
in a matrix form as
! "
! 1 "
1
01

2
2
=
,
(2.113)
1
2 10
2 12
2
!
"
! 1 "
1
01

2
2
=
.
(2.114)
12
2 10
2 12
That is, in ket notation

= |; x
,
Sx |; x
2
(2.115)
where the eigenvectors of Sx are given by

1
|; x
= (|+ |) .
2
(2.116)
Using this result the probability p+ is easily calculated
1
1
|2 =
+| (|+ + |)
= .
p+ = |+ |+; x
2
2
(2.117)
2.9 Unitary Operators

Unitary operators are useful for transforming from one orthonormal basis to
another.
Definition 2.9.1. An operator U is said to be unitary if U = U 1 , namely
if U U = U U = 1.
Consider two observables A and B, and two corresponding complete and
orthonormal bases of eigenvectors

A |an = an |an , am |an = nm ,
|an an | = 1 ,
(2.118)
n
B |bn = bn |bn , bm |bn = nm ,
The operator U, which is given by

Eyal Buks

n
|bn bn | = 1 .
(2.119)
32
2.10. Trace
U=

n
|bn an | ,
(2.120)
transforms each of the basis vector |an to the corresponding basis vector |bn
U |an = |bn .
(2.121)
It is easy to show that the operator U is unitary

U U =
|an bn |bm am | =
|an an | = 1 .

n,m
n
(2.122)
nm
The matrix elements of U in the basis {|an } are given by

an | U |am = an |bm ,
(2.123)
and those of U by
an | U |am = bn |am .
Consider a ket vector

| =
|an an | ,
(2.124)
which can be represented as a column vector in the basis {|an }. The nth
element of such a column vector is an |. The operator U can be employed
for finding the corresponding column vector representation of the same ketvector | in the other basis {|bn }

bn | =
bn |am am | =
an | U |am am | .
(2.125)
m
Similarly, Given an operator X the relation between the matrix elements

an | X |am in the basis {|an } to the matrix elements bn | X |bm in the
basis {|bn } is given by

bn | X |bm =
bn |ak ak | X |al al |bm
k,l

k,l
an | U |ak ak | X |al al | U |am .
(2.126)
2.10 Trace
Given an operator X and an orthonormal and complete basis {|an }, the
trace of X is given by
Eyal Buks
33

Tr (X) =

n
an | X |an .
(2.127)
It is easy to show that Tr (X) is independent on basis, as is shown below:

Tr (X) =
an | X |an
n
n,k,l
n,k,l

k,l

k
an |bk bk | X |bl bl |an

bl |an an |bk bk | X |bl
bl |bk bk | X |bl

kl
bk | X |bk .
(2.128)
The proof of the following relation

Tr (XY ) = Tr (Y X) ,
(2.129)
is left as an exercise.
2.11 Commutation Relation

The commutation relation of the operators A and B is defined as
[A, B] = AB BA .
(2.130)
As an example, the components Sx , Sy and Sz of the spin angular momentum

operator, satisfy the following commutation relations
[Si , Sj ] = iijk Sk ,
(2.131)
where
ijk
i, j, k are not all different

0
1 i, j, k is an even permutation of x, y, z
=
1 i, j, k is an odd permutation of x, y, z
(2.132)
is the Levi-Civita symbol. Equation (2.131) employs the Einsteins convention, according to which if an index symbol appears twice in an expression,
it is to be summed over all its allowed values. Namely, the repeated index k
should be summed over the values x, y and z:
ijk Sk = ijx Sx + ijy Sy + ijz Sz .
Eyal Buks
(2.133)
34
2.12. Simultaneous Diagonalization of Commuting Operators

Moreover, the following relations hold
1
Sx2 = Sy2 = Sz2 = 2 ,
4
(2.134)
3
S2 = Sx2 + Sy2 + Sz2 = 2 .
(2.135)
4
The relations below, which are easy to prove using the above definition,
are very useful for evaluating commutation relations
[F, G] = [G, F ] ,
[F, F ] = 0 ,
[E + F, G] = [E, G] + [F, G] ,
[E, F G] = [E, F ] G + F [E, G] .
(2.136)
(2.137)
(2.138)
(2.139)
2.12 Simultaneous Diagonalization of Commuting

Operators
Consider an observable A having a set of eigenvalues {an }. Let gn be the
degree of degeneracy of eigenvalue an , namely gn is the dimension of the
corresponding eigensubspace, which is denoted by Fn . Thus the following
holds
A |an,i = an |an,i ,
(2.140)
where i = 1, 2, , gn , and
an ,i |an,i = nn ii .
(2.141)
The set of vectors {|an,1 , |an,2 , , |an,gn } forms an orthonormal basis for
the eigensubspace Fn . The closure relation can be written as
1=
gn

n i=1
|an,i an,i | =
Pn ,
(2.142)
where
Pn =
gn

i=1
|an,i an,i | .
(2.143)
Now consider another observable B, which is assumed to commute with

A, namely [A, B] = 0.
Claim. The operator B has a block diagonal matrix in the basis {|an,i },
namely am,j | B |an,i = 0 for n = m.
Eyal Buks
35

Proof. Multiplying Eq. (2.140) from the left by am,j | B yields
am,j | BA |an,i = an am,j | B |an,i .
(2.144)
On the other hand, since [A, B] = 0 one has

am,j | BA |an,i = am,j | AB |an,i = am am,j | B |an,i ,
(2.145)
thus
(an am ) am,j | B |an,i = 0 .
(2.146)
For a given n, the gn gn matrix an,i | B |an,i is Hermitian, namely
an,i | B |an,i = an,i | B |an,i . Thus, there exists a unitary transformation

Un , which maps Fn onto Fn , and which diagonalizes the block of B in the
subspace Fn . Since Fn is an eigensubspace of A, the block matrix of A in the
new basis remains diagonal (with the eigenvalue an ). Thus, we conclude that
a complete and orthonormal basis of common eigenvectors of both operators
A and B exists. For such a basis, which is denoted as {|n, m}, the following
holds
A |n, m = an |n, m ,
B |n, m = bm |n, m .
(2.147)
(2.148)
2.13 Uncertainty Principle

Consider a quantum system in a state |n, m, which is a common eigenvector
of the commuting observables A and B. The outcome of a measurement of
the observable A is expected to be an with unity probability, and similarly,
the outcome of a measurement of the observable B is expected to be bm
with unity probability. In this case it is said that there is no uncertainty
corresponding to both of these measurements.
Definition 2.13.1. The variance in a measurement
'
(of a given observable A
2
of a quantum system in a state | is given by (A) , where A = AA,
namely
'
( '
( ) *
(2.149)
(A)2 = A2 2A A + A2 = A2 A2 ,
where
A = | A | ,
) 2*
A = | A2 | .
Eyal Buks
(2.150)
(2.151)
36
2.13. Uncertainty Principle

Example 2.13.1. Consider a spin 1/2 system in a state | = |+;
z. Using
Eqs. (2.99), (2.102) and (2.134) one finds that
'
( ) *
2
2
(Sz ) = Sz2 Sz = 0 ,
(2.152)
'
( ) *
1
(2.153)
(Sx )2 = Sx2 Sx 2 = 2 .
4
The last example raises the question: can one find a state | for which
the variance in the measurements of both Sz and Sx vanishes? According to
the uncertainty principle the answer is no.
Theorem 2.13.1. The uncertainty principle - Let A and B be two observables. For any ket-vector | the following holds
'
('
( 1
(A)2 (B)2 |[A, B]|2 .
(2.154)
4
Proof. Applying the Schwartz inequality [see Eq. (2.166)], which is given by
u |u v |v |u |v|2 ,
(2.155)
for the ket-vectors

|u = A | ,
|v = B | ,
(2.156)
(2.157)
and exploiting the fact that (A) = A and (B) = B yield

'
('
(
(A)2 (B)2 |AB|2 .
(2.158)
The term AB can be written as

AB =
1
1
[A, B] + [A, B]+ ,
2
2
(2.159)
where
[A, B] = AB BA ,
[A, B]+ = AB + BA .
(2.160)
(2.161)
While the term [A, B] is anti-Hermitian, whereas the term [A, B]+ is
Hermitian, namely
([A, B]) = (AB BA) = BA AB = [A, B] ,

[A, B]+ = (AB + BA) = BA + AB = [A, B]+ .
In general, the following holds

+
| X |
if X is Hermitian
| X | = | X | =
, (2.162)
| X | if X is anti-Hermitian
Eyal Buks
37

thus
*
1)
1
AB = [A, B] + [A, B]+ ,

2
2

I
(2.163)
and consequently
|AB|2 =
*
2
1
1
)
|[A, B]|2 +
[A, B]+
.
4
4
(2.164)
Finally, with the help of the identity [A, B] = [A, B] one finds that
'
('
( 1
2
(B)2 |[A, B]|2 .
(2.165)
(A)
4
2.14 Problems
1. Derive the Schwartz inequality

|u |v| u |u v |v ,
where |u and |v are any two vectors of a vector space F.
2. Derive the triangle inequality:

(u| + v|) (|u + |v) u |u + v |v .
(2.166)
(2.167)
3. Show that if a unitary operator U can be written in the form U = 1+iF ,

where is a real infinitesimally small number, then the operator F is
Hermitian.
4. A Hermitian operator A is said to be positive-definite if, for any vector
|u, u| A |u 0. Show that the operator A = |a a| is Hermitian and
positive-definite.
5. Show that if A is a Hermitian positive-definite operator then the following
hold

|u| A |v| u| A |u v| A |v .
(2.168)
6. Find the expansion of the operator (A B)1 in a power series in ,

assuming that the inverse A1 of A exists.
7. The derivative of an operator A () which depends explicitly on a parameter is defined to be
dA ()
A ( + ) A ()
= lim
.
0
d
(2.169)
Show that
d
dA
dB
(AB) =
B+A
.
d
d
d
Eyal Buks
(2.170)
38
2.14. Problems
8. Show that
d 1
dA 1
A
= A1
A .
d
d
(2.171)
9. Let |u and |v be two vectors of finite norm. Show that
Tr (|u v|) = v |u .
(2.172)
10. If A is any linear operator, show that A A is a positive-definite Hermitian operator whose trace is equal to the sum of the square moduli of
the matrix
elements of A in any arbitrary representation. Deduce that
Tr A A = 0 is true if and only if A = 0.
11. Show that if A and B are two positive-definite observables, then Tr (AB)
0.
12. Show that for any two operators A and L
eL AeL = A + [L, A] +
1
1
[L, [L, A]] + [L, [L, [L, A]]] + . (2.173)
2!
3!
13. Show that if A and B are two operators satisfying the relation [[A, B] , A] =
0 , then the relation
[Am , B] = mAm1 [A, B]
(2.174)
holds for all positive integers m .

14. Show that
eA eB = eA+B e(1/2)[A,B] ,
(2.175)
provided that [[A, B] , A] = 0 and [[A, B] , B] = 0.

15. Proof Kondos identity
,
A, eH = eH
eH [H, A] eH d ,
(2.176)
where A and H are any two operators and is real.

16. Show that Tr (XY ) = Tr (Y X).
17. Consider the two normalized spin 1/2 states | and |. The operator
A is defined as
A = | | | | .
(2.177)
Find the eigenvalues of the operator A.

18. A molecule is composed of six identical atoms A1 , A2 , . . . , A6 which
form a regular hexagon. Consider an electron, which can be localized on
each of the atoms. Let |n be the state in which it is localized on the
nth atom (n = 1, 2, , 6). The electron states will be confined to the
Eyal Buks
39

space spanned by the states |n , which is assumed to be orthonormal.
The Hamiltonian of the system is given by
H = H0 + W .
(2.178)
The eigenstates of H0 are the six states |n , with the same eigenvalue
E0 . The operator W is described by
W |1 = a |2 a |6 ,
W |2 = a |3 a |1 ,
..
.
W |6 = a |1 a |5 .
(2.179)
Find the eigenvalues and eigen vectors of H. Clue: Consider a solution of

the form
|k =
6

n=1
eikn |n .
(2.180)
2.15 Solutions
1. Let
| = |u + |v ,
(2.181)
where C. The requirement | 0 leads to

u |u + u |v + v |u + ||2 v |v 0 .
(2.182)
By choosing
=
v |u
,
v |v
(2.183)
one has
v |u
2
v |u
u |v
v |v 0 ,
u |u
u |v
v |u +
v |v
v |v
v |v
(2.184)
thus
|u |v|
Eyal Buks

u |u v |v .
(2.185)
40
2.15. Solutions
2. The following holds
(u| + v|) (|u + |v) = u |u + v |v + 2 Re (u |v)
u |u + v |v + 2 |u |v| .
(2.186)
Thus, using Schwartz inequality one has

(u| + v|) (|u + |v) u |u + v |v + 2 u |u v |v

2

=
u |u + v |v .
(2.187)
3. Since

1 = U U = 1 iF (1 + iF ) = 1 + i F F + O 2 , (2.188)
one has F = F .
4. In general, (| |) = | |, thus clearly the operator A is Hermitian.

Moreover it is positive-definite since for every |u the following holds
u| A |u = u |a a |u = |a |u|2 0 .
(2.189)
5. Let
v| A |u
|v .
v| A |v
| = |u
Since A is Hermitian and positive-definite the following holds

0 | A |

u| A |v
v| A |u
= u|
v| A |u
|v
v| A |v
v| A |v
= u| A |u
|u| A |v|2 |u| A |v|2 |u| A |v|2
+
,
v| A |v
v| A |v
v| A |v
(2.190)
thus
|u| A |v|

u| A |u v| A |v .
(2.191)
Note that this result allows easy proof of the following: Under the same
conditions (namely, A is a Hermitian positive-definite operator) Tr (A) =
0 if and only if A = 0.
6. The expansion is given by

1
(A B)1 = A 1 A1 B

1 1
= 1 A1 B
A

1 2 1 3
1
= 1 + A B + A B + A B + A1 .
(2.192)
Eyal Buks
41

7. By definition:
A ( + ) B ( + ) A () B ()
d
(AB) = lim
0
d
(A ( + ) A ()) B ()
A ( + ) (B ( + ) B ())
+ lim
= lim
0
0
dA
dB
=
B+A
.
d
d
(2.193)
8. Taking the derivative of both sides of the identity 1 = AA1 on has
0=
dA 1
dA1
A +A
,
d
d
(2.194)
thus
d 1
dA 1
A
= A1
A .
d
d
(2.195)
9. Let {|n} be a complete orthonormal basis, namely

|n n| = 1 .
(2.196)
In this basis
Tr (|u v|) =

n
n |u v |n = v|
!

n
"
|n n| |u = v |u . (2.197)

10. The operator A A is Hermitian since A A = A A, and positivedefinite since the norm of A |u is nonnegative for every |u, thus one
has u| A A |u 0. Moreover, using a complete orthonormal basis {|n}
one has

Tr A A =
n| A A |n
n

n,m

n,m
n| A |m m| A |n

2
|m| A |n| .
(2.198)
11. Let {|b } be a complete orthonormal basis made of eigenvectors of B

(i.e., B |b = b |b ). Using this basis for evaluating the trace one has

Tr (AB) =
b | AB |b =
b b | A |b 0 .
(2.199)

Eyal Buks
42
2.15. Solutions
12. Let f (s) = esL AesL , where s is real. Using Taylor expansion one has
1 df
1 d2 f
f (1) = f (0) +
+
+ ,
(2.200)
1! ds
s=0 2! ds2
s=0
thus
eL AeL = A +
1 df
1 d2 f
+
+ ,
1! ds
s=0 2! ds2
s=0
where
df
= LesL AesL esL AesL L = [L, f (s)] ,
ds

d2 f
df
= [L, [L, f (s)]] ,
= L,
ds2
ds
(2.201)
(2.202)
(2.203)
therefore
eL AeL = A + [L, A] +
1
1
[L, [L, A]] + [L, [L, [L, A]]] + . (2.204)
2!
3!
13. The identity clearly holds for the case m = 1. Moreover, assuming it
holds for m, namely assuming that
[Am , B] = mAm1 [A, B] ,
(2.205)
one has
, m+1 A
, B = A [Am , B] + [A, B] Am
= mAm [A, B] + [A, B] Am .
(2.206)
It is easy to show that if [[A, B] , A] = 0 then [[A, B] , Am ] = 0, thus one
concludes that
, m+1 A
, B = (m + 1) Am [A, B] .
(2.207)
14. Define the function f (s) = esA esB , where s is real. The following holds
df
= AesA esB + esA BesB
ds

= A + esA BesA esA esB
Using Eq. (2.174) one has
Eyal Buks
43

sA
m

(sA)
B=
B
m!
m=0

sm (BAm + [Am , B])
m!
m=0

m
m
s BA + mAm1 [A, B]
=
m!
m=0
= BesA + s
m1

(sA)
[A, B]
(m 1)!
m=1
= BesA + sesA [A, B] ,
(2.208)
thus
df
= AesA esB + BesA esB + sesA [A, B] esB
ds
= (A + B + [A, B] s) f (s) .
(2.209)
The above differential equation can be easily integrated since [[A, B] , A] =
0 and [[A, B] , B] = 0. Thus
f (s) = e(A+B)s e[A,B]
s2
2
(2.210)
eA eB = eA+B e(1/2)[A,B] .
(2.211)
For s = 1 one gets
15. Define
,
f () A, eH ,
H
g () e
eH [H, A] eH d .
(2.212)
(2.213)
Clearly, f (0) = g (0) = 0 . Moreover, the following holds

df
= AHeH + HeH A = Hf + [H, A] eH ,
d
dg
= Hg + [H, A] eH ,
d
(2.214)
(2.215)
namely, both functions satisfy the same differential equation. Therefore

f = g.
16. Using a complete orthonormal basis {|n} one has
Eyal Buks
44
2.15. Solutions
Tr (XY ) =

n

n,m
n,m

m
n| XY |n
n| X |m m| Y |n
m| Y |n n| X |m
m| Y X |m
= Tr (Y X) .
(2.216)
+
Note that using this result it is easy to show that Tr (U XU ) = Tr (X)

, provided that U is a unitary operator.
17. Clearly A is Hermitian, namely A = A, thus the two eigenvalues 1
and 2 are expected to be real. Since the trace of an operator is basis
independent, the following must hold
Tr (A) = 1 + 2 ,
(2.217)
and

Tr A2 = 21 + 22 .
(2.218)
On the other hand, with the help of Eq. (2.172) one finds that
Tr (A) = Tr (| |) Tr (| |) = 0 ,
(2.219)
and

Tr A2 = Tr (| | |) + Tr (| | |) Tr (| | |) Tr (| | |)
= 2 | Tr (| |) | Tr (| |)

= 2 1 | ||2 ,
(2.220)
thus
.
= 1 | ||2 .
(2.221)
| = a | + c | ,
(2.222)
Alternatively, this problem can also be solved as follows. In general, the

state | can be decomposed into a parallel to and a perpendicular to |
terms, namely
where a, c C, the vector | is orthogonal to |, namely | = 0, and

in addition | is assumed to be normalized, namely | = 1. Since |
Eyal Buks
45

is normalized one has |a|2 + |c|2 = 1. The matrix representation of A in
the orthonormal basis {| , |} is given by

2

| A | | A |
|c| ac
A .
(2.223)
A=
=
| A | | A |
a c |c|2
Thus,
and

Tr A = 0 ,
(2.224)

2
2
2
2
Det A = |c| |c| + |a| = 1 | || ,
(2.225)
therefore the eigenvalues are

.
= 1 | ||2 .
(2.226)
18. Following the clue, we seek a solution to the eigenvalue equation

H |k = Em |k ,
(2.227)
where
|k =
6

n=1
eikn |n .
(2.228)
thus
H |k = E0 |k a
6

n=1
(
eikn
(n1) +
(n+1)
= E |k , (2.229)
where n is the modulus of n divided by 6 (e.g., 1 = 1, 0 = 6, 7 = 1).

A solution is obtained if
e6ik = 1 ,
(2.230)
or
km =
m
,
3
(2.231)
where m = 1, 2, , 6. The corresponding eigen vectors are denoted as

|km =
6

n=1
eikm n |n ,
(2.232)
and the following holds

Eyal Buks
46
2.15. Solutions
H |km = E0 |km aeikm
6

n=1
(
(

eikm (n1)
(n1) aeikm
eikm (n+1)
(n+1)
n=1
= (E0 2a cos km ) |km ,
(2.233)
thus
H |km = Em |k ,
(2.234)
where
Em = E0 2a cos km .
Eyal Buks
(2.235)
47
3. The Position and Momentum Observables
Consider a point particle moving in a 3 dimensional space. We first treat

the system classically. The position of the particle is described using the
Cartesian coordinates qx , qy and qz . Let
pj =
L
qj
(3.1)
be the canonically conjugate variable to the coordinate qj , where j {x, y, z}

and where L is the Lagrangian. As we have seen in exercise 4 of set 1, the
following Poissons brackets relations hold
{qj , qk } = 0 ,
{pj , pk } = 0 ,
{qj , pk } = jk .
(3.2)
(3.3)
(3.4)
In quantum mechanics, each of the 6 variables qx , qy , qz , px , py and pz is

represented by an Hermitian operator, namely by an observable. It is postulated that the commutation relations between each pair of these observables
is related to the corresponding Poissons brackets according to the rule
{, }
1
[, ] .
i
(3.5)
Namely the following is postulated to hold

[qj , qk ] = 0 ,
[pj , pk ] = 0 ,
[qj , pk ] = i jk .
(3.6)
(3.7)
(3.8)
Note that here we use the same notation for a classical variable and its
quantum observable counterpart. In this chapter we will derive some results
that are solely based on Eqs. (3.6), (3.7) and (3.8).
3.1 The One Dimensional Case

In this section, which deals with the relatively simple case of a one dimensional motion of a point particle, we employ the less cumbersome notation
Chapter 3. The Position and Momentum Observables

x and p for the observables qx and px . The commutation relation between
these operators is given by [see Eq. (3.8)]
[x, p] = i .
(3.9)
The uncertainty principle (2.154) employed for x and p yields

('
( 2
'
2
2
(px )
.
(x)
4
(3.10)
3.1.1 Position Representation

Let x be an eigenvalue of the observable x, and let |x be the corresponding
eigenvector, namely
x |x = x |x .
(3.11)
Note that x R since x is Hermitian. As we will see below transformation

between different eigenvectors |x can be performed using the translation
operator J (x ).
Definition 3.1.1. The translation operator is given by

ix p
,
J (x ) = exp
(3.12)
where x R.
Recall that in general the meaning of a function of an operator can be
understood in terms of the Taylor expansion of the function, that is, for the
present case

n

1
ix p
J (x ) =
.
n!

n=0
(3.13)
It is easy to show that J (x ) is unitary

J (x ) = J (x ) = J 1 (x ) .
(3.14)
Moreover, the following composition property holds

J (x1 ) J (x2 ) = J (x1 + x2 ) .
(3.15)
Theorem 3.1.1. Let x be an eigenvalue of the observable x, and let |x be

the corresponding eigenvector. Then the ket-vector J (x ) |x is a normalized
eigenvector of x with an eigenvalue x + x .
Eyal Buks
50
3.1. The One Dimensional Case

Proof. With the help of Eq. (3.77), which is given by
[x, B (p)] = i
dB
,
dp
(3.16)
and which is proven in exercise 1 of set 3, one finds that

[x, J (x )] = i
x
J (x ) .
i
(3.17)
Using this result one has

xJ (x ) |x = ([x, J (x )] + J (x ) x) |x = (x + x ) J (x ) |x , (3.18)
thus the ket-vector J (x ) |x is an eigenvector of x with an eigenvalue x +
x . Moreover, J (x ) |x is normalized since J is unitary.
In view of the above theorem we will in what follows employ the notation
J (x ) |x = |x + x .
(3.19)
An important consequence of the last result is that the spectrum of eigenvalues of the operator x is continuous and contains all real numbers. This point
will be further discussed below.
The position wavefunction (x ) of a state vector | is defined as:
(x ) = x | .
(3.20)
Given the wavefunction (x ) of state vector |, what is the wavefunction

of the state O |, where O is an operator? We will answer this question below
for some cases:
1. The operator O = x. In this case
x | x | = x x | = x (x ) ,
(3.21)
namely, the desired wavefunction is obtained by multiplying (x ) by

x .
2. The operator O is a function A (x) of the operator x. Let

A (x) =
an xn .
(3.22)
n
be the Taylor expansion of A (x). Exploiting the fact that x is Hermitian

one finds that

x | A (x) | =
an x | xn | =
an xn x | = A (x ) (x ) .

n
n
xn x |
Eyal Buks
(3.23)
51

3. The operator O = J (x ). In this case
x | J (x ) | = x | J (x ) | = x x | = (x x ) .
(3.24)
4. The operator O = p. In view of Eq. (3.12), the following holds

ipx
ix
=1+
p + O (x )2 ,
J (x ) = exp

(3.25)
thus
x | J (x ) | = (x ) +

ix
2
x | p | + O (x ) .
(3.26)
On the other hand, according to Eq. (3.24) also the following holds
x | J (x ) | = (x + x ) .
(3.27)
Equating the above two expressions for x | J (x ) | yields

x | p | = i
(x + x ) (x )
+ O (x ) .
x
(3.28)
Thus, in the limit x 0 one has

x | p | = i
d
.
dx
(3.29)
To mathematically understand the last result, consider the differential

operator

d
J (x ) = exp x
dx

2
1
d
d
= 1 + x
+
x
+ .
dx 2!
dx
(3.30)
In view of the Taylor expansion of an arbitrary function f (x)
df
(x )2 d2 f
f (x0 + x ) = f (x0 ) + x
+
+
dx
2! dx2

d
= exp x
f
dx
x=x0
= J (x ) f
,
x=x0
(3.31)
one can argue that the operator J (x ) generates translation.

Eyal Buks
52
3.1. The One Dimensional Case

As we have pointed out above, the spectrum (i.e., the set of all eigenvalues) of x is continuous. On the other hand, in the discussion in chapter
2 only the case of an observable having discrete spectrum has been considered. Rigorous mathematical treatment of the case of continuous spectrum
is nontrivial mainly because typically the eigenvectors in such a case cannot
be normalized. However, under some conditions one can generalize some of
the results given in chapter 2 for the case of an observable having a continuous spectrum. These generalization is demonstrated below for the case of the
position operator x:
1. The closure relation (2.23) is written in terms of the eigenvectors |x as
dx |x x | = 1 ,
(3.32)
namely, the discrete sum is replaced by an integral.

2. With the help of Eq. (3.32) an arbitrary ket-vector can be written as
| =
dx |x x | =
dx (x ) |x ,
(3.33)
and the inner product between a ket-vector | and a bra-vector | as

| =
dx |x x | =
dx (x ) (x ) .
(3.34)
3. The normalization condition reads

1 = | =
dx | (x )| .
(3.35)
4. The orthonormality relation (2.64) is written in the present case as

x |x = (x x ) .
(3.36)
Note that the above orthonormality relation (3.36) is consistent with

the closure relation (3.32). This can be seen by evaluating the operator
12 = 1 1 using Eqs. (3.32) and (3.36)
2
1 =
dx
dx |x x |x x | =

(x x )
dx |x x | ,
(3.37)
thus, as expected 12 = 1.
Eyal Buks
53

5. In a measurement of the observable x, the quantity
2
f (x ) = |x || = | (x )|
(3.38)
represents the probability distribution function to find the particle at the

point x = x .
6. That is, the probability to find the particle in the interval (x1 , x2 ) is given
by
x2
p(x1 ,x2 ) =
dx f (x ) .
(3.39)
x1
This can be rewritten as

p(x1 ,x2 ) = | P(x1 ,x2 ) | ,
(3.40)
where the projection operator P(x1 ,x2 ) is given by

x2
P(x1 ,x2 ) =
dx |x x | .
(3.41)
x1
The operator P(x1 ,x2 ) is considered to be a projection operator since for

every x0 (x1 , x2 ) the following holds
x2
P(x1 ,x2 ) |x0 =
dx |x x |x0 = |x0 .
(3.42)

x1
(x x0 )
7. Any realistic measurement of a continuous variable such as position is

subjected to finite resolution. Assuming that a particle has been measured to be located in the interval (x x /2, x + x /2), where x is the
resolution of the measuring device, the collapse postulate implies that
the state of the system undergoes the following transformation
P(x x /2,x +x /2) |
|
.
| P(x x /2,x +x /2) |
(3.43)
8. Some observables have a mixed spectrum containing both a discrete and

continuous subsets. An example of such a mixed spectrum is the set of
eigenvalues of the Hamiltonian operator of a potential well of finite depth.
3.1.2 Momentum Representation
Let p be an eigenvalue of the observable p, and let |p be the corresponding
eigenvector, namely
p |p = p |p .
Eyal Buks
(3.44)
54
3.2. Transformation Function

Note that p R since p is Hermitian. Similarly to the case of the position
observable, the closure relation is written as

dp |p p | = 1 ,
(3.45)
and the orthonormality relation as
p |p = (p p ) .
(3.46)
The momentum wavefunction (p ) of a given state | is defined as

(p ) = p | .
(3.47)
The probability distribution function to measure a momentum value of p = p

is
2
| (p )| = |p || .
(3.48)
Any ket-vector can be decomposed into momentum eigenstates as

| =
dp |p p | =
dp (p ) |p .
(3.49)
The inner product between a ket-vector | and a bra-vector | can be

expressed as
| =
dp |p p | =
dp (p ) (p ) .
(3.50)
The normalization condition reads

1 = | =
dp | (p )| .
(3.51)
3.2 Transformation Function

What is the relation between the position wavefunction (x ) and its momentum counterpart (p )?
Claim. The transformation function x |p is given by

1
ip x
x |p =
exp
.

2
Eyal Buks
(3.52)
55

Proof. On one hand, according to Eq. (3.44)
x | p |p = p x |p ,
(3.53)
and on the other hand, according to Eq. (3.29)

x | p |p = i
x |p ,
x
(3.54)
thus
x |p .
x
The general solution of this differential equation is

ip x
,
x |p = N exp

p x |p = i
(3.55)
(3.56)
where N is a normalization constant. To determine the constant N we employ

Eqs. (3.36) and (3.45):
(x x )
= x |x

= dp x |p p |x
=
dp |N | exp
2
= |N |
ip (x x )
dkeik(x x ) .

2(x x )
(3.57)
Thus, by choosing N to be real one finds that

1
ip x
x |p =
exp
.

2
(3.58)
The last result together with Eqs. (3.32) and (3.45) yield
/ ip x
dp e (p )
dp x |p p | =
(x ) = x | =
,
2
(p ) = p | =
(3.59)
ip x
dx e (x )
dx p |x x | =
. (3.60)
2
That is, transformations relating (x ) and (p ) are the direct and inverse
Fourier transformations.
Eyal Buks
56
3.3. Generalization for 3D
3.3 Generalization for 3D

According to Eq. (3.6) the observables qx , qy and qz commute with each
other, hence, a simultaneous diagonalization is possible. Denoting the common eigenvectors as
*
(3.61)
|r =
qx , qy , qz ,
one has
*
qx |r = qx
qx , qy , qz ,
*
qy |r = qy
qx , qy , qz ,
*
qz |r = qz
qx , qy , qz .
(3.62)
(3.63)
(3.64)
The closure relation is written as

1=

dx dy dz |r r | ,
(3.65)

r |r = (r r ) .
(3.66)
Similarly, according to Eq. (3.7) the observables px , py and pz commute

with each other, hence, a simultaneous diagonalization is possible. Denoting
the common eigenvectors as
*
|p =
px , py , pz ,
(3.67)
one has
*
px |p = px
px , py , pz ,
*
py |p = py
px , py , pz ,
*
pz |p = pz
px , py , pz .
(3.68)
(3.69)
(3.70)
The closure relation is written as

1=

dpx dpy dpz |p p | ,
(3.71)

p |p = (p p ) .
(3.72)
The translation operator in three dimensions can be expressed as

i p
J () = exp
,
(3.73)

Eyal Buks
57

where = (x , y , z ) R3 , and where
J () |r = |r + .
The generalization of Eq. (3.52) for three dimensions is

1
ip r
r |p =
exp
.
3/2

(2)
(3.74)
(3.75)
3.4 Problems
1. Show that
dA
,
dx
dB
[x, B (p)] = i
,
dp
[p, A (x)] = i
(3.76)
(3.77)
where A (x) is a differentiable function of x and B (p) is a differentiable

function of p.
2. Show that the mean value of x in a state described by the wavefunction
(x), namely
+

x =
dx (x) x (x) ,
(3.78)
is equal to the value of a for which the expression

+

dx (x + a) x2 (x + a)
F (a)
(3.79)
obtains a minimum, and that this minimum has the value

) *
Fmin = (x)2 = x2 x2 .
(3.80)
3. Consider a Gaussian wave packet, whose x space wavefunction is given

by

1
x2
(x ) =
exp ikx 2 .
(3.81)
2d
1/4 d
Calculate
'
('
(
a) (x)2 (p)2
b) p |
Eyal Buks
58
3.4. Problems
4. Show that the state | with wave function
+
1/ 2a for |x| a
x | =
0 for |x| > a
(3.82)
the uncertainty in momentum is infinity.

5. Show that
p = i
dx |x
d
x | .
dx
(3.83)
6. Show that
1
(2)
d3 p exp
ip (r r )
= (r r ) .
(3.84)
7. Find eigenvectors and corresponding eigenvalues of the operator

O = p + Kx ,
(3.85)
where K is a real constant, p is the momentum operator, which is canonically conjugate to the position operator x. Calculate the wavefunction
of the eigenvectors.
8. Let | be the state vector of a point particle having mass m that moves
in one dimension along the x axis. The operator p is defined by the
following requirements: (1) p is Hermitian (i.e. p = p ) (2) [x, p ] = 0
(i.e. p commutes with the position operator x) and (3)
| (p p )2 | = min | (p O)2 | ,
O
(3.86)
where p is the momentum operator (i.e. the minimum value of the quantity | (p O)2 | is obtained when the operator O is chosen to be
p ).
a) Calculate the matrix elements x | p |x of the operator p in the
position representation.
b) The operator P is the difference between the true momentum operator and p
P = p p .
(3.87)
'
(
(P)2 = | P 2 | | P |2 .
(3.88)
'
(
Calculate the variance (P)2 with respect to the state |
Eyal Buks
59

c) Use your results to prove the uncertainty relation (3.10)
'
('
( 2
(x)2 (p)2
.
4
where
(
'
(x)2 = | x2 | | x |2 ,
and where
'
(
(p)2 = | p2 | | p |2 .
(3.89)
(3.90)
(3.91)
3.5 Solutions
1. The commutator [x, p] = i is a constant, thus the relation (2.174) can
be employed
[p, xm ] = imxm1 = i
[x, pm ] = impm1 = i
dxm
,
dx
(3.92)
dp
.
dp
(3.93)
This holds for any m, thus, for any differentiable function A (x) of x and
for any differentiable function B (p) of p one has
dA
,
dx
dB
[x, B (p)] = i
.
dp
[p, A (x)] = i
(3.94)
(3.95)

+

F (a) =
dx (x + a) x2 (x + a)
+

2
=
dx (x ) (x a) (x)
'
(
= (x a)2
) *
= x2 2a x + a2 .
(3.96)
The requirement
dF
=0
da
Eyal Buks
(3.97)
60
3.5. Solutions
leads to 2 x + 2a = 0, or a = x. At that point one has
'
( ) *
2
2
Fmin = (x x) = x2 x .
(3.98)
3. The following hold
+

x =
dx (x ) x (x )
+
2

x
= 1/2
dx exp 2 x
d
d
=0,
(3.99)
)
+

=
dx (x ) x2 (x )
+
2

x
dx exp 2 x2
= 1/2
d
d
d3 1/2
1/2 d 2
d2
=
,
2
(3.100)
+

d
p = i
dx (x )
dx
+
2

x
x
= 1/2
dx exp 2
ik 2
d
d
d
i
i
1/2 d
= k ,
ikd1/2
(3.101)
Eyal Buks
61

) 2*
p = (i)2
+

d2
dx (x )
dx2
""
+
2
2 !!

x
x
1
ik 2
= (i) 1/2
dx exp 2
2
d
d
d
d

2d4 k2 + d2
1
1
= (i)2 1/2
d
2
d4
d
!
"
1
2
,
= (k) 1 +
2
2 (dk)
2
(3.102)
a) thus
'
('
( d2
(x)2 (p)2 =
2
(k)
1+
1
2 (dk)
"
(k)
"
2
.
4
(3.103)
b) Using Eq. (3.60) onehas

1
ip x
p | =
dx exp
(x )

2

ip
x2
1
1
=
dx exp
ik
x 2

2d
2 1/4 d
!
"
d
(k p )2 d2
exp
=
.
22
1/4
(3.104)
4. The momentum wavefunction is found using Eq. (3.60)

1
ip x
(p ) =
dx exp
x |

2
a

1
ip x
=
dx exp

4a
a
0
a sin ap

=
.
ap

(3.105)
The momentum wavefunction )(p )*is normalizable, however, the integrals for evaluating both p and p2 do not converge.
Eyal Buks
62
3.5. Solutions
5. Using Eqs. (3.29) and (3.32) one has

p | =
dx |x x | p |
= i
dx |x
d
x | ,
dx
(3.106)
thus, since | is an arbitrary ket vector, the following holds

p = i
dx |x
d
x | .
dx
(3.107)
6. With the help of Eqs. (3.66), (3.71) and (3.75) one finds that
(r r ) = r |r

= d3 p r |p p |r

1
ip (r r )
3
=
d
p
exp
.

(2)3
(3.108)
7. Using the identity (2.173), which is given by

eL AeL = A + [L, A] +
1
1
[L, [L, A]] + [L, [L, [L, A]]] + . (3.109)
2!
3!
and the identity (3.76), which is given by

[g (x) , p] = i
dg
,
dx
(3.110)
one finds that

g(x)
g(x)
pe

dg i
dg
i
dg
= p + i +
g (x) ,
+
g (x) , g (x) ,
+ .
dx 2!
dx
3!
dx
(3.111)
Choosing g (x) to be given by

g (x) =
Kx2
2i
(3.112)
yields
UpU = p + Kx = O ,
Eyal Buks
(3.113)
63

where the unitary operator U is given by
U = e
iKx2
2
Thus, the vectors | (p ), which are define as

| (p ) = U |p ,
(3.114)
where |p is an eigenvector of p with eigenvalue p (i.e. p |p = p |p ),

are eigenvectors of O, and the following holds
O | (p ) = p | (p ) .
(3.115)
With the help of Eq. (3.52), which is given by

x |p =
ip x
1
e ,
2
(3.116)
one finds that the wavefunction (x ; p ) = x | (p ) of the state

| (p ) is given by
(x ; p ) = e
iKx2
2
x |p
i
1
e
=
2
p x Kx
2
.
(3.117)
8. With the help of Eq. (3.32) one finds that

p =
dx
dx |x x | p |x x | .
(3.118)
The requirement [x, p ] = 0 implies that

dx
dx |x x | p |x (x x ) x | = 0 ,
(3.119)
hence x | p |x = 0 unless x = x . Thus by using the notation

x | p |x = (x ) (x x ) ,
(3.120)
the operator p can be expressed as

p =
dx |x (x ) x | .
(3.121)
The requirement that p is Hermitian implies that (x ) is real.

Eyal Buks
64
3.5. Solutions
a) With the help of Eq. (3.83), which is given by
p = i
d
x | ,
dx
dx |x
(3.122)
one finds that

p p =

d
dx |x i (x ) x | ,
dx
(3.123)
hence in terms of the wavefunction (x ) = x | of | one has

d
(p p ) | =
dx |x i (x ) (x )
dx
=

d log
dx |x (x ) i
(x ) .
dx
(3.124)
Similarly
| (p p ) =
dx
(x )

d log
i
(x ) x | , (3.125)
dx
and thus
| (p p )2 |
2

d log

=
dx (x )
i
+ (x )
.
dx
(3.126)
where
(x ) = | (x )| .
(3.127)
The minimum value is obtained when (recall that (x ) is required

to be real)

d log d log
(x ) =
2i
dx
dx
d log
,
2i dx
(3.128)
and thus
Eyal Buks
65
d log
(x x ) .
2i dx
x | p |x =
(3.129)
Note: Comparing this result with the expression for the current density J associated with the state | [see Eq. (4.200)] yields the following relation

d
Im
J=
m
dx

(x ) d log d log
=
m 2i
dx
dx
(x )
=
(x ) .
m
(3.130)
b) As can be seen from Eqs. (3.123) and (3.128) the following holds
d log
d
1

dx |x +
x | ,
(3.131)
P = i
dx
2 dx
hence
| P | = i
i
=
2
=
i
2
=0,

d
1
d
d
dx +

dx
2
dx
dx

d
d
dx +
dx
dx
dx
d (x )
dx
(3.132)
thus'[see Eqs.
( (3.126) and (3.128)]
2
(P) = | P 2 |
2
2

d log

d log
=
dx (x )
+
2
dx
dx
2
2

d log (x )
=
dx (x )
.
2
dx
(3.133)
Note that the result | P | = 0 implies that p and p have the
same expectation value, i.e. | p | = | p |. On the other hand,
Eyal Buks
66
3.5. Solutions
contrary to p, the operator p commutes with the position operator
x.
c) '
Using the
|
( relation
'
( p | = | p | one finds that
2
2
(p) (p ) = | p2 | | p2 |

= | (p p )2 | + | pp + p p 2p2 |

= | P 2 | + | pp + p p 2p2 | .
(3.134)
As can be see from Eq. (3.128), the following holds
p =
dx |x Im
d log
dx
x | ,
(3.135)
thus

| pp + p p 2p2 |
|p x
| x |p|

+
d
log
d log
=
dx
2 Im
Im

dx
dx
=0,
(3.136)
and 'therefore
'
(
(
(p)2 = | P 2 | + (p )2
| P 2 |
2

2

d log (x )
=
dx (x )
.
2
dx
(3.137)
For general real functions f (x ) , g (x ) : R R the Schwartz inequality (2.166) implies that

2
2
dx f (x ) g (x )

dx (f (x ))
dx (g (x )) . (3.138)
Implementingthis inequality for the functions

f (x ) = (x ) (x x) ,

d log (x )
g (x ) = (x )
,
dx
where
x =
Eyal Buks
(3.139)
(3.140)
dx (x ) x
67

is the expectation value of x, yields
d
log
x
(
)
dx (x ) (x x) dx
2

d log (x )
'
(
dx (x )
,
2
dx
(x)
(3.141)
where

'
(
2
2
dx (x ) (x x)
(x) =
(3.142)
is the variance of x. By integrating by parts one finds that

d log (x )
d (x )
dx (x ) (x x)
=
dx (x x)
dx
dx
dx (x )
= 1 .
(3.143)
Combining these results [see Eqs. (3.137) and (3.141)] lead to
'
('
( 2
(x)2 (p)2
.
2
Eyal Buks
(3.144)
68
4. Quantum Dynamics
The time evolution of a state vector | is postulated to be given by the

Schrdinger equation
i
d |
= H | ,
dt
(4.1)
where the Hermitian operator H = H is the Hamiltonian of the system.

The Hamiltonian operator is the observable corresponding to the classical
Hamiltonian function that we have studied in chapter 1. The time evolution
produced by Eq. (4.1) is unitary, as is shown below:
Claim. The norm | is time independent.
Proof. Since H = H , the dual of the Schrdinger equation (4.1) is given by
i
d |
= | H .
dt
(4.2)
Using this one has

d |
d |
1
d |
=
|+|
=
( | H | + | H |) = 0 . (4.3)
dt
dt
dt
i
4.1 Time Evolution Operator

The time evolution operator u (t, t0 ) relates the state vector at time | (t0 )
with its value | (t) at time t:
| (t) = u (t, t0 ) | (t0 ) .
(4.4)
Claim. The time evolution operator satisfies the Schrdinger equation (4.1).
Proof. Expressing the Schrdinger equation (4.1) in terms of Eq. (4.4)
i
d
u (t, t0 ) | (t0 ) = Hu (t, t0 ) | (t0 ) ,
dt
and noting that | (t0 ) is t independent yield
(4.5)
Chapter 4. Quantum Dynamics

i

d
u (t, t0 ) | (t0 ) = Hu (t, t0 ) | (t0 ) .
dt
(4.6)
Since this holds for any | (t0 ) one concludes that

i
du (t, t0 )
= Hu (t, t0 ) .
dt
(4.7)
This results leads to the following conclusion:

Claim. The time evolution operator is unitary.
Proof. Using Eq. (4.7) one finds that

d u u
du du
= u
+
u
dt
dt
dt

1
u Hu u Hu
=
i
=0.
(4.8)
Furthermore, for t = t0 clearly u (t0 , t0 ) = u (t0 , t0 ) = 1. Thus, one concludes

that u u = 1 for any time, namely u is unitary.
4.2 Time Independent Hamiltonian

A special case of interest is when the Hamiltonian is time independent. In
this case the solution of Eq. (4.7) is given by

iH (t t0 )
u (t, t0 ) = exp
.
(4.9)

The operator u (t, t0 ) takes a relatively simple form in the basis of eigenvectors
of the Hamiltonian H. Denoting these eigenvectors as |an,i , where the index
i is added to account for possible degeneracy, and denoting the corresponding
eigenenergies as En one has
H |an,i = En |an,i ,
(4.10)
where
an ,i |an,i = nn ii .
(4.11)
By using the closure relation, which is given by

1=
gn

n i=1
Eyal Buks
|an,i an,i | ,
(4.12)
70

and Eq. (4.9) one finds that

iH (t t0 )
u (t, t0 ) = exp
1

gn

iH (t t0 )
=
exp
|an,i an,i |

n i=1

gn

iEn (t t0 )
exp
=
|an,i an,i | .

n i=1
(4.13)
Using this results the state vector | (t) can be written as

| (t) = u (t, t0 ) | (t0 )

gn

iEn (t t0 )
=
exp
an,i | (t0 ) |an,i .

n i=1
(4.14)
Note that if the system is initially in an eigenvector of the Hamiltonian

with eigenenergy En , then according to Eq. (4.14)

iEn (t t0 )
| (t) = exp
| (t0 ) .
(4.15)

However, the phase factor multiplying | (t0 ) has no effect on any measurable physical quantity of the system, that is, the systems properties are
time independent. This is why the eigenvectors of the Hamiltonian are called
stationary states.

In classical mechanics, the potential energy U of a magnetic moment in a
magnetic field B is given by
U = B .
(4.16)
The magnetic moment of a spin 1/2 is given by [see Eq. (2.90)]

spin =
2B
S,
(4.17)
where S is the spin angular momentum vector and where

B =
e
2me c
Eyal Buks
(4.18)
71

is the Bohrs magneton (note that the electron charge is taken to be negative
e < 0). Based on these relations we hypothesize that the Hamiltonian of a
spin 1/2 in a magnetic field B is given by
e
H=
SB.
(4.19)
me c
Assume the case where
B = B
z,
(4.20)
where B is a constant. For this case the Hamiltonian is given by

H = Sz ,
(4.21)
where
=
|e| B
me c
(4.22)
is the so-called Larmor frequency. In terms of the eigenvectors of the operator

Sz

Sz | = | ,
(4.23)
2
where the compact notation | stands for the states |;
z, one has

| ,
(4.24)
2
namely the states | are eigenstates of the Hamiltonian. Equation (4.13) for
the present case reads
H | =
u (t, 0) = e
it
2
|+ +| + e
it
2
| | .
(4.25)
Exercise 4.3.1. Consider spin 1/2 in magnetic field given by B = B

z, where
B is a constant. Given that | (0) = |+; x
at time t = 0 calculate (a) the
probability p (t) to measure Sx = /2 at time t; (b) the expectation value
Sx (t) at time t.
Solution 4.3.1. Recall that [see Eq. (2.102)]
1
|; x
= (|+ |)
2
(4.26)
(a) Using Eq. (4.25) one finds

p (t) = |; x
| u (t, 0) | (0)|2
2
it
it
2
2
=
(+| |) e
|+ +| + e | | (|+ + |)

1
it
it
=
e 2 e 2
,
2
(4.27)
Eyal Buks
72
4.4. Connection to Classical Dynamics

thus

t
,
2

t
p (t) = sin2
.
2
p+ (t) = cos2
(4.28)
(4.29)
(b)Using the results for p+ and p one has

(p+ p )
2

t
2 t
2
=
cos
sin
2
2
2

= cos (t) .
2
Sx =
(4.30)
4.4 Connection to Classical Dynamics

In chapter 1 we have found that in classical physics, the dynamics of a variable
A(c) is governed by Eq. (1.38), which is given by
dA(c) 1 (c) (c) 2 A(c)
+
= A ,H
.
dt
t
(4.31)
We seek a quantum analogy to this equation. To that end, we derive an

equation of motion for the expectation value A of the observable A that
corresponds to the classical variable A(c) . In general, the expectation value
can be expressed as
A = (t)| A | (t) = (t0 )| u (t, t0 ) Au (t, t0 ) | (t0 ) = (t0 )| A(H) | (t0 ) ,
(4.32)
where u is the time evolution operator and
A(H) = u (t, t0 ) Au (t, t0 ) .
(4.33)
The operator A(H) is called the Heisenberg representation of A. We first

derive an equation of motion for the operator A(H) . By using Eq. (4.7) one
finds that the following holds
du
1
= Hu ,
dt
i
du
1
= u H ,
dt
i
therefore
Eyal Buks
(4.34)
(4.35)
73

dA(H)
du
du
A
=
Au + u A
+ u
u
dt
dt
dt
t

1
A
=
u HAu + u AHu + u
u
i
t

1
A
=
u Huu Au + u Auu Hu + u
u
i
t

1
A(H)
=
H(H) A(H) + A(H) H(H) +
.
i
t
(4.36)
Thus, we have found that

dA(H)
1 3 (H) (H) 4 A(H)
=
A ,H
+
.
dt
i
t
(4.37)
Furthermore, the desired equation of motion for A is found using Eqs. (4.32)
and (4.37)
5
6
d A
1
A
=
[A, H] +
.
(4.38)
dt
i
t
We see that the Poissons brackets in the classical equation of motion (4.31)
for the classical variable A(c) are replaced by a commutation relation in the
quantum counterpart equation of motion (4.38) for the expectation value A
{, }
1
[, ] .
i
(4.39)
Note that for the case where the Hamiltonian is time independent, namely
for the case where the time evolution operator is given by Eq. (4.9), u commutes with H, namely [u, H] = 0, and consequently
H(H) = u Hu = H .
(4.40)
4.5 Symmetric Ordering

What is in general the correspondence between a classical variable and its
quantum operator counterpart? Consider for example the system of a point
particle moving in one dimension. Let x(c) be the classical coordinate and let
p(c) be the canonically conjugate momentum. As we have done in chapter 3,
the quantum observables corresponding to x(c) and p(c) are the Hermitian
operators x and p. The commutation
relation [x, p] is derived from the cor
responding Poissons brackets x(c) , p(c) according to the rule
{, }
Eyal Buks
1
[, ] ,
i
(4.41)
74
4.5. Symmetric Ordering

namely
1
2
x(c) , p(c) = 1 [x, p] = i .
(4.42)
However,
what
is the quantum operator corresponding to a general func
tion A x(c) , p(c) of x(c) and p(c) ? This question raises the issue of order

ing. As an example, let A x(c) , p(c) = x(c) p(c) . Classical variables obviously
commute, therefore x(c) p(c) = p(c) x(c) . However, this is not true for quantum
operators xp = px. Moreover, it is clear that both operators xp and px cannot
be considered as observables since they are not Hermitian
(4.43)
(4.44)
(xp) = px = xp ,
(px) = xp = px .
A better candidate to serve as the quantum operator corresponding to the

classical variables x(c) p(c) is the operator (xp + px) /2, which is obtained from
x(c) p(c) by a procedure called symmetric ordering. A general transformation
that produces a symmetric ordered observable
A (x, p) that corresponds to

a given general function A x(c) , p(c) of the classical variable x(c) and its
canonical conjugate p(c) is given below

A (x, p) =
A x(c) , p(c) dx(c) dp(c) ,

(4.45)
where
=
1
2
(2)

(c)
e ((x
i
x)+(p(c) p))
dd .
(4.46)
This transformation is called the Weyl transformation. The identity

dkeik(x x ) = 2 (x x ) ,
implies that

(c)
i
1
e (x x) d = x(c) x ,
2

(c)
i
1
e (p p) d = p(c) p .
2
(4.47)
(4.48)
(4.49)
At first glance these relations

may
leadto the (wrong) conclusion that the
term equals to x(c) x p(c) p , however, this is incorrect since x
and p are non-commuting operators.
Eyal Buks
75
4.6 Problems
1. Consider spin 1/2 in magnetic field given by B = B
z, where B is a
constant. At time t = 0 the system is in the state |+; x
. Calculate Sx ,
Sy and Sz as a function of time t.
2. Consider a point particle having mass m moving in one dimension under
the influence of the potential V (x). Let |n be a normalized eigenvector
of the Hamiltonian of the system with eigenvalue En . Show that the corresponding wavefunction n (x ) in the coordinate representation satisfies
the following equation
2 d2 n (x )
+ V (x ) n (x ) = En n (x ) .
2m dx2
(4.50)
3. Consider the Hamiltonian operator

H=
p2
+ V (r) ,
2m
(4.51)
where r = (x, y, z) is the vector of position operators, p = (px , py , pz )

is the vector of canonical conjugate operators, and the mass m is a constant. Let |n be a normalizable eigenvector of the Hamiltonian H with
eigenvalue En . Show that
n | p |n = 0 .
(4.52)
4. Show that in the p representation the Schrdinger equation

i
d |
= H | ,
dt
(4.53)
where H is the Hamiltonian

H=
p2
+ V (r) ,
2m
can be transformed into the integro-differential equation

d
p2
i =
+ dp U (p p ) ,
dt
2m
(4.54)
(4.55)
where = (p , t) = p | is the momentum wave function and

where

i
U (p) = (2)3 dr V (r) exp p r .
(4.56)

5. Consider a particle of mass m in a scalar potential energy V (r). Prove
Ehrenfests theorem
m
Eyal Buks
d2
r = V (r) .
dt2
(4.57)
76
4.6. Problems
6. Show that if the potential energy V (r) can be written as a sum of functions of a single coordinate, V (r) = V1 (x1 ) + V2 (x2 ) + V3 (x3 ), then the
time-independent Schrdinger equation can be decomposed into a set of
one-dimensional equations of the form
d2 i (xi ) 2m
+ 2 [Ei Vi (xi )] i (xi ) = 0 ,
dx2i
(4.58)
where i {1, 2, 3}, with (r) = 1 (x1 ) 2 (x2 ) 3 (x3 ) and E = E1 +

E2 + E3 .
7. Show that, in one-dimensional problems, the energy spectrum of the
bound states is always non-degenerate.
8. Let n (x) (n = 1, 2, 3, ) be the eigen-wave-functions of a onedimensional Schrdinger equation with eigen-energies En placed in order
of increasing magnitude (E1 < E2 < . ). Show that between any two
consecutive zeros of n (x), n+1 (x) has at least one zero.
9. What conclusions can be drawn about the parity of the eigen-functions
of the one-dimensional Schrdinger equation
d2 (x) 2m
+ 2 (E V (x)) (x) = 0
dx2
(4.59)
if the potential energy is an even function of x , namely V (x) = V (x).

10. Show that the first derivative of the time-independent wavefunction is
continuous even at points where V (x) has a finite discontinuity.
11. A particle having mass m is confined by a one dimensional potential given
by
+
W if |x| a
Vs (x) =
,
(4.60)
0 if |x| > a
where a > 0 and W > 0 are real constants. Show that the particle has
at least one bound state (i.e., a state having energy E < 0).
12. Consider a particle having mass m confined in a potential well given by
+
0
if 0 x a
V (x) =
.
(4.61)
if x < 0 or x > a
The eigenenergies are denoted by En and the corresponding eigen states
are denoted by |n , where n = 1, 2, (as usual, the states are numbered in increasing order with respect to energy). The state of the system
at time t = 0 is given by
| (0) = a1 |1 + a2 |2 + a3 |3 .
(4.62)
(a) The energy E of the system is measured at time

t =0 . What is the
probability to measure a value smaller than 3 2 2 / ma2 ? (b) Calculate
Eyal Buks
77

.
the standard deviation E = E 2 E2 at time t = 0 . (c) the same
as (b), however for any time t > 0 . (d) The energy was measured at
time t and the value of 22 2 / ma2 was found. The energy is measured
again at later time t0 > t . Calculate E and E at time t0 .
13. Consider a point particle having mass m in a one dimensional potential
given by
V (x) = (x) ,
(4.63)
where (x) is the delta function. The value of the parameter suddenly
changes from 1 at times t < 0 to the value 2 at times t > 0. Both
1 and 2 are positive real numbers. Given that the particle was in the
ground state at times t < 0, what is the probability p that the particle
will remain bounded at t > 0?
given by
V (x) = (x) ,
(4.64)
where (x) is the delta function, and where > 0. Let | 0 be the
ground state and let E0 be the energy of the ground state. The particle
is prepared in the state

ip0 x
|g (p0 ) = exp
(4.65)
| 0 ,

where p0 is real and where x is the position operator. Calculate the
probability s (p0 ) that a measurement of energy will yield the result E0 .
given by
+
(x) |x| < a
V (x) =
,
(4.66)
|x| a
where (x) is the delta function and is a constant. Let E0 be the energy
of the ground state. Under what conditions E0 < 0?
16. The same as the previous exercise, however for the potential
+
x<0
V (x) =
,
(4.67)
(x x0 ) x 0
where is real and x0 is positive.
17. Thomas-Reiche-Kuhn sum rule - Let
H=
p2
+ V (r)
2m
(4.68)
be the Hamiltonian of a particle of mass m moving in a potential V (r).

Show that
Eyal Buks
78
4.6. Problems

k
(Ek El ) |k| x |l| =
2
,
2m
(4.69)
where the sum is taken over all energy eigen-states of the particle (where
H |k = Ek |k ), and x is the x component of the position vector operator
r (the Thomas-Reiche-Kuhn sum rule).
18. A particle having mass m is confined in a one dimensional potential well
given by
+
0 0<x<a
V (x) =
.
else
a) At time t = 0 the position was measured and the result was x = a/2.
The resolution of the position measurement is x , where x << a.
After time 1 the energy was measured. Calculate the probability pn
to measure that the energy of the system is En , where En are the
eigenenergies of the particle in the well, and where n = 1, 2, .
b) Assume that the result of the measurement in the previous section
was E2 . At a later time 2 > 1 the momentum p of the particle
was measured. Calculate the expectation value p.
19. A particle having mass m is in the ground state of an infinite potential
well of width a, which is given by
+
0 0<x<a
V1 (x) =
.
(4.70)
else
At time t = 0 the potential suddenly changes and becomes
+
0 0 < x < 2a
V2 (x) =
,
else
(4.71)
namely the width suddenly becomes 2a. (a) Find the probability p to
find the particle in the ground state of the new well. (b) Calculate the
expectation value of the energy H before and after the change in the
potential.
'
(
20. Calculate the uncertainties in position (x)2 and in momentum
'
(
(p)2 of the energy eigenstates of a particle having mass m, which
is confined in a one dimensional potential well given by
+
0 0<x<a
V (x) =
.
(4.72)
else
21. The continuity equation - Consider a point particle having mass m
and charge q placed in an electromagnetic field. Show that
d
+ J = 0 ,
dt
Eyal Buks
(4.73)
79

where
=
(4.74)
is the probability distribution function, (x ) is the wavefunction,

J=

q
Im ( )
A
m
mc
(4.75)
is the current density, and A is the electromagnetic vector potential.

22. A particle having mass m moves in one dimension under the influence of
the potential V (x ). In the range |x | > a the potential V (x ) vanishes.
Consider a solution to the time independent Schrdinger equation, whose
wavefunction (x ) in the range |x | > a is taken to be given by
+
A1 eikx + B1 eikx x < a

(x ) =
,
(4.76)
A2 eikx + B2 eikx x > a

where A1 , B1 , A2 , B2 and k are all constants. Find a relation that the
constants A1 , B1 , A2 and B2 must satisfy.
23. A particle having mass m moves in one dimension under the influence of
a rectangular potential barrier given by
+
Ub |x | a2
V (x ) =
.
(4.77)
0 |x | > a2
Consider a solution to the time independent Schrdinger equation, whose
wavefunction in the range |x| > a/2 has the form
+ ikx
e
+ reikx x < a2
(x ) =
,
(4.78)
teikx
x > a2
where k is a constant. Calculate the transmission and reflection coefficients t and r respectively.

24. Calculate the Weyl transformation A (x, p) of the classical variable A x(c) , p(c) =
p(c) x(c) .

25. Invert Eq. (4.45), i.e. express the variable A x(c) , p(c) as a function of
the operator A (x, p).
4.7 Solutions
1. The operators Sx , Sy and Sz are given by Eqs. (2.102), (2.103) and (2.99)
respectively. The Hamiltonian is given by Eq. (4.21). Using Eqs. (4.38)
and (2.131) one has
Eyal Buks
80
4.7. Solutions
d Sx
=
[Sx , Sz ] = Sy ,
dt
i
d Sy
=
[Sy , Sz ] = Sx ,
dt
i
d Sz
=
[Sz , Sz ] = 0 ,
dt
i
where
=
|e| B
.
me c
(4.79)
(4.80)
(4.81)
(4.82)
At time t = 0 the system is in state

1
|+; x
= (|+ + |) ,
2
(4.83)
thus

(+| + |) (|+ | + | +|) (|+ + |) = .
4
2

Sy (t = 0) = (+| + |) (i |+ | + i | +|) (|+ + |) = 0 .
4

Sz (t = 0) = (+| + |) (|+ +| | |) (|+ + |) = 0 .
4
The solution is easily found to be given by

Sx (t) =
cos (t) ,
(4.84)
2

Sy (t) =
sin (t) ,
(4.85)
2
Sz (t) = 0 .
(4.86)
Sx (t = 0) =
2. The Hamiltonian operator H is given by

H=
p2
+ V (x) .
2m
(4.87)
Multiplying the relation

H |n = En |n
(4.88)
from the left by x | yields [see Eqs. (3.23) and (3.29)]
2 d2 n (x )
+ V (x ) n (x ) = En n (x ) ,
2m dx2
(4.89)
where
n (x ) = x |n
(4.90)
is the wavefunction in the coordinate representation.

Eyal Buks
81

3. Using [x, px ] = [y, py ] = [z, pz ] = i one finds that
2
p
,r
[H, r] =
2m
1 , 2 - , 2 - , 2 -
=
px , x , py , y , pz , z
2m

(px , py , pz )
=
im

p.
=
im
(4.91)
Thus
im
n | [H, r] |n

im
=
n | (Hr rH) |n

imEn
=
n | (r r) |n

=0.
n | p |n =
(4.92)
4. Multiplying Eq. (4.53) from the left by the bra p | and inserting the
closure relation

1 = dp |p p |
(4.93)
yields
i
d (p )
=
dt
dp p | H |p (p ) .
(4.94)
The following hold

p | p2 |p = p2 (p p ) ,
and
dr p |r r | V (r) |r r |p

ip r
ip r
3
V (r ) (r r ) exp
= (2)
dr dr exp

i (p p ) r
= (2)3 dr exp
V (r )

= U (p p ) ,
(4.96)
p | V (r) |p =
Eyal Buks
(4.95)
dr
82
4.7. Solutions
thus the momentum wave function (p ) satisfies the following equation

d
p2
i
=
+ dp U (p p ) .
(4.97)
dt
2m
5. The Hamiltonian is given by
H=
p2
+ V (r) .
2m
(4.98)
Using Eq. (4.38) one has
and
-* px
d x
1
1 ),
=
[x, H] =
x, p2x =
,
dt
i
i2m
m
d px
1
=
[px , V (r)] ,
dt
i
or with the help of Eq. (3.76)

5
6
d px
V
=
.
dt
x
This together with Eq. (4.99) yield
5
6
d2 x
V
m
=
.
dt2
x
(4.99)
(4.100)
(4.101)
(4.102)
Similar equations are obtained for y and z, which together yield Eq.
(4.57).
6. Substituting a solution having the form
(r) = 1 (x1 ) 2 (x2 ) 3 (x3 )
(4.103)
into the time-independent Schrdinger equation, which is given by

2 (r) +
2m
[E V (r)] (r) = 0 ,
2
(4.104)
and dividing by (r) yield

3

i=1

1 d2 i (xi ) 2m
2m
2 Vi (xi ) = 2 E .
2
i (xi ) dxi

(4.105)
In the sum, the ith term (i {1, 2, 3}) depends only on xi , thus each
term must be a constant
1 d2 i (xi ) 2m
2m
2 Vi (xi ) = 2 Ei ,
2
i (xi ) dxi

(4.106)
where E1 + E2 + E3 = E.
Eyal Buks
83

7. Consider two eigen-wave-functions 1 (x) and 2 (x) having the same
eigenenergy E. The following holds
d2 1 2m
+ 2 (E V (x)) 1 = 0 ,
dx2

d2 2 2m
+ 2 (E V (x)) 2 = 0 ,
dx2
(4.107)
(4.108)
thus
1 d2 1
1 d2 2
=
,
1 dx2
2 dx2
(4.109)
or
d2 1
d2 2
d
=
2
1
2
2
dx
dx
dx

d1
d 2
2
1
=0,
dx
dx
(4.110)
therefore
2
d1
d
1 2 = C ,
dx
dx
(4.111)
where C is a constant. However, for bound states

lim (x) = 0 ,
(4.112)
thus C = 0, and consequently

1 d1
1 d2
=
.
1 dx
2 dx
(4.113)
Integrating the above equation yields

log 1 = log 2 + ,
(4.114)
where is a constant. Therefore

1 = e 2 ,
(4.115)
and therefore 2 is just proportional to 1 (both represent the same

physical state).
8. Consider two eigen-wave-functions n (x) and n+1 (x) with En < En+1 .
As we saw in the previous exercise, the spectrum is non-degenerate. Moreover, the Schrdinger equation
d2 2m
+ 2 (E V (x)) = 0 ,
dx2
(4.116)
which the eigen-wave-functions satisfy, is real. Therefore given that (x)

is a solution with a given eigenenergy E, then also (x) is a solution
Eyal Buks
84
4.7. Solutions
with the same E. Therefore, all eigen-wave-functions can be chosen to be
real (i.e., by the transformation (x) ( (x) + (x)) /2). We have
d2 n 2m
+ 2 (En V (x)) n = 0 ,
(4.117)
dx2

d2 n+1 2m
+ 2 (En+1 V (x)) n+1 = 0 .
(4.118)
dx2

By multiplying the first Eq. by n+1 , the second one by n , and subtracting one has
n+1
d2 n+1 2m
d2 n
+ 2 (En En+1 ) n n+1 = 0 ,

n
dx2
dx2
(4.119)
or
d
dx
d
d
n+1 n n n+1
dx
dx
2m
[En En+1 ] n n+1 = 0 . (4.120)
2
Let x1 and x2 be two consecutive zeros of n (x) (i.e., n (x1 ) =

n (x2 ) = 0). Integrating from x1 to x2 yields
x2
x2
d
2m
d
n+1
n+1 n n
=
(E
E
)
dx n n+1 .
n+1
n
dx

x1
dx
2
=0
x1
>0
(4.121)
Without lost of generality, assume that n (x) > 0 in the range (x1 , x2 ).
Since n (x) is expected to be continuous, the following must hold
dn
>0,
(4.122)
dx
x=x1
dn
<0.
(4.123)
dx
x=x2
As can be clearly seen from Eq. (4.121), the assumption that n+1 (x) > 0
in the entire range (x1 , x2 ) leads to contradiction. Similarly, the possibility that n+1 (x) < 0 in the entire range (x1 , x2 ) is excluded. Therefore,
n+1 must have at least one zero in this range.
9. Clearly if (x) is an eigen function with energy E, also (x) is an
eigen function with the same energy. Consider two cases: (i) The level E
is non-degenerate. In this case (x) = c (x), where c is a constant.
Normalization requires that |c|2 = 1. Moreover, since the wavefunctions
can be chosen to be real, the following holds: (x) = (x). (ii) The
level E is degenerate. In this case every superposition of (x) and (x)
can be written as a superposition of an odd eigen function odd (x) and
an even one even (x), which are defined by
Eyal Buks
85

odd (x) = (x) (x) ,
even (x) = (x) + (x) .
(4.124)
(4.125)
10. The time-independent Schrdinger equation reads

d2 (x) 2m
+ 2 (E V (x)) (x) = 0 .
dx2
(4.126)
Assume V (x) has a finite discontinuity at x = x0 . Integrating the

Schrdinger equation in the interval (x0 , x0 + ) yields

x0 +
x +
d (x)
0
2m
= 2
(V (x) E) (x) = 0 .
dx
x0
(4.127)
x0
In the limit 0 the right hand side vanishes (assuming (x) is

bounded). Therefore d (x) /dx is continuous at x = x0 .
11. Since Vs (x) = Vs (x) the ground state wavefunction is expected to be
an even function of x. Consider a solution having an energy E and a
wavefunction of the form
if x > a
Aex
(x) = B cos (kx) if a x a ,
(4.128)
Aex
if x < a
where
2mE
=
,
(4.129)
and

2m (W + E)
k=
.
(4.130)
Requiring that both (x) and d (x) /dx are continuous at x = a yields
Aea = B cos (ka) ,
(4.131)
Aea = kB sin (ka) ,
(4.132)
and
or in a matrix form

A
0
C
=
,
B
0
(4.133)
where
Eyal Buks
86
4.7. Solutions
C=
ea cos (ka)
ea k sin (ka)
(4.134)
A nontrivial solution exists iff Det (C) = 0, namely iff
= tan (ka) .
k
(4.135)
This condition can be rewritten using Eqs. (4.129) and (4.130) and the
dimensionless parameters
K = ka ,
2mW
K0 =
a,
(4.136)
(4.137)
as
cos2 K =
1
1
=
2 =
2
1 + tan K
1 + k
K
K0
2
(4.138)
Note, however, that according to Eq. (4.135) tan K > 0. Thus, Eq. (4.135)
is equivalent to the set of equations
K
,
K0
tan K > 0 .
|cos K| =
(4.139)
(4.140)
This set has at least one solution (this can be seen by plotting the functions |cos K| and K/K0 ).
12. Final answers: (a) |a1 |2 + |a2 |2 . (b)
7
! 3
"2
8 3
2 2 8

9
2 4
2 2
E =
|an | n
|an | n
.
(4.141)
2ma2 n=1
n=1

(c) The same as at t = 0. (d) E = 22 2 / ma2 , E = 0.
13. The Schrdinger equation for the wavefunction (x) is given by
2

d
2m
+ 2 E (x) = 0 .
(4.142)
dx2

The boundary conditions at x = 0 are

0+ = 0 ,
+
2
d (0 ) d (0 )
= (0) ,
dx
dx
a0
where
a0 =
Eyal Buks
2
.
m
(4.143)
(4.144)
(4.145)
87

Due to symmetry V (x) = V (x) the solutions are expected to have
definite symmetry (even (x) = (x) or odd (x) = (x)). For
the ground state, which is expected to have even symmetry, we consider
a wavefunction having the form
(x) = Ae|x| ,
where A is a normalization constants and where
2mE
.
=
(4.146)
(4.147)
The parameter is real for E < 0. This even wavefunction satisfies the
Schrdinger equation for x = 0 and the boundary condition (4.143). The
condition (4.144) leads to a single solution for the energy of the ground
state
E=
m2
.
22
(4.148)
Thus the normalized wavefunction of the ground state 0 (x) is given by

0
m
m
0, (x) =
exp
2 |x| .
(4.149)
2

The probability p that the particle will remain bounded is given by
p=
0,1 (x) 0,2 (x) dx
2

4m2 1 2

m (1 + 2 )
exp
x
dx
4
2

0
41 2
=
.
(1 + 2 )2
=
(4.150)
14. The normalized wavefunction of the ground state is given by [see Eq.
(4.149)]
0
m
m
0 (x) =
exp
2 |x| .
(4.151)
2

Thus, the probability s (p0 ) is given by
Eyal Buks
88
4.7. Solutions
2

ip0 x
| 0
s (p0 ) =
0 | exp

m
ip0 x
2m
=
2
exp
exp 2 |x| dx

=
1+
2 p20
4m2 2
2 .
(4.152)
15. The Schrdinger equation for the wavefunction (x) is given by

2

d
2m
+ 2 (E V ) (x) = 0 .
(4.153)
dx2

The boundary conditions imposed upon (x) by the potential are [see
Eq. (4.127)]
(a) = 0 ,

0+ = 0 ,
d (0+ ) d (0 )
2
= (0) ,
dx
dx
a0
where
a0 =
2
.
m
(4.154)
(4.155)
(4.156)
(4.157)
Due to symmetry V (x) = V (x) the solutions are expected to have

definite symmetry (even (x) = (x) or odd (x) = (x)). For
the ground state, which is expected to have even symmetry, we consider
a wavefunction having the form
+
A sinh ( (x a)) x > 0
(x) =
,
(4.158)
A sinh ( (x + a)) x < 0
where A is a normalization constants and where
2mE0
=
.
(4.159)

The parameter is real for E0 < 0. This even wavefunction satisfies Eq.
(4.153) for x = 0 and the boundary conditions (4.154) and (4.155). The
condition (4.156) reads
a0 = tanh (a) .
(4.160)
Nontrivial ( = 0) real solution exists only when a > a0 , thus E0 < 0 iff
a > a0 =
Eyal Buks
2
.
m
(4.161)
89

16. For the present case the boundary conditions imposed upon (x) by the
potential are [see Eq. (4.127)]
(0) = 0 ,

+
,
x0 = x
0
+

d x0
d x0
2
= (x0 ) ,
dx
dx
a0
where
a0 =
2
.
m
(4.162)
(4.163)
(4.164)
(4.165)
Consider a solution having the form

+
A sinh (x) + B cosh (x) 0 x < x0
(x) =
,
e(xx0 )
x > x0
(4.166)
where
=
2mE0
.
(4.167)
The boundary conditions yield

B=0,
1 = A sinh (x0 ) ,
2
(1 + A cosh (x0 )) = ,
a0
thus
x0 (1 + coth (x0 )) =
2x0
.
a0
(4.168)
(4.169)
(4.170)
(4.171)
Note that for x 0 the following holds x (1 + coth x) 1, thus a solution

with E0 < 0 (i.e. a solution with a real positive ) is possible only if
2x0
1,
a0
(4.172)
or
x0
2
.
2m
(4.173)
17. Using Eq. (4.37) one has

dx(H)
1 3 (H) 4
=
x ,H ,
dt
i
(4.174)
therefore
Eyal Buks
90
4.7. Solutions
k|
dx(H)
1
i (Ek El )
|l =
k| x(H) H Hx(H) |l =
k| x(H) |l .
dt
i

(4.175)
Integrating yields
(H)
k| x
(H)
(t) |l = k| x
(t = 0) |l exp
i (Ek El ) t
(4.176)
Using this result one has

(Ek El ) |k| x |l|2
k

k

k
(Ek El )
k| x(H) |l
(Ek El ) k| x(H) |l l| x(H) |k

dx(H)
dx(H)

(H)
(H)
=
k|
|l l| x |k k| x |l l|
|k
2i
dt
dt
k

dx(H)
dx(H)
(H)
(H)
=
|l l|
|k k| x |l
l| x |k k|
2i k
dt
dt
=

dx(H) dx(H) (H)
l| x(H)
x |l .
2i
dt
dt
(4.177)
Using again Eq. (4.37) one has
dx(H)
1 3 (H) 4 px
=
x ,H =
,
dt
i
m
(H)
(4.178)
therefore
3
4

(Ek El ) |k| x |l|2 =
l| x(H) , p(H)
|l
x
2im
k

i
2im
2
=
.
2m
=
(4.179)
18. The wavefunctions of the normalized eigenstates are given by
0
2
nx
n (x) =
sin
,
a
a
(4.180)
and the corresponding eigenenergies are

Eyal Buks
91

En =
2 2 n2
.
2ma2
(4.181)
a) The wavefunction after the measurement is a normalized wavepacket

centered at x = a/2 and having a width x
x
+ 1
2
2
x
(x) =
.
(4.182)
0
else
Thus in the limit x << a
a
x 2 n
pn =
dxn (x) (x)
2
sin
.
a
2
0
(4.183)
Namely, pn = 0 for all even n, and the probability of all energies with
odd n is equal.
b) Generally, for every bound state in one dimension p = 0 [see Eq.
(4.52)].
19. For a well of width a the wavefunctions of the normalized eigenstates are
given by
0
2
nx
(a)
n (x) =
sin
,
(4.184)
a
a
En(a) =
2 2 n2
.
2ma2
(a) The probability is given by
a
32
(a)
(2a)
p=
dx1 (x) 1 (x)
= 2 .
9
0
(4.185)
(4.186)
(a)
(b) For times t < 0 it is given that H = E1 . Immediately after the
change (t = 0+ ) the wavefunction remains unchanged. A direct evaluation
(a)
of H using the new Hamiltonian yields the same result H = E1 as for
t < 0. At later times t > 0 the expectation value H remains unchanged
due to energy conservation.
20. The wavefunctions of the normalized eigenstates are given by [see Eq.
(4.181)]
0
2
nx
n (x) =
sin
,
(4.187)
a
a
and the corresponding eigenenergies are [see Eq. (4.180)]
En =
Eyal Buks
2 2 n2
,
2ma2
(4.188)
92
4.7. Solutions
where n = 1, 2, . By symmetry for all states x = a/2. Furthermore,
for all states p = 0 [see Eq. (4.52)]. For the nth state the following
holds

) 2 * 2 a 2 2 nx
a2 2n2 2 3
=
x =
dx x sin
,
(4.189)
a 0
a
6n2 2
thus

'
( ) *
1
1
2
2
(x) = x2 x = a2
2 2
12 2n
and [see Eq. (3.29)]

2
'
( ) * 2 n 2 a
nx
n
(p)2 = p2 =
dx sin2
=
, (4.190)
a
a
a
a
0
thus [compare with the uncertainty principle (3.10)]

'
('
( 2 n2 2
(x)2 (p)2 =
2 .
4
3
(4.191)
21. The Schrdinger equation is given by

i
d |
= H | ,
dt
(4.192)
where the Hamiltonian is given by [see Eq. (1.62)]

2
p qc A
H=
+ q .
2m
(4.193)
Multiplying from the left by x | yields

i
where
d
1
q 2
=
i A + q ,
dt
2m
c
= (x ) = x | .
(4.194)
(4.195)
Multiplying Eq. (4.194) by , and subtracting the complex conjugate of

Eq. (4.194) multiplied by yields

d
1
q 2
q 2
=
i A i A ,
(4.196)
i
dt
2m
c
c
where
=
Eyal Buks
(4.197)
93

is the probability distribution function. Moreover, the following holds

q 2
q 2
i A i A
c
c

q 2
iq
iq
2 2
2
A+
A
= +
A +
c
c
c

q 2
iq
2 2
2 iq
+
A
A
A
c
c
c

= 2 2 2
iq
+
( A + A + A + A )
c
iq
= 2 ( ) +
( A + A ) .
c
(4.198)
Thus, Eq. (4.196) can be written as
d
+ J = 0 ,
dt
(4.199)
where
J=

q
Im ( )
A.
m
mc
(4.200)
22. The current density J [see Eq. (4.75)] that is associated with the wave
function (x ) = Aeikx + Beikx is given by

J=
Im
m
x

=
Im ik A eikx + B eikx
Aeikx Beikx
m

=
Im ik |A|2 |B|2 + AB e2ikx A Be2ikx
m

k 2
=
|A| |B|2 .
m
(4.201)
Thus for a solution to the time independent Schrdinger equation, for
which the current density = is time independent, the continuity
equation (4.73) yields the relation
|A1 |2 |B1 |2 = |A2 |2 |B2 |2 .
(4.202)
23. In this problem the potential is piecewise constant. At the points where
the piecewise constant potential abruptly changes the solution has to
satisfy the requirements that both (x) and d/dx [see Eq. (4.127)] are
Eyal Buks
94
4.7. Solutions
continuous. Consider first a general case, where a given potential is taken
to be given by
+
Ul x x0
V (x ) =
,
(4.203)
Ur x > x0
where Ul and Ur are constants, and the wavefunction is expressed as
+
Al eikl x + Bl eikl x x x0
,
(4.204)
(x ) =
Ar eikr x + Br eikr x x > x0

where Al , Bl , Ar and Br are constants, and where the constants kl and
kr is related to the energy of the particle E by [see Eq. (4.50)]
2 kl2
= E Ul ,
(4.205)
2m
2 kr2
= E Ur .
(4.206)
2m
The requirements that both (x) and d/dx [see Eq. (4.127)] are continuous yield a linear relation between the amplitudes on the left Al and
Bl and those on the right Ar and Br

Al
Ar
=M
,
(4.207)
Bl
Br
where
1 ik x

eikl x0
eikl x0
e r 0
eikr x0
kl eikl x0 kl eikl x0
kr eikr x0 kr eikr x0
!
"
1 + kkrl 1 kkrl
1
= (kl x0 )
(kr x0 ) ,
1 kkrl 1 + kkrl
2
M=
(4.208)
and where the matrix () is defined by

i
e
0
() =
.
0 ei
(4.209)
The above general result (4.207) is employed below for the given piecewise
constant potential (4.77). Assume first that the wave function on the right
side of the barrier in the region x > a/2 is given by (x ) = eikx , and
on the left side of the barrier in the region x < a/2 it is given by
(x ) = Aeikx + Beikx . With the help of Eq. (4.207) one finds that
Eyal Buks
95

A
B

1
ka
1 + k 1 k

(a)
1 k 1 + k
4
2

ka
1 + k 1 k
1
0
1 k 1 + k
2
ika

i k
cos a
2 + k sin a
e

,
=
1
k
2 i k sin a
where
2 k2
=E,
2m
2 2

= E Ub ,
2m
and thus
(4.210)
(4.211)
(4.212)
eika
1

=
,
A
cos a 2i k + k sin a
k ika
1
sin a
B
2 i k e

.
r=
=
A
cos a 2i k + k sin a
t=
(4.213)
(4.214)
Note that, as is expected from current conservation [see Eq. (4.202)], the
following holds |t|2 + |r|2 = 1.
A (x, p) =
1
2
(2)

p(c) x(c) e [(x

i
(c)
x)+(p(c) p)]
dddx(c) dp(c) .
(4.215)

eA eB = eA+B e(1/2)[A,B] ,
(4.216)
one has
i
e x e p = e (x+p) e 2 2 [x,p] ,
(4.217)
thus

(c)
(c)
i
1
i
i
i
A (x, p) =
p(c) x(c) e (x +p ) e 2 e x e p dddx(c) dp(c)
2
(2)

(c)
(c)
i
1
i
i
=
p(c) x(c) e [((x + 2 )+p )] e x e p dddx(c) dp(c) .
2
(2)
(4.218)
Changing the integration variable
Eyal Buks
96
4.7. Solutions
x(c) = x(c)
one has
,
2
(4.219)

i (x(c) +p(c) ) i x i p
p(c) x(c)
e
e e dddx(c) dp(c)
2
(2)

1
i (x(c) x) i (p(c) p)
(c)
(c)
=
p
x
e
e
dddx(c) dp(c) .
2
(2)2
(4.220)
A (x, p) =
Using the identity

dkeik(x x ) = 2 (x x ) ,
one finds that

(c)
i
1
e (x x) d = x(c) x ,
2

(c)
i
1
e (p p) d = p(c) p ,
2
(4.221)
(4.222)
(4.223)
thus

(c)
i
1
i (p(c) p)
(c)
(c)
(c)
(c) 1
A (x, p) =
p
x
e
ddx dp
e (x x) d
2
2
2

i (p(c) p)
1
(c)
(c)
(c)
(c)
=
x
ddx dp x(c) x
p
e
2
2

i (c)
1
=
p(c) x
e (p p) ddp(c)
2
2

(c)
i
1
1
i (c)
= p(c) xdp(c)
e (p p) d
p(c) e (p p) ddp(c)
2
2
2

(c)
i
1
= px
p(c) e (p p) ddp(c)
2
2
(c)
i

p)
(p
1
(c) e
= px
p
ddp(c)
2 2i
p(c)

(c)
i
1
= px
dp(c) p(c) (c)
de (p p) .
2i
p 2

(p(c) p)
(4.224)
Integration by parts yields
Eyal Buks
97

(c)

p
p(c) p dp(c)
(c)
2i
p

= px
2i
[x, p]
= px +
2
xp + px
=
.
2
A (x, p) = px
(4.225)
(c) (c)
25. Below we derive an expression for the variable A x , p
in terms
of the matrix elements of the operator A (x, p) in the basis of position
eigenvectors
|x . To
thatend
( we begin by evaluating the matrix element
'

x
x
x 2
A (x, p)
x + 2 using Eqs. (4.218), (3.19) and (4.222)
5
6
x
x
x
A (x, p)
x +
2
2

i
(c)
(c)
1
(c) (c)
A
x
=
,
p
e [((x + 2 )+p )]
2
(2)
5
6
x
i x i p
x
e
x
+
e
dddx(c) dp(c)
2

i
i
x
(c)
1
+ 2 )+p(c) )] x 2
(c) (c)
[( (x
=
A
x
,
p
e
e
(2)2
5
6
x
x
+
x
x
+
dddx(c) dp(c)
2
2

i (c)
(c)
i
1
(c) (c)
x p
(c)
(c)
A
x
e [(x x )] d
=
,
p
e
dx
dp
2
(2)

i (c)

1
=
A x(c) , p(c) e x p dx(c) dp(c) x(c) x
2

i (c)

1
=
A x , p(c) e x p dp(c) .
2
i
Applying the inverse Fourier transform, i.e. multiplying by e x

integrating over x yields
6
5
x
x
x
A (x, p)
x +
e x p dx
2
2

(c)
i
1
=
A x , p(c) dp(c) e x (p p ) dx ,
2
and
(4.226)
thus with the help of Eq. (4.223) one finds the desired inversion of Eq.
(4.45) is given by
Eyal Buks
98
4.7. Solutions
6
5
x
x
x +
A (x , p ) =
x
A
(x,
p)
e x p dx .
2
2
(4.227)
Note that A (x , p ), which appears on the left hand side of the above
equation (4.227) is a classical variable, whereas A (x, p) on the right hand
side is the corresponding quantum operator. A useful relations can be
obtained by integrating A (x , p ) over p . With the help of Eq. (4.222)
one finds that
6
5

x
x
x +
A
(x,
p)
e x p dp
A (x , p ) dp = dx x
2
= 2 x | A (x, p) |x .
(4.228)
Another useful relations can be obtained by integrating A (x , p ) over
x .With the help of Eqs. (3.52) and (4.223) one finds that
6

5
x
i
x
A (x , p ) dx =
x
A (x, p)
x +
e x p dx dx
2
2
6
5
x
x
i

=
x
|p p | A (x, p) |p p
x +
e x p dx dx dp dp
2
2

i
i x
1
i
=
e x (p p ) e 2 (p p ) p | A (x, p) |p e x p dx dx dp dp
2

i x
i
=
(p p ) e 2 (p p ) p | A (x, p) |p e x p dx dp dp

i
=
p | A (x, p) |p e x (p p ) dx dp

= 2 p | A (x, p) |p (p p ) dp
= 2 p | A (x, p) |p .
Eyal Buks
(4.229)
99
5. The Harmonic Oscillator
Consider a particle of mass m in a parabolic potential well

1
U (x) = m2 x2 ,
2
where the angular frequency is a constant. The classical equation of motion
for the coordinate x is given by [see Eq. (1.19)]
m
x=
U
= m 2 x .
x
(5.1)
It is convenient to introduce the complex variable , which is given by

1
i
=
x + x ,
(5.2)
x0
where x0 is a constant having dimension of length. Using Eq. (5.1) one finds
that

1
i
1
i 2
=
(5.3)
=
x + x
x x = i .
x0
x0
The solution is given by

= 0 eit ,
where 0 = (t = 0). Thus, x and x oscillate in time according to

x = x0 Re 0 eit ,

x = x0 Im 0 eit .
(5.4)
(5.5)
(5.6)
The Hamiltonian is given by [see Eq. (1.34)]

H=
p2
m 2 x2
+
.
2m
2
(5.7)
In quantum mechanics the variables x and p are regarded as operators satisfying the following commutation relations [see Eq. (3.9)]
[x, p] = xp px = i .
(5.8)
Chapter 5. The Harmonic Oscillator
5.1 Eigenstates
The annihilation and creation operators are defined as
0

m
ip
x+
,
a=
2
m
0

m
ip
a =
x
.
2
m
The inverse transformation is given by
0

x=
a + a ,
2m
0

m
p=i
a + a .
2
The following holds
,
i
([p, x] [x, p]) = 1 ,
a, a =
2
(5.9)
(5.10)
(5.11)
(5.12)
(5.13)
The number operator, which is defined as

N = a a,
(5.14)
can be expressed in terms of the Hamiltonian

N = a a

m
ip
ip
=
x
x+
2
m
m
2

m
i [x, p]
p
2
=
+x +
2 m2 2
m
2

2 2
p
m x
1
1
+
=
2m
2
2
H
1
=
.
2
(5.15)
Thus, the Hamiltonian can be written as

1
H = N +
.
2
(5.16)
The operator N is Hermitian, i.e. N = N , therefore its eigenvalues are

expected to be real. Let {|n} be the set of eigenvectors of N and let {n} be
the corresponding set of eigenvalues
Eyal Buks
102
5.1. Eigenstates
N |n = n |n .
(5.17)
According to Eq. (5.16) the eigenvectors of N are also eigenvectors of H

H |n = En |n ,
where the eigenenergies En are given by

1
.
En = n +
2
(5.18)
(5.19)
Theorem 5.1.1. Let |n be a normalized eigenvector of the operator N with

eigenvalue n. Then (i) the vector
|n + 1 = (n + 1)1/2 a |n
(5.20)
is a normalized eigenvector of the operator N with eigenvalue n + 1; (ii) the

vector
|n 1 = n1/2 a |n
(5.21)
is a normalized eigenvector of the operator N with eigenvalue n 1

Proof. Using the commutation relations
,
,
N, a = a a, a = a ,
,
[N, a] = a , a a = a ,
(5.22)
(5.23)
one finds that
N a |n =
and
,

N, a + a N |n = (n + 1) a |n ,
N a |n = ([N, a] + aN ) |n = (n 1) a |n .
(5.24)
(5.25)
Thus, the vector a |n, which is proportional to |n + 1, is an eigenvector of

the operator N with eigenvalue n + 1 and the vector a |n, which is proportional to |n 1, is an eigenvector of the operator N with eigenvalue n 1.
Normalization is verified as follows
,
n + 1 |n + 1 = (n + 1)1 n| aa |n = (n + 1)1 n| a, a + a a |n = 1 ,
(5.26)
and
n 1 |n 1 = n1 n| a a |n = 1 .
Eyal Buks
(5.27)
103

As we have seen from the above theorem the following hold
a |n = n |n 1 ,
a |n = n + 1 |n + 1 .
(5.28)
(5.29)
Claim. The spectrum (i.e. the set of eigenvalues) of N are the nonnegative
integers {0, 1, 2, }.
Proof. First, note that since the operator N is positive-definite the eigenvalues are necessarily non negative
n = n| a a |n 0 .
(5.30)
On the other hand, according to Eq. (5.28), if n is an eigenvalue also n 1

is an eigenvalue, unless n = 0. For the later case according to Eq. (5.28)
a |0 = 0. Therefore, n must be an integer, since otherwise one reaches a
contradiction with the requirement that n 0.
According to exercise 6 of set 4, in one-dimensional problems the energy
spectrum of the bound states is always non-degenerate. Therefore, one concludes that all eigenvalues of N are non-degenerate. Therefore, the closure
relation can be written as
1=
n=0
|n n| .
(5.31)
Furthermore, using Eq. (5.29) one can express the state |n in terms of the
ground state |0 as
n
a
(5.32)
|n = |0 .
n!
5.2 Coherent States

As can be easily seen from Eqs. (5.11), (5.12), (5.28) and (5.29), all energy
eigenstates |n have vanishing position and momentum expectation values
n| x |n = 0 ,
n| p |n = 0 .
(5.33)
(5.34)
Clearly these states dont oscillate in phase space as classical harmonic oscillators do. Can one find quantum states having dynamics that resembles
classical harmonic oscillators?
Eyal Buks
104
5.2. Coherent States

Definition 5.2.1. Consider a harmonic oscillator having ground state |0.
A coherent state | with a complex parameter is defined by
| = D () |0 ,
(5.35)
where

D () = exp a a ,
(5.36)
is the displacement operator.
In the set of problems at the end of this chapter the following results are
obtained:
The displacement operator is unitary D () D () = D () D () = 1.
The coherent state | is an eigenvector of the operator a with an eigenvalue
, namely
a | = | .
(5.37)

For any function f a, a having a power series expansion the following
holds

D () f a, a D () = f a + , a + .
(5.38)
The displacement operator satisfies the following relations
||2
||2
D () = e 2 ea e a = e 2 e a ea ,
m
2
2
+
i
x m
p 4
2
2
D () = e
e
e
,

D ( + ) .
D () D ( ) = exp
2
(5.39)
(5.40)
(5.41)
Coherent state expansion in the basis of number states
| = e
||2
2

n
|n .
n!
n=0
(5.42)
The following expectation values hold
Eyal Buks
105

2
H = | H | = || + 1/2 ,

| H2 | = 2 2 ||4 + 2 ||2 + 1/4 ,
.
2
H = | (H) | = || ,
0
2
Re () ,
x = | x | =
m
p = | p | = 2m Im () ,
0
.

2
x = | (x) | =
,
2m
0
.
m
2
,
p = | (p) | =
2

x p = .
2
The wave function of a coherent state is given by
(5.43)
(5.44)
(5.45)
(5.46)
(5.47)
(5.48)
(5.49)
(5.50)
(x ) = x |

2

2
x x
2 m 1/4
x
= exp
exp
+ i p
.
4

2x
(5.51)
The following closure relation holds

1
1=
| | d2 ,
(5.52)
where d2 denotes infinitesimal area in the complex plane, namely d2 =

d {Re } d {Im }.
Given that at time t = 0 the oscillator is in a coherent state with parameter 0 , namely | (t = 0) = |0 , the time evolution can be found with
the help of Eqs. (4.14), (5.19) and (5.42)

| |2
iEn t n0
20
|n
| (t) = e
exp

n!
n=0
n
exp (int) 0 |n
n!
n=0

n
|0 |2
0 eit
= eit/2 e 2
|n
n!
n=0
*
= eit/2
= 0 eit .
= eit/2 e
Eyal Buks
|0 |2
2
(5.53)
106
5.3. Problems
In view of Eqs. (5.43), (5.45) (5.48) and (5.49), we see from this results that
H , H , x and p are all time independent. On the other hand, as
can be seen from Eqs. (5.46) and (5.47) the following holds
0

2
Re 0 eit ,
x = | x | =
(5.54)
m

p = | p | = 2m Im 0 eit .
(5.55)
These results show that indeed, x and p have oscillatory time dependence identical to the dynamics of the position and momentum of a classical
harmonic oscillator [compare with Eqs. (5.5) and (5.6)].
5.3 Problems
1. Calculate the wave functions n (x ) = x |n of the number states |n
of a harmonic oscillator.
2. Show that

tn
exp 2Xt t2 =
Hn (X)
,
n!
n=0
(5.56)
where Hn (X) is the Hermite polynomial of order n, which is defined by

2

n

X
X2
d
Hn (X) = exp
exp
X
.
(5.57)
2
dX
2
3. Show that
n=0
n
2
Hn (X) Hn (Y )
=
n!
exp
(2XY X 2 Y 2 )
12
1 2
(5.58)
where Hn (X) is the Hermite polynomial of order n.

4. Show that for the state |n of a harmonic oscillator
2
'
('
(
1
2
2
(x)
(p) = n +
2 .
2
(5.59)
5. Consider a free particle in one dimension having mass m. Express the

Heisenberg operator x(H) (t) in terms x(H) (0) and p(H)
' (0). At
( time t = 0
2
the system in in the state |0 . Express the variance (x) (t) at time
t, where x = x x, in terms of the following expectation values at
time t = 0
Eyal Buks
107

x0 = 0 | x |0 ,
p0 = 0 | p |0 ,
(xp)0 = 0 | xp |0 ,
(x)20 = 0 | (x x0 )2 |0 ,
(p)20
= 0 | (p p0 ) |0 .
(5.60)
(5.61)
(5.62)
(5.63)
(5.64)
6. Consider a harmonic oscillator of angular frequency and mass m.

a) Express the Heisenberg picture x(H) (t) and p(H) (t) in terms x(H) (0)
and p(H) (0).
,
- ,
b) Calculate the following commutators p(H) (t1 ) , x(H) (t2 ) , p(H) (t1 ) , p(H) (t2 )
,
and x(H) (t1 ) , x(H) (t2 ) .
7. Consider a particle having mass m confined by a one dimensional potential V (x), which is given by
+ m2
2
2 x x>0 ,
(5.65)
V (x) =
x0
where is a constant.
a) Calculate the eigenenergies of the) system.
*
b) Calculate the expectation values x2 of all energy eigenstates of the
particle.
8. Calculate the possible energy values of a particle in the potential given
by
V (x) =
m2 2
x + x .
2
(5.66)
9. A particle is in the ground state of harmonic oscillator with potential

energy
V (x) =
m2 2
x .
2
(5.67)
Find the probability p to find the particle in the classically forbidden

region.
10. Consider a particle having mass m in a potential V given by
+ m2 z2 a
2 x a2 and a2 y a2
2
V (x, y, z) =
,
(5.68)
else
where and a are positive real constants. Find the eigenenergies of the
system.
11. Consider a harmonic oscillator having angular resonance frequency 0 .
At time t = 0 the systems state is given by
1
| (t = 0) = (|0 + |1) ,
2
Eyal Buks
(5.69)
108
5.3. Problems
where the states |0 and |1 are the ground and first excited states, respectively, of the oscillator. Calculate as a function of time t the following
quantities:
a) x
b) )p *
c) x2
d) xp
12. Harmonic oscillator having angular resonance frequency is in state
1
| (t = 0) = (|0 + |n)
2
(5.70)
at time t = 0, where |0 is the ground state and |n is the eigenstate
with eigenenergy (n + 1/2) (n is a non zero integer). Calculate the
expectation value x for time t 0.
13. Consider a harmonic oscillator having mass m and angular resonance
frequency . At time t = 0 the systems state is given by |(0) = c0 |0+
c1 |1 , where |n are the eigenstates with energies En = (n +.1/2).
Given that H = , |(0) is normalized, and x (t = 0) =

calculate x (t) at times t > 0.
14. Show that
D () = e
||2
2
ea e
=e
||2
2
e a ea .
1
2

m ,
(5.71)
15. Show that the displacement operator D () is unitary.

16. Show that
| = e
||2
2

n
|n .
n!
n=0
(5.72)
17. Show that the coherent state | is an eigenvector of the operator a with
an eigenvalue , namely
a | = | .
18. Show that
(5.73)
0

m
x
D () = exp

2

2

+
2
i
p exp
exp
.
4
2
m
(5.74)

19. Show that for any function f a, a having a power series expansion the
following holds

D () f a, a D () = f a + , a + .
(5.75)
Eyal Buks
109

20. Show that the following holds for a coherent state |:

2
a) | H | = || + 1/2 .

4
2
b) | H2 | = 2 2 || + 2 || + 1/4 .
.
c) | (H)2 | = ||.
.
2
d) x = | x | = m
Re ().
e) p = |
Im ().
. p | = 2m.
2

f) x = | (x) | = 2m
.
.
.
2
g) p = | (p) | = m
2 .
21. Consider a harmonic oscillator of mass m and angular resonance frequency . The Hamiltonian is given by
H=
1
p2
+ m 2 x2 .
2m 2
(5.76)
The system at time t is in a normalized state |, which is an eigenvector

of the annihilation operator a, thus
a | = | ,
(5.77)
where the eigenvalue is a complex number. At time t > 0 the energy of

the system is measured. What are the possible results En and what are
the corresponding probabilities pn (t)?
22. Show that the wave function of a coherent state is given by
(x ) = x |

2

2
x x
x
2 m 1/4
exp
+ i p
= exp
.
4

2x
(5.78)
23. Show that

D () D ( ) = exp
D ( + ) .
24. Show that the following closure relation holds

1
| | d2 ,
1=
(5.79)
(5.80)
where d2 denotes infinitesimal area in the complex plane, namely

d2 = d {Re } d {Im }.
25. Calculate the inner product between two coherent states | and |,
where , C.
Eyal Buks
110
5.3. Problems
26. A one dimensional potential acting on a particle having mass m is given
by
1
V1 (x) = m 2 x2 + m 2 x .
2
(5.81)
a) Calculate the Heisenberg representation of the position operator

x(H) (t) and its canonically conjugate operator p(H) (t).
b) Given that the particle at time t = 0 is in the state |0, where the
state |0 is the ground state of the potential
1
V1 (x) = m2 x2 .
2
(5.82)
Calculate the expectation value x at later times t > 0.

27. A particle having mass m is in the ground state of the one-dimensional
potential well V1 (x) = (1/2) m 2 (x x )2 for times t < 0 . At time
t = 0 the potential suddenly changes and becomes V2 (x) = (1/2) m 2 x2 .
a) Calculate the expectation
' value
( x at times t > 0.
2
b) Calculate the variance (x) at times t > 0 , where x = x x.
c) The energy of the particle is measured at time t > 0 . What are the
possible results and what are the probabilities to obtain any of these
results.
28. Consider a particle having mass m in the ground state of the potential
well Va (x) = (1/2) m2 x2 for times t < 0 . At time t = 0 the potential
suddenly changes and becomes Vb (x) = gx . (a) Calculate the
' expecta(
tion value x at times t > 0 . (b) Calculate the variance (x)2 at
times t > 0 , where x = x x.
29. Consider a particle of mass m in a potential of a harmonic oscillator
having angular frequency . The operator S (r) is defined as
4
3 r 2
S (r) = exp
,
(5.83)
a a2
2
where r is a real number, and a and a are the annihilation and creation
operators respectively. The operator T is defined as
T = S (r) aS (r) .
(5.84)
a) Find an expression for the operator T of the form T = Aa + Ba ,

where both A and B are constants.
b) The vector state |r is defined as
|r = S (r) |0 ,
(5.85)
where |0 is the ground state of the harmonic oscillator. Calculate

the expectation values r| x |r of the operator x (displacement) and
the expectation value r| p |r of the operator p (momentum).
Eyal Buks
111

c) Calculate the variance (x)2 of x and the variance (p)2 of p.
30. The state |r from the previous exercise, which is called a squeezed state,
can be alternatively defined as a normalized state that satisfies the relation
Q (r) |r = 0 ,
(5.86)
where the operator Q (r) is defined by

Q (r) = a cosh r + a sinh r ,
(5.87)
r is a real number, and a and a are the annihilation and creation operators respectively. Based on the above definition calculate the expectation
values r| x |r of the position operator x, the expectation value r| p |r
2
of the momentum operator p, the variance (x) of x and the variance
(p)2 of p with respect to the state |r.
31. Consider one dimensional motion of a particle having mass m. The Hamiltonian is given by
H = 0 a a + 1 a a aa ,
where
a=
m 0
2

ip
x+
,
m0
(5.88)
(5.89)
is the annihilation operator, x is the coordinate and p is its canonical

conjugate momentum. The frequencies 0 and 1 are both positive.
a) Calculate the eigenenergies of the system.
b) Let |0 be the ground state of the system. Calculate
i. 0|x|0
ii. 0|p|0
(
'
iii. 0| (x)2 |0
'
(
iv. 0| (p)2 |0
32. The Hamiltonian of a system is given by

H = N ,
(5.90)
where the real non-negative parameter has units of energy, and where
the operator N is given by
N = b b .
(5.91)
The following holds

b b + bb = 1 ,
b2 = 0 ,
2
b
=0.
Eyal Buks
(5.92)
(5.93)
(5.94)
112
5.3. Problems
a) Find the eigenvalues of H. Clue: show first that N 2 = N .
b) Let |0 be the ground state of the system, which is assumed to be
non-degenerate.
Define
the two states
|+ = A+ 1 + b |0 ,
(5.95a)

(5.95b)
| = A 1 b |0 ,
where the real non-negative numbers A+ and A are normalization
constants. Calculate A+ and A . Clue: show first that b |0 is an
eigenvector of N .
c) At time t = 0 the system is in the state
| (t = 0) = |+ ,
(5.96)
Calculate the probability p (t) to find the system in the state | at

time t > 0.

33. Normal ordering - Let f a, a be a function of
the annihilation a and
creation
a
operators.
The
normal
ordering
of
f
a, a , which is denoted

by : f a, a : places the a operators on the right and the a operators
on the left. Some examples are given below
: aa : = a a ,
: a a : = a a ,
n
n
: a a : = a an .
(5.97)
(5.98)
(5.99)
Normal ordering is linear, i.e. : f + g :=: f : + : g :. Show that the projection operator Pn = |n n|, where |n is an eigenvector of the Hamiltonian
of a harmonic oscillator, can be expressed as
Pn =

1 n
exp a a an : .
: a
n!
(5.100)
34. Consider a harmonic oscillator of angular frequency and mass m. A

time dependent force is applied f (t). The function f (t) is assumed to
vanish f (t) 0 in the limit t . Given that the oscillator was
initially in its ground state |0 at t calculate the probability pn
to find the oscillator in the number state |n in the limit t .
35. The parity operator P is defined by
P=
dx |x x | ,
(5.101)
where |x is an eigenvector of the position operator x having eigenvalue

x , i.e. x |x = x |x . Express the parity operator P as a function of the
number operator N = a a.
36. Show that
,

ea a =: exp e 1 a a : .
(5.102)
Eyal Buks
113

37. Consider a harmonic oscillator having mass m and angular resonance
frequency . Show that
2

2
x
exp 2x
2 xx0 a a2
2 +
0
|x =
|0 ,
(5.103)
1/2
1/4 x0
where |x is an eigenvector of the position operator x with eigenvalue x ,
i.e. x |x = x |x ,
0

m
ip
a=
x+
(5.104)
2
m
is the annihilation operator, |0 is the ground state and
0

.
x0 =
m
(5.105)
38. Show that
1
exp
1+
x2
1 + x20
2
x
=: exp 2 : ,
x0
where x is the position operator, is real and

0

x0 =
.
m
(5.106)
(5.107)
39. Let F (X) be a smooth function of the normalized position operator X

X=
x
a + a
=
,
x
2
0
(5.108)
where a is the annihilation operator, x is the position operator and

0

x0 =
.
(5.109)
m
Show that
d
dF
: F (X) : =:
: .
dX
dX
40. Consider a system having Hamiltonian H given by
k
H = a a + 1 a a ,
(5.110)
(5.111)
where a and a are the annihilation and creation operators, both and
1 are positive, and where k is integer. At initial time t = 0 the state
of the system is an eigenstate of the operator a with eigenvalue , i.e.
| (t = 0) = |c , where a |c = |c .
Eyal Buks
114
5.4. Solutions
a) Find a general expression for the state of the system | (t) at time
t > 0.
b) Evaluate | (t) at time t = 2/ 1 .
c) Evaluate | (t) at time t = /1 .
d) Evaluate | (t) at time t = /2 1 for the case where k is even.
41. Consider two normalized coherent states | and |, where , C.
The operator A is defined as
A = | | | | .
(5.112)
Find the eigenvalues of the operator A.
5.4 Solutions
H=
p2
m2 x2
+
.
2m
2
Using Eqs. (3.21), (3.29), (5.9) and (5.10) one has

2 1/2
2 d n
x | a |n = 2x0
x n (x ) + x0
,
dx

1/2
d
x | a |n = 2x20
x n (x ) x20 n
,
dx
(5.113)
(5.114)
where
x0 =

.
m
(5.115)
For the ground state |0, according to Eq. (5.28), a |0 = 0, thus
x 0 (x ) + x20
d0
=0.
dx

!
1
0 (x ) = A0 exp
2
(5.116)
x
x0
2 "
(5.117)
where the normalization constant A0 is found from the requirement

2
|0 (x )| dx = 1 ,
(5.118)
thus
Eyal Buks
115

! "
2
x
|A0 |
exp
dx = 1 .
x0

2
(5.119)
x0
Choosing A0 to be real leads to

!
2 "
1
1 x
0 (x ) =
.
exp
1/2
2 x0
1/4 x
(5.120)
All other wavefunctions are found using Eqs. (5.32) and (5.114)

n
1
2 d
n (x ) =
x
x
0 (x )
0
dx
(2x0 )n/2 n!
!
2 "

n
1
1
1 x
2 d
=
x x0
exp
.
dx
2 x0
1/4 2n n! x0n+1/2
(5.121)
Using the notation
2

n

X
X2
d
exp
Hn (X) = exp
X
,
2
dX
2
the expression for n (x ) can be rewritten as
2

x
exp 2x
Hn xx0
2
0
n (x ) =
.
1/2 n
1/4 x0
2 n!
(5.122)
(5.123)
The term Hn (X), which is called the Hermite polynomial of order n, is

calculated below for some low values of n
H0 (X) = 1 ,
H1 (X) = 2X ,
H2 (X) = 4X 2 2 ,
H3 (X) = 8X 3 12X ,
H4 (X) = 16X 4 48X 2 + 12 .
(5.124)
(5.125)
(5.126)
(5.127)
(5.128)
2. Therelation (5.56),
which is a Taylor expansion of the function f (t) =

exp 2Xt t2 around the point t = 0, implies that

dn
Hn (X) = n exp 2Xt t2
.
(5.129)
dt
t=0
2
The identity 2Xt t2 = X 2 (X t) yields
Eyal Buks
116
5.4. Solutions

2 dn
2
Hn (X) = exp X
exp (X t)
.
dtn
t=0
(5.130)
Moreover, using the relation

d
d
2
2
exp (X t) =
exp (X t) ,
dt
dX
(5.131)
one finds that

dn
2
Hn (X) = exp X 2 (1)n

exp
(X
t)
dX n
t=0
n

2

n d
2
= exp X (1)
exp X .
dX n
Note that for an arbitrary function g (X) the following holds

d

d
exp X 2
exp X 2 g = 2X
g,
dX
dX
(5.132)
(5.133)
and
exp
X2
2

d
X2
d
X
exp
g = 2X
g,
dX
2
dX
(5.134)
thus
Hn (X) = exp
X2
2

n

d
X2
X
exp
.
dX
2
(5.135)
3. With the help of Eq. (5.132) and the general identity

exp ax2 + bx + c dx =
2
14 4ca+b
a
e
,
a
(5.136)
according to which the following holds (for the case a = 1, b = 2iX and
c = 0)

1
exp X 2 =

exp x2 + 2iXx dx ,
(5.137)
one finds that
Eyal Buks
117

n

exp X 2
d
Hn (X) =
exp x2 + 2iXx dx
dX
1
=
1
=

n
(2ix) exp X 2 x2 + 2iXx dx
2
(2ix) e(X+ix) dx ,
(5.138)
thus the following holds [see Eq. (5.136)]

n
n

Hn (X) Hn (Y )
2
2 (2xy)
1
2
dx
dy e(X+ix) e(Y +iy)
=
n!
n!
n=0
n=0

e2xy
1
=
dy e(Y +iy) e2xy

x(x2iY )
e
2
2
2
dx e(1 )x +2i(XY )x+X
exp
=
dx e(X+ix)
(2XY X 2 Y 2 )
12
1 2
.
(5.139)
4. With the help of Eqs. (5.9), (5.10), (5.11), (5.12) and (5.13) one finds
n| x |n = 0 ,

n| x2 |n =
n| aa + a a |n =
(2n + 1) ,
2m
2m
n| p |n = 0 ,
m
m
n| p2 |n =
n| aa + a a |n =
(2n + 1) ,
2
2
thus
2
'
('
(
1
2
2
(x)
(p) = n +
2 .
2
(5.140)
(5.141)
(5.142)
(5.143)

H=
Eyal Buks
p2
.
2m
(5.144)
118
5.4. Solutions
Using Eqs. (4.37) and (5.8) one finds that
and
dx(H)
1 3 (H) (H) 4 p(H) 3 (H) (H) 4 p(H)
=
x ,H
=
x ,p
=
,
dt
i
im
m
(5.145)
dp(H)
1 3 (H) (H) 4
=
p ,H
=0.
dt
i
(5.146)
The solution is thus
x(H) (t) = x(H) (0) +
1 (H)
p (0) t .
m
(5.147)
With the help of Eq. (5.147) one finds that

(
'
) *
(x)2 (t) = x2 (t) (x (t))2

2
1 (H)
(H)
= 0 | x (0) + p (0) t |0
m

2
1
0 | x(H) (0) + p(H) (0) t |0
m
2
t
2t
= (x)20 + 2 (p)20 + ((xp)0 x0 p0 ) .
m
m
(5.148)
H=
p2
m2 x2
+
.
2m
2
(5.149)
and
dx(H)
1 3 (H) (H) 4 p(H)
=
x ,H
=
,
dt
i
m
(5.150)
dp(H)
1 3 (H) (H) 4
=
p ,H
= m2 x(H) .
dt
i
(5.151)
a) The solutions of the above equations are given by

x(H) (t) = x(H) (0) cos (t) +
sin (t) (H)

p (0) ,
m
(5.152)
and
p(H) (t) = p(H) (0) cos (t) m sin (t) x(H) (0) .
Eyal Buks
(5.153)
119

b) Using the expressions for x(H) (t) and p(H) (t) and Eq. (5.8) one finds
that 3
4
p(H) (t1 ) , x(H) (t2 )
3
4
= (cos (t1 ) cos (t2 ) + sin (t1 ) sin (t2 )) x(H) (0) , p(H) (0)
= i cos ( (t1 t2 )) ,
3
4
p(H) (t1 ) , p(H) (t2 )
(5.154)
3
4
= m (cos (t1 ) sin (t2 ) sin (t1 ) cos (t2 )) x(H) (0) , p(H) (0)
= im sin ( (t1 t2 )) ,
(5.155)
and 3
4
x(H) (t1 ) , x(H) (t2 )
3
4
1
(cos (t1 ) sin (t2 ) sin (t1 ) cos (t2 )) x(H) (0) , p(H) (0)
m
i
=
sin ( (t1 t2 )) .
m
(5.156)
7. Due to the infinite barrier for x 0 the wavefunction must vanish at
x = 0. This condition is satisfied by the wavefunction of all number
states |n with odd value of n (the states |n are
of the regu
eigenstates
lar harmonic oscillator with potential V (x) = m 2 /2 x2 ). These wavefunctions obviously satisfy the Schrdinger equation for x > 0.
a) Thus the possible energy values are Ek = (2k + 3/2) where k =
0, 1, 2, .
b) The corresponding normalized wavefunctions are given by
+
22k+1 (x) x > 0
k (x) =
,
(5.157)
0
x0
=
where n (x) is the wavefunction of the number states |n. Thus for
a given k
Eyal Buks
120
5.4. Solutions
) 2*
x k=

0
=2
2
dx
k (x)
x

0
2
dx
2k+1 (x)
x2
2
dx
2k+1 (x)
x2
= 2k + 1| x2 |2k + 1 ,
thus with the help of Eq. (5.141) one finds that

) 2*

3
2k +
.
x k=
m
2
(5.158)
(5.159)
8. The potential can be written as

V (x) =
m2
2
2
.
x+
2
m 2
2m 2
(5.160)
This describes a harmonic oscillator centered at x0 = /m 2 having

angular resonance frequency . The last constant term represents energy
shift. Thus, the eigenenergies are given by
En = (n + 1/2) 2 /2m2 ,
(5.161)
where n = 0, 1, 2, .
9. In the classically forbidden region V (x) > E0 = /2, namely |x| > x0
where
0

x0 =
.
(5.162)
m
Using Eq. (5.120) one finds

p=2
|0 (x)|2 dx
x0
! 2 "

2
x
= 1/2
exp
dx
x0
x0 x0
= 1 erf (1)
= 0.157 .
(5.163)
Eyal Buks
121

10. The answer is [see Eqs. (4.181) and (5.19)]

2 2 n2x + n2y
1
,
Enx ,ny ,nz =
+

n
+
z
2ma2
2
(5.164)
where nx and ny are positive integers and nz is a nonnegative integer.

11. With the help of Eq. (4.14) one has

i0 t
1
| (t) = e 2 |0 + ei0 t |1 .
2
Moreover, the following hold

0

a + a ,
x=
2m0
0

m0
p=i
a + a ,
2
a |n = n |n 1 ,
a |n = n + 1 |n + 1 ,
, a, a = 1 ,
thus
a)
(5.165)
(5.166)
(5.167)
(5.168)
(5.169)
(5.170)

(t)| a + a | (t)
2m 0
0

1
=
0| + ei0 t 1| a + a |0 + ei0 t |1
2m 0 2
0

1 i0 t
=
+ ei0 t
e
2m 0 2
0

=
cos ( 0 t) .
2m 0
(5.171)
x =
b)

m 0
(t)| a + a | (t)
2

m 0 1
=i
0| + ei0 t 1| a + a |0 + ei0 t |1
2 2
0
m 0
=
sin ( 0 t) .
2
(5.172)
p = i
0
c)
Eyal Buks
122
5.4. Solutions
) 2*
x =
2

(t)| a + a | (t)
2m 0

2 ,

=
(t)| a2 + a + a, a + 2a a | (t)
2m 0

1
=
1+2
2m 0
2

=
.
m 0
(5.173)
d) Similarly
) 2*

2
m 0
p =
(t)| a + a | (t)
2

2 ,
m 0
(t)| a2 + a a, a 2a a | (t)
=
2
= m0 ,
(5.174)
thus
:
0
cos2 ( 0 t)
sin2 (0 t)
xp = 1
1
2
2
0

1
=
2 + sin2 (2 0 t) .
2
4
(5.175)
12. The state | (t) is given by

1
iE0 t
iEn t
| (t) = exp
|0 + exp
|n ,
(5.176)

2
where

1
En = n +
,
2
thus, using
0
x=
and

a + a ,
2m
a |n = n |n 1 ,
a |n = n + 1 |n + 1 ,
(5.177)
(5.178)
(5.179)
(5.180)
one finds that x (t) = 0 if n > 1, and for n = 1
Eyal Buks
123

0

(t)| a + a | (t)
2m
0

cos (t) .
=
2m
x (t) =
(5.181)
13. Since H = and |(0) is normalized one has

|c0 |2 = |c1 |2 =
1
,
2
thus |(0) can be written as

0

1
|(0) =
|0 + ei |1 ,
2
where is real. Given that at time t = 0

0
1

x (t = 0) =
,
2 m
one finds using the identities
0

x=
a + a ,
2m
a |n = n |n 1 ,
a |n = n + 1 |n + 1 ,
that
cos =
2
.
2
(5.182)
(5.183)
(5.184)
(5.185)
(5.186)
(5.187)
(5.188)
Using this result one can evaluate p (t = 0), where

0

m
a + a ,
p=i
2
(5.189)
thus
p (t = 0) =
m
m 2
sin =
= mx (t = 0) . (5.190)
2
2 2
Using these results together with Eq. (5.152) yields

0
1

x (t) =
(cos (t) sin (t))
2 m
0

=
cos t
.
2m
4
Eyal Buks
(5.191)
124
5.4. Solutions
14. According to identity (2.175), which states that
1
eA+B = eA eB e 2 [A,B] = eB eA e 2 [A,B] ,
(5.192)
provided that
[A, [A, B]] = [B, [A, B]] = 0 ,
(5.193)
one finds with the help of Eq. (5.13) that

D () = exp a a
= e
=e
||2
2
||2
2
ea e
e a ea .
(5.194)

D () = e
||2
2
ea e
=e
||2
2
e a ea ,
(5.195)
thus
D () D () = D () D () = 1 .
(5.196)
16. Using Eqs. (5.35), (5.28) and (5.29) one finds that
| = e
||2
2
ea e a |0 = e
n
||2
|n .
= e 2
n!
n=0
||2
2
ea |0
(5.197)
17. Using Eqs. (5.42) and (5.28) one has

a | = e
||2
2
= e

n
a |n
n!
n=0
||2
2
= | .
n1

|n 1
(n 1)!
n=1
18. Using Eqs. (5.36), (5.9) and (5.10) one has

0

0
m
1
D () = exp
( ) x i
( + ) p ,
2
2m
(5.198)
(5.199)
thus with the help of Eqs. (2.175) and (5.8) the desired result is obtained
Eyal Buks
125

m
x

2

2

i
+
2
p exp
exp
.
4
2
m
D () = exp
0
(5.200)
19. Using the operator identity (2.173)

eL AeL = A + [L, A] +
1
1
[L, [L, A]] + [L, [L, [L, A]]] + , (5.201)
2!
3!
and the definition (5.36)

D () = exp a a ,
(5.202)
one finds that
D () aD () = a + ,
D () a D () = a + .
(5.203)
(5.204)
Exploiting the unitarity of D ()

D () D () = 1

it is straightforward to show that for any function f a, a having a
power series expansion the following holds

D () f a, a D () = f a + , a +
(5.205)
(e.g., D a2 D = D aDD aD = (a + )2 ).
20. Using Eq. (5.75) and the following identities

1
H = a a +
,
(5.206)
2
0

a + a ,
x=
(5.207)
2m
0

m
p=i
a + a ,
(5.208)
2
all these relations are easily obtained.
21. Expressing the state | in the basis of eigenvectors of the Hamiltonian
|n
| =
n=0
cn |n ,
(5.209)
using
Eyal Buks
126
5.4. Solutions
a | = | ,
(5.210)
and
a |n =
n |n 1 ,
(5.211)
one finds
n=0
cn n |n 1 =
cn |n ,
(5.212)
n=0
thus
cn+1 =
cn ,
n+1
(5.213)
therefore
| = A

n
|n .
n!
n=0
The normalization constant A is found by

n
||2

2
1 = |A|2
= |A|2 e|| .
n!
n=0
(5.214)
(5.215)
Choosing A to be real yields

A = e
||2
2
(5.216)
n
.
n!
(5.217)
thus
cn = e
||2
2
Note that this result is identical to Eq. (5.42), thus | is a coherent

state. The possible results of the measurement are

1
En = n +
,
(5.218)
2
and the corresponding probabilities, which are time independent, are
given by

n
2
||
2
pn (t) = |cn |2 = e||
.
(5.219)
n!
Eyal Buks
127

22. Using the relations
0
2
Re () ,
x =
m
p = 2m Im () ,
(5.220)
(5.221)
Eq. (5.74) can be written as

2

i p x
i x p
2
D () = exp
exp
exp
. (5.222)

4

i x p
|x = |x + x ,
exp

thus

2

i x p
i p x
2
x | = x | exp
exp
exp
|0

4
2

i p x
2
= exp
exp
x x |0 .
4

(5.223)
Using Eq. (5.120) the wavefunction of the ground state is given by

!
2 "
1
1
x
x |0 =
exp
,
(5.224)
2x
(2)1/4 x
where
x =

,
2m
(5.225)
thus

2
x x

exp
2x
i p x
2 2
x | = exp
exp
1/4
4

(2)
x

2

2

2
1/4
x x

m
x
= exp
exp
+ i p
.
4

2x

(5.226)
23. Using Eqs. (5.36) and (2.175) this relation is easily obtained.
24. With the help of Eq. (5.42) one has

2
1
1
1
| | d2 =
|n m|
e|| n m d2 .
n,m
n!m!
Eyal Buks
128
5.4. Solutions
(5.227)
Employing polar coordinates in the complex plane = ei , where is
non-negative real and is real, leads to
1

2
1
1
| | d =
|n m|
dn+m+1 e
dei(nm)
n,m
n!m!
0
0

2
2nm

n

n

n
|n n|
2
n!
d2n+1 e
1
|n n| (n + 1)
n!
=n!
|n n|
=1.
(5.228)
25. Using Eqs. (5.35) and (5.41) one finds that
| = 0| D () D () |0
= 0| D () D () |0

+
= exp
0| D ( + ) |0
2

+
= exp
0 | .
2
(5.229)
Thus, with the help of Eq. (5.42) one has

2
+ ||
2
| = exp
e
2
!
"
2
||
||2
= exp
+
2
2
!
"
| |2
= exp
+ i Im ( ) .
2
(5.230)
Eyal Buks
129

1
1
m2 (x + )2 m 2 2
2
2
1
1
= m2 x2 m 2 2 ,
2
2
V1 (x) =
(5.231)
where
x = x + .
(5.232)
a) Thus, using Eqs. (5.152) and (5.153) together with the relations
x(H) (t) = x(H) (t) + ,
(5.233)
(H)
(H)
p (t) = p
(t) ,
(5.234)
one finds

sin (t) (H)
x(H) (t) = x(H) (0) + cos (t) +
p (0) , (5.235)
m

p(H) (t) = p(H) (0) cos (t) m sin (t) x(H) (0) + . (5.236)
b) For '
this case (at time t = 0
x(H) (0) = 0 ,
(
'
p(H) (0) = 0 ,
thus
'
(
x(H) (t) = (cos (t) 1) .
27. The state of the system at time t = 0 is given by

ix
p |0 ,
| (t = 0) = exp
(5.237)
(5.238)
(5.239)
(5.240)
where |0 is the ground state of the potential V2 . In general a coherent

state with parameter can be written as
| = exp
0

2

m
+
i
2
x exp
p exp
|0 .

4
2
2
m
(5.241)
a) Thus | (t = 0) = |0 , where

0
m
0 = x
.
2
The time evolution of a coherent state is given by
*
| (t) = eit/2
= 0 eit ,
Eyal Buks
(5.242)
(5.243)
130
5.4. Solutions
0
,
2
x (t) =
Re 0 eit = x cos (t) ,
m
b) According to Eq. (5.48)

(
'

2
.
(x) (t) =
2m
(5.244)
(5.245)
c) In general a coherent state can be expanded in the basis of number

states |n
| = e||
/2
n
|n ,
n!
n
(5.246)
thus the probability to measure energy En = (N + 1/2) at time t

is given by
Pn = |n| (t)|2 =
2

n
e|0 | 2n
1
m2x
m2x
0
.
=
exp
n!
n!
2
2
(5.247)
28. At time t = 0 the following holds

x = 0 ,
p = 0 ,
'
( ) *

(x)2 = x2 =
,
2m
'
( ) * m
(p)2 = p2 =
.
2
Moreover, to calculate xp it is convenient to use
0

x=
a + a ,
2m
0

m
p=i
a + a ,
2
, a, a = 1 ,
(5.248)
(5.249)
(5.250)
(5.251)
(5.252)
(5.253)
(5.254)
thus at time t = 0

xp = i 0| aa a a |0 = i .
2
2
(5.255)
The Hamiltonian for times t > 0 is given by

H=
Eyal Buks
p2
+ gx .
2m
(5.256)
131

Using the Heisenberg equation of motion for the operators x and x2 one
finds
dx(H)
1 ,
(5.257)
=
x(H) , H ,
dt
i
dp(H)
1 ,
=
p(H) , H ,
(5.258)
dt
i
4
dx2(H)
1 3 2
=
x(H) , H ,
(5.259)
dt
i
or using [x, p] = i
thus
dx(H)
p(H)
=
,
dt
m
dp(H)
= g ,
dt
dx2(H)

1
1
=
x p + p(H) x(H) =
2x(H) p(H) i ,
dt
m (H) (H)
m
p(H) (t) = p(H) (0) gt ,

x(H) (t) = x(H) (0) +
(5.260)
(5.261)
(5.262)
(5.263)
p(H) (0) t gt2
,
(5.264)
m
2m

2 t
+
x (t ) p(H) (t ) dt
m 0 (H)

p(H) (0) t gt2 ,
2 t
x(H) (0) +
+
p(H) (0) gt dt
m 0
m
2m
it
m
it
= x2(H) (0)
m
it
= x2(H) (0)
m
"
t!
2
2 3
p2(H) (0) t gt2
p
(0)
gt
2
g
t
(H)
+
x(H) (0) p(H) (0) +
p (0) x(H) (0) gt

+
dt
m 0
m
2m (H)
m
2m
x2(H) (t) = x2(H) (0)
it
= x2(H) (0)
m
!
"
p2(H) (0) t2 p(H) (0) gt3 x(H) (0) gt2 p(H) (0) gt3 g2 t4
2
+
x(H) (0) p(H) (0) t +
+
.
m
2m
6m
2
3m
8m
(5.265)
Using the initial conditions Eqs. (5.248), (5.249), (5.250), (5.251) and
(5.255) one finds
x (t) =
Eyal Buks
gt2
,
2m
(5.266)
132
5.4. Solutions
g 2 t4
,
4m2
p (t) = gt ,
2
x (t) =
and
) 2 *
x (t) =
(5.267)
(5.268)

it
2
+
2m
m
m

it t2 g2 t4
+
+
,
2
4
8m
'
( )
*
(x)2 (t) = x2 (t) x (t)2 =
(5.269)

t2

+
=
1 + 2 t2 .
2m
2m
2m
(5.270)
29. Using the operator identity (2.173), which is given by

eL OeL = O + [L, O] +
1
1
[L, [L, O]] + [L, [L, [L, O]]] + , (5.271)
2!
3!
for the operators

O=a,

r 2
L=
a2 ,
a
2
and the relations
, a, a = 1 ,
(5.272)
(5.273)
(5.274)
[L, O] = ra ,
[L, [L, O]] = r2 a ,
[L, [L, [L, O]]] = r3 a ,
[L, [L, [L, [L, O]]]] = r4 a ,
etc., one finds

r 2 r4
r3
T = 1+
+
+ a r +
+ a + ,
2!
4!
3!
(5.275)
(5.276)
(5.277)
(5.278)
(5.279)
a) Thus
T = Aa + Ba ,
where
A = cosh r ,
B = sinh r .
b) Using the0relations

x=
a + a ,
2m
0

m
p=i
a + a .
2
Eyal Buks
(5.280)
(5.281)
(5.282)
(5.283)
(5.284)
133

one finds

0| S (r) a + a S (r) |0
2m
0

=
0| T |0 + 0| T |0
2m
=0,
r| x |r =
(5.285)

m
r| p |r = i
0| S (r) a + a S (r) |0
2
0

0| T |0 + 0| T |0
=
2m
=0.
(5.286)
c) Note that S (r) is unitary, namely S (r) S (r) = 1, since the operator
2
a2 a is anti Hermitian. Thus

r| x2 |r =
0| S (r) a + a a + a S (r) |0
2m

=
0| S (r) a + a S (r) S (r) a + a S (r) |0
2m

2

=
0| T + T |0
2m

2
(A + B)2
=
0| a + a |0
2m
(cosh r sinh r)2
=
2m
e2r
=
,
2m
(5.287)
and

2
m
r| p2 |r =
0| S (r) a a S (r) |0
2
2

m
=
0| T T |0
2

2
m (A B)2
=
0| a a |0
2
m (cosh r + sinh r)2
=
2
me2r
=
.
2
(5.288)
Thus
Eyal Buks
134
5.4. Solutions
e2r
,
2m
me2r
2
,
(p) =
2

(x) (p) = .
2
30. With the help of Eqs. (5.9) and (5.10) one finds that
0

r
Q (r) + Q (r) ,
x=e
2m
0

m
r
Q (r) Q (r) ,
p = ie
2
2
(x) =
(5.289)
(5.290)
(5.291)
(5.292)
(5.293)

r| x |r = 0 ,
r| p |r = 0 .
(5.294)
(5.295)
Using the commutation relation

-
,
,
Q (r) , Q (r) = cosh2 r sinh2 r a, a = 1 ,
one obtains
2

e2r
r| Q (r) + Q (r) |r
2m
e2r
=
r| Q (r) Q (r) |r
2m
e2r
=
,
2m
r| x2 |r =
(5.296)
and similarly
2

me2r
r| Q (r) Q (r) |r
2
me2r
=
r| Q (r) Q (r) |r
2
me2r
,
=
2
r| p2 |r =
(5.297)
thus
e2r
,
2m
me2r
(p)2 =
.
2
(x)2 =
Eyal Buks
(5.298)
(5.299)
135

31. Using the commutation relation
, a, a = 1 ,
(5.300)
one finds

H = 0 N + 1 N 2 N ,
(5.301)
where
N = a a
(5.302)
is the number operator.

a) The eigenvectors of N
N |n = n |n ,
(5.303)
(where n = 0, 1, ) are also eigenvectors of H and the following

holds
H |n = En |n ,
(5.304)
where
,

En = 0 n + 1 n2 n .
(5.305)
Note that
En+1 En
= 0 + 2 1 n ,

thus En+1 > En .
b) Using the relations
0

x=
a +a ,
2m 0
0

m 0
p=i
a a ,
2

x2 =
a a + aa + 2N + 1 ,
2m 0

m0
2
p =
a a aa + 2N + 1 ,
2
a |n = n |n 1 ,
a |n = n + 1 |n + 1 ,
one finds
i. 0|x|0 = 0
ii. 0|p|0 = 0
Eyal Buks
(5.306)
(5.307)
(5.308)
(5.309)
(5.310)
(5.311)
(5.312)
136
5.4. Solutions
'
(
2
0| (x) |0 =
'
(
iv. 0| (p)2 |0 =
iii.

2m0
m0
2
32. The proof of the clue is:

N 2 = b bb b = b 1 b b b = N .
(5.313)
Moreover, N is Hermitian, thus N is a projector.

a) Let |n be the eigenvectors of N and n the corresponding real eigenvalues (N is Hermitian)
N |n = n |n .
(5.314)
Using the clue one finds that n2 = n, thus the possible values of n
are 0 (ground state) and 1 (excited state). Thus, the eigenvalues of
H are 0 and .
b) To verify the statement in the clue we calculate
N b |0 = b bb |0 = b (1 N ) |0 = b |0 ,
(5.315)
thus the state b |0 is indeed an eigenvector of N with eigenvalue 1

(excited state). In what follows we use the notation
|1 = b |0 .
(5.316)
Note that |1 is normalized since

1|1 = 0| bb |0 = 0| (1 N ) |0 = 0|0 = 1 .
(5.317)
Moreover, since |0 and |1 are eigenvectors of an Hermitian operator

with different eigenvalues they must be orthogonal to each other
0|1 = 0 .
(5.318)
Using Eqs. (5.316), (5.317) and (5.318) one finds

+|+ = 2 |A+ |2 ,
(5.319)
| = 2 |A | .
(5.320)
choosing the normalization constants to be non-negative real numbers lead to
1
A+ = A = .
2
(5.321)
c) Using N 2 = N one finds
Eyal Buks
137

n

iHt
1
iHt
exp
= 1+

n!

n=1

n

1
it
= 1+N
n!

n=1
!
"
1 it n
= 1 + N 1 +
n!

n=0

it
.
= 1 + N 1 + exp
(5.322)
Thus

iHt
p0 (t) =
| exp
|+
2

it
=
(0| 1|) 1 + N 1 + exp

(|0 + |1)
4

2
1
it
1 exp
4

t
= sin2
.
2
(5.323)
33. The closure relation (5.31) can be written as
1=
n,m=0
|n m| n,m .
With the help of Eq. (5.32) together with the relation

n
1
d
m
= n,m ,
n! d
=0
(5.324)
(5.325)
which is obtained using the Taylor power expansion series of the function
m , one finds that
Eyal Buks
138
5.4. Solutions
1=
n,m=0
|n m| n,m
n
d
1
|n m|
m
d
n! m!
=0
n,m=0

n
m

a
a
d
=
|0 0|
m
n!
m! d
=0
n,m=0
! n n "
!
"
d
m m
a

a
d
=
|0 0|
n!
m!
n=0
m=0
=0

d
|0 0| exp (a)
= exp a
.
d
=0
=
(5.326)
Denote the normal ordering representation of the operator |0 0| by Z,

i.e.
|0 0| = : Z : .
(5.327)
For general functions f , g and h of the operators a and a it is easy to

show that the following holds
: f g : =: gf : ,
(5.328)
: f gh : =: f hg : ,
(5.329)
: f (: g : ) : = : f g : .
(5.330)
and
Thus
Eyal Buks
139

1=
=
=
=

d
exp a
: Z : exp (a)
=0

d
Z exp (a)
: exp a
:
d
=0

d
exp (a) Z
: exp a
:
d
=0
d n
m
a d
(a)
:
Z
n!
m!
n,m
=0
d n
n d
am
:
a
m
Z:
n!
m!
n,m
=0

n,m

= : exp a a Z :

= : exp a a ( : Z : ) : ,
(5.331)
and therefore

|0 0| = : exp a a : .
(5.332)
Using again Eq. (5.32) one finds that

Pn = |n n| =

n
1 n
: a
exp a a a : .
n!
(5.333)
34. The Hamiltonian H, which is given by

H=
p2
m2 x2
+
+ xf (t) ,
2m
2
(5.334)
can be expressed in terms of the annihilation a and creation a operators

[see Eqs. (5.11) and (5.12)] as
0

1

H = a a +
+ f (t)
a + a .
(5.335)
2
2m
The Heisenberg equation of motion for the operator a is given by [see

Eq. (4.37)]
0
da
1
= ia i
f (t) .
(5.336)
dt
2m
The solution of this first order differential equation is given by
0
t
1
i(tt0 )
a (t) = e
a (t0 ) i
dt ei(tt ) f (t ) ,
2m t0
Eyal Buks
(5.337)
140
5.4. Solutions
where the initial time t0 will be taken below to be . The Heisenberg
operator a (t) is found from the Hermitian conjugate of the last result.
Let Pn (t) be the Heisenberg representation of the projector |n n|. The
probability pn (t) to find the oscillator in the number state |n at time t
is given by
pn (t) = 0| Pn (t) |0 .
(5.338)
To evaluate pn (t) it is convenient to employ the normal ordering representation of the operator Pn (5.100). In normal ordering the first term of
Eq. (5.337), which is proportional to a (t0 ) does not contribute to pn (t)
since a (t0 ) |0 = 0 and also 0| a (t0 ) = 0. To evaluate pn = pn (t )
the integral in the second term of Eq. (5.337) is evaluate from t0 =
to t = +. Thus one finds that
pn =
e n
,
n!
(5.339)
where
=
it

dt
e
f
(t
)
.
2m
(5.340)
35. As can be seen from the definition of P, the following holds

x | P | =
dx x |x x |
= x | ,
(5.341)
thus the wave function of P | is (x ) given that the wave function

of | is (x ). For the wavefunctions n (x ) = x |n of the number
states |n, which satisfy N |n = n |n, the following holds
+
n (x ) n odd
n (x ) =
,
(5.342)
n (x ) n even
thus
P |n =
|n n odd
,
|n n even
(5.343)
or P |n = (1)n |n ,and consequently the parity operator P can be

expressed as a function of N
P = eiN .
Eyal Buks
(5.344)
141

36. Using Eqs. (5.31), (5.32) and (5.332) together with the relation
a a |n = n |n ,
(5.345)
yields
ea
=
=
=
=
n=0
en |n n|
n
n=0
n
a
an
|0 0|
n!
n!

en n
a
: exp a a : an
n!
n=0

en
n!
n=0
n

: a exp a a an :

en n
a a exp a a :
n!
n=0

=: exp e a a exp a a : ,
=:
(5.346)
thus
ea
=: exp
,

e 1 a a : .
(5.347)
37. The following holds [see Eq. (5.31)]

|x =
n=0
|n n |x ,
(5.348)
where
n
a
|n = |0 ,
n!
(5.349)
thus with the help of Eq. (5.123) and the generating function of Hermite
polynomials (5.56) one finds that (note that x |n is real)
2

n
x
x
H
exp 2x

2
n x0
a
0
|0
|x =
1/2
2n n!
n!
1/4 x0
n=0
2

2
x
exp 2x
2 xx0 a a2
2 +
0
=
|0 .
1/2
1/4 x0
(5.350)
Eyal Buks
142
5.4. Solutions
38. Using the relation x |x = x |x and Eq. (3.32) one finds that

exp kx2 =
dx ekx |x x | .
(5.351)
Eqs. (5.332) and (5.103) yield

2
1
|x x | =
: e(X X ) : ,
x0
(5.352)
where
a + a
x
=
,
x0
2
X=
(5.353)
and where
X =
x
.
x0
(5.354)
Thus

1
exp kx2 =
1
=
dX : e(X
X ) +KX 2
dX : e(1K)X
+2X XX 2
: ,
(5.355)
where K = kx20 . With the help of the identity (5.136) this becomes

Kx2
1
K x2
exp
: exp
=
: .
(5.356)
x20
1 K x20
1K
Using the notation
=
K
,
1K
the results can be also expressed as

2
1
x2
x
exp
=: exp 2 : .
2
1
+
x
x
1+
0
0
(5.357)
(5.358)
39. The orthogonality between number states yields according to Eq. (5.123)
Eyal Buks
143
m |n =
=
2

exp xx2 Hm xx0 Hn xx0
0
dx
2m m!2n n!x0
dX
exp

X 2 Hm (X ) Hn (X )
2m m!2n n!
= nm .
(5.359)
2
Multiplying Eq. (5.56) by the factor ez Hm (z)

2
e(zt) Hm (z) = ez Hm (z)
Hn (z)
n=0
tn
,
n!
(5.360)
and integrating over z

dz e(zt) Hm (z)
n

t
n!
n=0
dz ez Hm (z) Hn (z) ,
(5.361)
yields with the help of Eq. (5.359)

dz e(zt) Hm (z) = (2t)m
(5.362)

The relations x |x = x |x , X = a + a / 2 = x/x0 together with
Eq. (5.352) yield
Hn (X) =
dx Hn
1
= :
x
x0
|x x |
dX e(X
X )
Hn (X ) : ,
(5.363)
thus, with the help of Eq. (5.362) one finds that

Hn (X) =: (2X)n : .
Eyal Buks
(5.364)
144
5.4. Solutions
The last result together with the identity
dHn
= 2nHn1 (X ) ,
dX
(5.365)
yields
d
1 dHn (X)
: Xn : = n
dX
2
dX
Hn1 (X)
=n
2n1
= n : X n1 : ,
(5.366)
thus
d
d n
: X n : =:
X : .
dX
dX
(5.367)
Thus, for a general smooth function F (X) of the operator X the following
holds
d
dF
: F (X) : =:
: .
dX
dX
(5.368)
40. Initially, the system is in a coherent state given by Eq. (5.42)

| (t = 0) = |c = e
||2
2

n
|n .
n!
n=0
(5.369)
The notation |c is used to label coherent states satisfying a |c =

|c .
k
a) Since a a commutes with a a , the time evolution operator is given
by [see Eq. (4.9)]

k
iHt
u (t) = exp
(5.370)
= ei1 (a a) t eia at ,

thus
| (t) = u (t) | (t = 0)

n
|n
n!
n=0
it n
k
||2 e
= ei1 (a a) t e 2
|n
n!
n=0

it n
||2 e
= e 2
ein |n ,
n!
n=0
2
k
||
= ei1 (a a) t eia at e 2
Eyal Buks
(5.371)
145

where
n = 1 tnk .
(5.372)
b) At time t = 2/ 1 the phase factor n is given by n = 2nk , thus

ein = 1 ,
(5.373)
and therefore
6
2i
2
=
e 1
.
1
c
(5.374)
c) At time t = /1 the phase factor n is given by n = nk . Using

the fact that
+

0 n is even
mod nk , 2 =
,
(5.375)
1 n is odd
one has
n
ein = (1) ,
(5.376)
and therefore

n
i
6
1
e
2

||
(1)n |n
= e 2
1
n!
n=0
i
.
=
e 1
c
(5.377)
d) At time t = /2 1 the phase factor n is given by n = (/2) nk .
For the case where k is even one has
+

0 n is even
mod nk , 4 =
,
(5.378)
1 n is odd
thus
in
1 n is even
,
i n is odd
(5.379)
and therefore

n
i
2

6
1
e
2
||
= e 2
ein |n .
2 1
n!
n=0
(5.380)
This state can be expressed as a superposition of two coherent states

6
(
(
1 i
i
i
4
e 2 1
4
e 2 1
=
e
+
e
. (5.381)
2 1
c
c
2
Eyal Buks
146
5.4. Solutions
41. Let {n } be the set of eigenvalues of A. Clearly A is Hermitian, namely
A = A, thus the eigenvalues n are expected to be real. Since the trace
of an operator is basis independent, the following must hold

Tr (A) =
n ,
(5.382)
n
and
2
Tr A2 =
n .
(5.383)
On the other hand, with the help of Eq. (2.172) one finds that
Tr (A) = Tr (| |) Tr (| |) = 0 ,
(5.384)
and

Tr A2 = Tr (| | |) + Tr (| | |)
Tr (| | |) Tr (| | |)
= 2 | Tr (| |) | Tr (| |)

= 2 1 | ||2 .
(5.385)
Clearly, A cannot have more than two nonzero eigenvalues, since the
dimensionality of the subspace spanned by the vectors {| , |} is at
most 2, and therefore A has three eigenvalues 0, + and , where [see
Eq. (5.230)]
.

2
(5.386)
= 1 | ||2 = 1 e|| .
Eyal Buks
147
6. Angular Momentum
Consider a point particle moving in three dimensional space. The orbital

angular momentum L is given by
x
y

z
px py pz
where r = (x, y, z) is the position vector and where p = (px , py , pz ) is the
momentum vector. In classical physics the following holds:
Claim.
{Li , Lj } = ijk Lk ,
where
ijk
i, j, k are not all different

0
1 i, j, k is an even permutation of x, y, z .
=
1 i, j, k is an odd permutation of x, y, z
(6.1)
(6.2)
Proof. Clearly, Eq. (6.1) holds for the case i = j. Using Eq. (1.48), which
reads
{xi , pj } = ij ,
(6.3)
one has
{Lx , Ly } = {ypz zpy , zpx xpz }
= {ypz , zpx } + {zpy , xpz }
= y {pz , z} px + x {z, pz } py
= ypx + xpy
= Lz .
(6.4)
In a similar way one finds that {Ly , Lz } = Lx and {Lz , Lx } = Ly . These
results together with Eq. (1.49) complete the proof.
Using the rule (4.41) {, } (1/i) [, ] one concludes that in quantum
mechanics the following holds:
[Li , Lj ] = iijk Lk .
(6.5)
Chapter 6. Angular Momentum
6.1 Angular Momentum and Rotation

We have seen before that the unitary operator u (t, t0 ) is the generator of time
evolution [see Eq. (4.4)]. Similarly, we have seen that the unitary operator

i p
J () = exp
(6.6)

[see Eq. (3.73)] is the generator of linear translations:
J () |r = |r + .
(6.7)
Below we will see that one can define a unitary operator that generates rotations.
Exercise 6.1.1. Show that

x
x
y
Dz ()
Dz () = Rz y ,
z
z
(6.8)
where

iLz
Dz () = exp
,
(6.9)
and where
cos sin 0
Rz = sin cos 0 .
0
0 1
(6.10)
Dz () xDz () = x cos y sin ,
(6.11)
= x sin + y cos ,
(6.12)
=z.
(6.13)
Solution 6.1.1. Equation (6.8) is made of 3 identities:

Dz () yDz ()
Dz () zDz ()
As an example, we prove below the first one. Using the identity (2.173), which
is given by
eL AeL = A + [L, A] +
1
1
[L, [L, A]] + [L, [L, [L, A]]] + ,
2!
3!
(6.14)
one has
Dz () xDz ()
= x+
i
1
[Lz , x] +

2!
Eyal Buks
i
2
[Lz , [Lz , x]] +
1
3!
i
3
[Lz , [Lz , [Lz , x]]] + .
(6.15)
150
6.1. Angular Momentum and Rotation

Furthermore with the help of
Lz = xpy ypx ,
[xi , pj ] = i ij ,
(6.16)
(6.17)
one finds that

[Lz , x] = y [px , x] = iy ,
[Lz , [Lz , x]] = ix [py , y] = (i) x ,
[Lz , [Lz , [Lz , x]]] = (i)2 [Lz , x] = (i)3 y ,

4
[Lz , [Lz , [Lz , [Lz , x]]]] = (i) x ,

..
.
(6.18)
thus
Dz () xDz ()
2 4
= x 1
+
+
2!
4!
= x cos y sin .

3
y
+
3!
(6.19)
The other 2 identities in Eq. (6.8) can be proven in a similar way.

The matrix Rz [see Eq. (6.10)] represents a geometrical rotation around
the z axis with angle . Therefore, in view of the above result, we refer to the
operator Dz () as the generator of rotation around the z axis with angle .
It is straightforward to generalize the above results and to show that rotation
around an arbitrary unit vector n
axis with angle is given by

iL n
Dn () = exp
.
(6.20)

In view of Eq. (3.73), it can be said that linear momentum p generates
translations. Similarly, in view of the above equation (6.20), angular momentum L generates rotation. However, there is an important distinction between
these two types of geometrical transformations. On one hand, according to
Eq. (3.7) the observables px , py and pz commute with each other, and consequently translation operators with different translation vectors commute
[J (1 ) , J (2 )] = 0 .
(6.21)
On the other hand, as can be seen from Eq. (6.5), different components of L do
not commute and therefore rotation operators Dn () with different rotations
axes n
need not commute. Both the above results, which were obtained from
commutation relations between quantum operators, demonstrate two well
known geometrical facts: (i) different linear translations commute, whereas
(ii) generally, different rotations do not commute.
Eyal Buks
151
6.2 General Angular Momentum

Elementary particles carry angular momentum in two different forms. The
first one is the above discussed orbital angular momentum, which is commonly labeled as L. This contribution L = r p has a classical analogue,
which was employed above to derive the commutation relations (6.5) from
the corresponding Poissons brackets relations. The other form of angular
momentum is spin, which is commonly labeled as S. Contrary to the orbital
angular momentum, the spin does not have any classical analogue. In a general discussion on angular momentum in quantum mechanics the label J is
commonly employed.
L - orbital angular momentum
S - spin angular momentum
J - general angular momentum
In the discussion below we derive some properties of angular momentum
in quantum mechanics, where our only assumption is that the components of
the angular momentum vector of operators J = (Jx , Jy , Jz ) obey the following
commutation relations
[Ji , Jj ] = iijk Jk .
(6.22)
Namely, we assume that Eq. (6.5), which was obtained from the corresponding
Poissons brackets relations for the case of orbital angular momentum holds
for general angular momentum.
6.3 Simultaneous Diagonalization of J2 and Jz

As we have seen in chapter 2, commuting operators can be simultaneously
diagonalized. In this section we seek such simultaneous diagonalization of the
operators J2 and Jz , where
J2 = Jx2 + Jy2 + Jz2 .
(6.23)
As is shown by the claim below, these operators commute.

Claim. The following holds
, 2
- ,
- ,
J , Jx = J2 , Jy = J2 , Jz = 0 .
(6.24)
Proof. Using Eq. (6.22) one finds that

, 2 - , 2 - , 2 J , Jz = Jx , Jz + Jy , Jz
= i (Jx Jy Jy Jx + Jy Jx + Jx Jy ) = 0 .
,
(6.25)
In a similar way one can show that J2 , Jx = J2 , Jy = 0.

Eyal Buks
152
6.3. Simultaneous Diagonalization of J2 and Jz

The common eigenvectors of the operators J2 and Jz are labeled as |a, b,
and the corresponding eigenvalues are labeled as a2 and b respectively
J2 |a, b = a2 |a, b ,
Jz |a, b = b |a, b .
(6.26)
(6.27)
Recall that we have shown in chapter 5 for the case of harmonic oscillator
that the ket-vectors a |n and a |n are eigenvectors of the number operator
N provided that |n is an eigenvector of N. Somewhat similar claim can
be made regrading the current problem under consideration of simultaneous
diagonalization of J2 and Jz :
Theorem 6.3.1. Let |a, b be a normalized simultaneous eigenvector of the
operators J2 and Jz with eigenvalues 2 a and b respectively, i.e.
J2 |a, b = a2 |a, b ,
Jz |a, b = b |a, b ,
a, b |a, b = 1 .
(6.28)
(6.29)
(6.30)
Then (i) the vector

|a, b + 1 1 [a b (b + 1)]1/2 J+ |a, b
(6.31)
where
J+ = Jx + iJy ,
(6.32)
is a normalized simultaneous eigenvector of the operators J2 and Jz with

eigenvalues 2 a and (b + 1) respectively, namely
J2 |a, b + 1 = a2 |a, b + 1 ,
Jz |a, b + 1 = (b + 1) |a, b + 1 .
(6.33)
(6.34)
(ii) The vector

|a, b 1 1 [a b (b 1)]1/2 J |a, b
(6.35)
where
J = Jx iJy ,
(6.36)
is a normalized simultaneous eigenvector of the operators J2 and Jz with

eigenvalues 2 a and (b 1) respectively, namely
J2 |a, b 1 = a2 |a, b 1 ,
Jz |a, b 1 = (b 1) |a, b 1 .
Eyal Buks
(6.37)
(6.38)
153

Proof. The following holds
J2 (J |a, b) = J2 , J + J J2 |a, b = a2 (J |a, b) .

(6.39)
Similarly
Jz (J |a, b) = ([Jz , J ] + J Jz ) |a, b ,
(6.40)
where
[Jz , J ] = [Jz , Jx iJy ] = (iJy Jx ) = J ,
(6.41)
thus
Jz (J |a, b) = (b 1) (J |a, b) .
(6.42)
Using the following relations
J+ = J J+
J+
= (Jx iJy ) (Jx + iJy )
= Jx2 + Jy2 + i [Jx , Jy ]
= J2 Jz2 Jz ,
(6.43)
J
J = J+ J
= (Jx + iJy ) (Jx iJy )
= Jx2 + Jy2 + i [Jy , Jx ]
= J2 Jz2 + Jz ,
(6.44)
one finds that
a, b| J+
J+ |a, b = a, b| J2 |a, b a, b| Jz (Jz + ) |a, b
= 2 [a b (b + 1)] ,
(6.45)
and
a, b| J
J |a, b = a, b| J2 |a, b a, b| Jz (Jz ) |a, b
= 2 [a b (b 1)] .
(6.46)
Thus the states |a, b + 1 and |a, b 1 are both normalized.

Eyal Buks
154
6.3. Simultaneous Diagonalization of J2 and Jz

What are the possible values of b? Recall that we have shown in chapter 5
for the case of harmonic oscillator that the eigenvalues of the number operator
N must be nonnegative since the operator N is positive-definite. Below we
employ a similar approach to show that:
Claim. b2 a
Proof. Both Jx2 and Jy2 are positive-definite, therefore
a, b| Jx2 + Jy2 |a, b 0 .
(6.47)
On the other hand, Jx2 + Jy2 = J2 Jz2 , therefore a b2 0.

As we did in chapter 5 for the case of the possible eigenvalues n of the
number operator N , also in the present case the requirement b2 a restricts
the possible values that b can take:
Claim. For a given value of a the possible values of b are {bmax , bmax + 1, , bmax 1, bmax }
where a = bmax (bmax + 1). Moreover, the possible values of bmax are 0, 1/2, 1, 3/2, 2, .
Proof. There must be a maximum value bmax for which
J+ |a, bmax = 0 .
(6.48)
Thus, also
J+
J+ |a, bmax = 0
(6.49)
holds. With the help of Eq. (6.43) this can be written as

2

J Jz2 Jz |a, bmax = [a bmax (bmax + 1)] 2 |a, bmax = 0 . (6.50)
Since |a, bmax = 0 one has
a bmax (bmax + 1) = 0 ,
(6.51)
a = bmax (bmax + 1) .
(6.52)
or
In a similar way with the help of Eq. (6.44) one can show that there exists a
minimum value bmin for which
a = bmin (bmin 1) .
(6.53)
From the last two equations one finds that

bmax (bmax + 1) = bmin (bmin 1) ,
(6.54)
or
Eyal Buks
155

(bmax + bmin ) (bmax bmin + 1) = 0 .
(6.55)
Thus, since bmax bmin + 1 > 0 one finds that

bmin = bmax .
(6.56)
The formal solutions of Eqs. (6.52) and (6.53) are given by

1 1
bmax =
1 + 4a ,
2 2
and
1 1
bmin =
1 + 4a .
2 2
(6.57)
(6.58)
Furthermore, a is an eigenvalue of a positive-definite operator J2 , therefore

a 0. Consequently, the only possible solutions for which bmax bmin are
1 1
bmax = +
1 + 4a 0 ,
(6.59)
2 2
and
1 1
1 + 4a = bmax 0 .
(6.60)
bmin =
2 2
That is, for a given value of a, both bmax and bmin are uniquely determined. The value bmin is obtained by successively applying the operator J to the state |a, bmax an integer number of times, and therefore
bmax bmin = 2bmax must be an integer. Consequently, the possible values of
bmax are 0, 1/2, 1, 3/2, .
We now change the notation |a, b for the simultaneous eigenvectors to
the more common notation |j, m, where
j = bmax ,
m=b.
(6.61)
(6.62)
Our results can be summarized by the following relations

J2 |j, m = j (j + 1) 2 |j, m ,
Jz |j, m = m |j, m ,

J+ |j, m = j (j + 1) m (m + 1) |j, m + 1 ,

J |j, m = j (j + 1) m (m 1) |j, m 1 ,
(6.63)
(6.64)
(6.65)
(6.66)
where the possible values j can take are
1
3
j = 0, , 1, , ,
(6.67)
2
2
and for each given j, the quantum number m can take any of the 2j + 1
possible values
m = j, j + 1, , j 1, j .
Eyal Buks
(6.68)
156
6.5. Orbital Angular Momentum

The vector space of a spin 1/2 system is the subspace spanned by the ketvectors |j = 1/2, m = 1/2 and |j = 1/2, m = 1/2. In this subspace the
spin angular momentum is labeled using the letter S, as we have discussed
above. The matrix representation of some operators of interest in this basis
can be easily found with the help of Eqs. (6.63), (6.64), (6.65) and (6.66):

2
10
2 3
S =
,
(6.69)
01
4

1 0

z ,
Sz =
(6.70)
2 0 1
2

S+ =

S =

01
00
00
10

(6.71)
(6.72)
The above results for S+ and S together with the identities

S+ + S
,
2
S+ S
=
,
2i
used to find the matrix representation of Sx and Sy

01

=
x ,
2 10
2

0 i

=
y .
2 i 0
2
Sx =
(6.73)
Sy
(6.74)
can be
Sx
Sy
(6.75)
(6.76)
The matrices x , y and z are called Paulis matrices, and are related to
the corresponding spin angular momentum operators by the relation

k .
Sk =
2
(6.77)
6.5 Orbital Angular Momentum

As we have discussed above, orbital angular momentum L = r p refers to
spatial motion. For this case the states |l, m (here, the letter l is used instead
of j since we are dealing with orbital angular momentum) can be described
using wave functions. In this section we calculate these wave functions. For
this purpose it is convenient to employ the transformation from Cartesian to
spherical coordinates
Eyal Buks
157

x = r sin cos ,
y = r sin sin ,
z = r cos ,
(6.78)
(6.79)
(6.80)
where
r0,
0,
0 2 .
(6.81)
(6.82)
(6.83)
Exercise 6.5.1. Show that:

1.
r | Lz | = i
x | .
(6.84)
2.

r | L | = i exp (i) i
cot
r | .
(6.85)
3.
r | L | =
1 2
1
2
2 + sin
sin

sin
r | .
Solution 6.5.1. Using the relations
x
y

z
px py pz
r | r | = r r | ,

r | p | = r | ,
i
[see Eqs. (3.21) and (3.29)] one finds that

r | Lx | =
y
z
(r ) ,
i
z
y

r | Ly | =
z
x
(r ) ,
i
x
z

x
y
(r ) ,
r | Lz | =
i
y
x
(6.86)
(6.87)
(6.88)
(6.89)
(6.90)
(6.91)
(6.92)
where
(r ) = r | .
Eyal Buks
(6.93)
158


r = x2 + y 2 + z 2 ,
z
cos =
,
x2 + y 2 + z 2
x
cot = .
y
(6.94)
(6.95)
(6.96)
x
y
z
=
+
+
x y z
= r sin sin
+ r sin cos
x
y
= y
+x
,
x
y
(6.97)
thus using Eq. (6.92) one has
r | Lz | = i
(r ) .

(6.98)
2. Using Eqs. (6.90) and (6.91) together with the relation L+ = Lx + iLy
one has
i
i
r | L+ | = r | Lx + iLy |

z
+ iz
ix
(r )
= y
z
y
x
z

= z i
i (x + iy)
(r )
x y
z

= z i
ir sin ei
(r ) .
x y
z
(6.99)
Thus, by using the identity
x
y
z
=
+
+
x y z

= r cos cos
+ sin
r sin
,
x
y
z
(6.100)
or
r sin
Eyal Buks

= r cos cos
+ sin
,
z
x
y
(6.101)
159

one finds that

i
r | L+ | = z i
ie
cot x
+y
(r )

x y
x
y

i
i
i
z + ie y cot
+ ie
(r )
= i z e x cot
x
y
i

i tan + iy + i (r )
= ei i cot ze
tan x x cot ze

y
xiy
xiy

= ei cot y
x
+i
(r )
x
y

i
i
cot
(r ) .
=e
(6.102)
In a similar way one evaluates r | L |. Both results can be expressed
as

(6.103)
r | L | = i exp (i) i
cot
(r ) .
3. Using the result of the previous section one has

1
r | Lx | = r | (L+ + L ) |
2

i i
i
=
e
cot
i
+e
cot
+i
(r )
2

= i cos cot
+ sin
(r ) .
(6.104)
Similarly

r | Ly | = i sin cot
cos
(r ) ,
(6.105)
thus
r | L2 | = r | L2x + L2y + L2z |

2
2
2
2
=
cos cot
+ sin
+ sin cot
cos
+ 2 (r )

2
2
= 2 1 + cot2
(r )
2 + cot +
2

1 2
1
+
= 2
sin
(r ) .
sin2 2 sin
(6.106)
Eyal Buks
160

Spherical Harmonics. The above exercise allows translating the relations
(6.63) and (6.64), which are given by
L2 |l, m = l (l + 1) 2 |l, m ,
Lz |l, m = m |l, m ,
(6.107)
(6.108)
into differential equations for the corresponding wavefunctions

1
1 2
sin
(r ) = l (l + 1) (r ) , (6.109)
sin2 2 sin
i (r ) = m (r ) ,
(6.110)
where
m = l, l + 1, , l 1, l .
(6.111)
We seek solutions having the form

(r ) = f (r) Ylm (, ) .
(6.112)
We require that both f (r) and Ylm (, ) are normalized

1=
1=

0
drr2 |f (r)|2 ,
d sin
2
0
(6.113)
d |Ylm (, )|2 .
(6.114)
These normalization requirements guarantee that the total wavefunction is

normalized
1=
dx
dy
dz | (r )| .
Substituting into Eqs. (6.109) and (6.110) yields

1 2
1 sin
+
Ylm = l (l + 1) Ylm ,
sin2 2 sin
i
m
Y = mYlm .
l
(6.115)
(6.116)
(6.117)
The functions Ylm (, ) are called spherical harmonics

In the previous section, which discusses the case of general angular momentum, we have seen that the quantum number m can take any half integer
value 0, 1/2, 1, 3/2, [see Eq. (6.67)]. Recall that the only assumption employed in order to obtain this result was the commutation relations (6.22).
Eyal Buks
161

However, as is shown by the claim below, only integer values are allowed for
the case of orbital angular momentum. In view of this result, one may argue
that the existence of spin, which corresponds to half integer values, is in fact
predicted by the commutation relations (6.22) only.
Claim. The variable m must be an integer.
Proof. Consider a solution having the form
Ylm (, ) = Flm () eim .
(6.118)
Clearly, Eq. (6.117) is satisfied. The requirement

Ylm (, ) = Ylm (, + 2) ,
(6.119)
namely, the requirement that Ylm (, ) is continuos, leads to

e2im = 1 ,
(6.120)
thus m must be an integer.

The spherical harmonics Ylm (, ) can be obtained by solving Eqs. (6.116)
and (6.117). However, we will employ an alternative approach, in which in
the first step we find the spherical harmonics Yll (, ) by solving the equation
L+ |l, l = 0 ,
(6.121)
which is of first order [contrary to Eq. (6.116), which is of the second order].
Using the identity (6.85), which is given by

i
r | L+ | = ie
(6.122)
i
cot
r | ,
one has

l cot Fll () = 0 .
(6.123)

Fll () = Cl (sin )l ,
(6.124)
where Cl is a normalization constant. Thus, Yll is given by

Yll (, ) = Cl (sin )l eil .
(6.125)
In the second step we employ the identity (6.66), which is given by

J |j, m = j (j + 1) m (m 1) |j, m 1 ,
(6.126)
Eyal Buks
162
6.6. Problems
and Eq. (6.85), which is given by

r | L | = i exp (i) i
cot
r | ,
(6.127)
to derive the following recursive relation

ei
m cot Ylm (, ) = l (l + 1) m (m 1)Ylm1 (, ) ,
(6.128)
which allows finding Ylm (, ) for all possible values of m provided that
Yll (, ) is given. The normalized spherical harmonics are found using this
method to be given by
(1)l
Ylm (, ) = l
2 l!
2l + 1 (l + m)! im
dlm
e
(sin )m
(sin )2l .
lm
4 (l m)!
d (cos )
(6.129)
As an example, closed form expressions for the cases l = 0 and l = 1 are

given below
1
Y00 (, ) =
,
4
0
3
1
Y1 (, ) =
sin ei ,
8
0
3
0
Y1 (, ) =
cos .
4
(6.130)
(6.131)
(6.132)
6.6 Problems
1. Let R (where i {x, y, z} ) be the 33 rotation matrices (as defined in
the lecture). Show that for infinitesimal angle the following holds

[Rx () , Ry ()] = 1 Rz 2 ,
(6.133)
where
[Rx () , Ry ()] = Rx () Ry () Ry () Rx () .
2. Show that

iJz
iJz
exp
Jx exp
= Jx cos Jy sin .

Eyal Buks
(6.134)
(6.135)
163

3. The components of the Pauli matrix vector are given by:

01
0 i
1 0
x =
, y =
, z =
.
10
i 0
0 1
(6.136)
a) Show that
( a) ( b) = a b + i (a b) ,
(6.137)
where a and b are vector operators which commute with , but not
necessarily commute with each other.
b) Show that

i n
exp
= 1 cos i n
sin ,
(6.138)
2
2
2
where n
is a unit vector and where 1 is the 2 2 identity matrix.
4. Find the eigenvectors and eigenvalues of the matrix n
(
n is a unit
vector).
5. Consider an electron in a state in which the component of its spin along
the z axis is +/2 . What is the probability that the component of the spin
along an axis z , which makes an angle with the z axis, will measured
to be +/2 or /2 . What is the average value of the component of the
spin along this axis?
6. The 2 2 matrix U is given by
U=
1 + i ( n
)
,
1 i ( n
)
(6.139)
where
= xx
+ y y
+ z
z
(6.140)
is the Pauli vector matrix,

+ ny y
+ nz
z
n
= nx x
(6.141)
is a unit vector, i.e. n

n
= 1, and nx , ny , nz and are all real parameters.
Note that generally for a matrix or an operator A1 A1 .
a) show that U is unitary.
b) Show that
dU
2i ( n
)
=
U.
d
1 + 2
(6.142)
c) Calculate U by solving the differential equation in the previous section.

Eyal Buks
164
6.6. Problems
7. A particle is located in a box, which is divided into a left and right
sections. The corresponding vector states are denoted as |L and |R
respectively. The Hamiltonian of the system is given by
H = EL |L L| + ER |R R| + (|L R| + |R L|) .
(6.143)
The particle at time t = 0 is in the left section

| (t = 0) = |L .
(6.144)
Calculate the probability pR (t) to find the particle in the state |R at
time t.
8. A magnetic field given by
z + B1 (
x cos (t) + y
sin (t))
B (t) = B0
(6.145)
is applied to a spin 1/2 particle. At time t = 0 the state is given by

| (t = 0) = |+;
z .
(6.146)
Calculate the probability P+ (t) to find the system in the state |;

z
at time t > 0.
9. A magnetic field given by
z + g (t) B1 (
x cos (t) + y
sin (t))
B (t) = B0
(6.147)
is applied to a spin 1/2 particle. While B0 , B1 and 1 are taken to be

constants, the function g (t) is assumed to be given by
0
t<0
0 t < p
1
p t < p + 0
g (t) = 0
,
(6.148)
1 p + 0 t < 2 p + 0
0
2 p + 0 t
i.e. two oscillatory magnetic field pulses, both having duration of p , are
applied, and the dwell time between these pulses is 0 . The normalized
pulse duration p is defined to be
p = 1 p ,
(6.149)
the normalized dwell time 0 is defined to be

0 = 0 ,
(6.150)
and the normalized detuning is defined to be

=
Eyal Buks
,
1
(6.151)
165

where
= 0 ,
and where
|e| B0
0 =
,
me c
|e| B1
1 =
.
me c
At time t = 0 the state is assumed to be given by
| (t = 0) = |+;
z .
(6.152)
(6.153)
(6.154)
Calculate the probability P++ (t) to find the system in the state |+;
z
at time t > 2 p + 0 . Assume that the normalized detuning is small, i.e.
|| 1, and expand P++ (t) to lowest nonvanishing order in for the
case where the normalized pulse duration is taken to be given by
p =
.
2
(6.155)
10. Find the time evolution of the state vector of a spin 1/2 particle in
a magnetic field along the z direction with time dependent magnitude
B (t) = B (t)
z.
11. A magnetic field given by B = B cos (t) z, where B is a constant, is
applied to a spin 1/2. At time t = 0 the spin is in state | (t), which
satisfies
Sx | (t = 0) =

| (t = 0) ,
2
(6.156)
Calculate the expectation value Sz at time t 0.

12. Consider a spin 1/2 particle. The time dependent Hamiltonian is given
by
H=
4Sz
,
1 + (t)2
(6.157)
where is a real non-negative constant and Sz is the z component of the

angular momentum operator. Calculate the time evolution operator u of
the system.
13. Consider a spin 1/2 particle in an eigenstate of the operator S n
with
eigenvalue +/2 , where S is the vector operator of angular momentum
and where n
is a unit vector. The angle between the unit vector n
and
the z axis is . Calculate the expectation values
a) S
(
' z
b) (Sz )2
Eyal Buks
166
6.6. Problems
14. An ensemble of spin 1/2 particles are in a normalized state
| = |+ + | ,
where the states |+ and | are the eigenstates of Sz (the z component
of the angular momentum operator). At what direction the magnetic field
should be aligned in a Stern-Gerlach experiment in order for the beam
not to split.
15. Consider a spin 1/2 particle having gyromagnetic ratio in a magnetic
field given by B (t) u
. The unit vector is given by
u
= (sin cos , sin sin , cos ) ,
(6.158)
where , are angles in spherical coordinates. The field intensity is given

by
t<0
0
B (t) = B0 0 < t < .
(6.159)
0
t>
At times t < 0 the spin was in state |+ , namely in eigenstate of Sz with
positive eigenvalue. Calculate the probability P (t) to find the spin in
state | at time t , where t > .
16. Consider a spin 1/2 particle. The Hamiltonian is given by
H = Sx ,
(6.160)
where is a Larmor frequency and where Sx is the x component of the

angular momentum operator. The z component of the angular momentum is measured at the times tn = nT /N where n = 0, 1, 2, , N , N is
integer and T is the time of the last measurement.
a) Find the matrix representation of the time evolution operator u (t)
in the basis of |;
z states.
b) What is the probability psame to get the same result in all N + 1
measurements. Note that the initial state of the particle is unknown.
c) For a fixed T calculate the limit lim psame .
N
17. Consider a spin 1/2 particle. No external magnetic field is applied. Three
measurements are done one after the other. In the first one the z component of the angular momentum is measured, in the second one the
component along the direction u
is measured and in the third measurement, again the z component is measured. The unit vector u
is described
using the angles and
u
= (sin cos , sin sin , cos ) .
(6.161)
Calculate the probability psame to have the same result in the 1st and
3rd measurements.
Eyal Buks
167

18. Let (t) be the expectation value of the magnetic moment associated
with spin 1/2 particle ( = S , where S is the angular momentum and
is the gyromagnetic ratio). Show that in the presence of a time varying
magnetic field B (t) the following holds
d
(t) = (t) B (t) .
dt
(6.162)
19. The Hamiltonian of an electron of mass m, charge q, spin 1/2, placed in

electromagnetic field described by the vector potential A (r, t) and the
scalar potential (r, t), can be written as [see Eq. (1.62)]
2

p qc A
q
+ q
B,
H=
2m
2mc
(6.163)
where B =A. Show that this Hamiltonian can also be written as

q 42
1 3
H=
p A
+ q .
(6.164)
2m
c
20. Show that

3
4

j, m| (Jx )2 + (Jy )2 |j, m = 2 j 2 + j m2 .
(6.165)
21. Find the condition under which the Hamiltonian of a charged particle in
a magnetic field
H=
1
q 2
p A .
2m
c
(6.166)
q
q2
1 2
p
pA+
A2 .
2m
mc
2mc2
(6.167)
can be written as
H=
22. Consider a point particle having mass m and charge q moving under the
influence of electric field E and magnetic field B, which are related to
the scalar potential and to the vector potential A by
E =
1 A
,
c t
(6.168)
and
B=A.
(6.169)
Find the coordinates representation of the time-independent Schrdinger

equation H | = E |.
Eyal Buks
168
6.6. Problems
23. A particle of mass m and charge e interacts with a vector potential
Ax = Az = 0 ,
Ay = Bx .
(6.170)
(6.171)
Calculate the ground state energy. Clue: Consider a wave function of the
form
(x, y, z) = (x) exp (iky y) exp (ikz z) .
(6.172)
24. Find the energy spectrum of a charged particle having mass m and charge
q moving in uniform and time-independent magnetic field B = B
z and
electric field E = E
x.
25. Consider a particle having mass m and charge e moving in xy plane under
the influence of the potential U (y) = 12 m 20 y 2 . A uniform and timeindependent magnetic field given by B = B
z is applied perpendicularly
to the xy plane. Calculate the eigenenergies of the particle.
26. Consider a particle with charge q and mass confined to move on a circle
of radius a in the xy plane, but is otherwise free. A uniform and time
independent magnetic field B is applied in the z direction.
a) Find the eigenenergies.
b) Calculate the current Jm for each of the eigenstates of the system.
27. The Hamiltonian of a non isotropic rigid rotator is given by
H=
L2y
L2x
L2
+
+ z ,
2Ixy
2Ixy
2Iz
(6.173)
where L is the vector angular momentum operator. At time t = 0 the

state of the system is described by the wavefunction
(, ) = A sin cos ,
(6.174)
where , are angles in spherical coordinates and A is a normalization

constant. Calculate the expectation value Lz at time t > 0 .
28. The eigenstates of the angular momentum operators L2 and Lz with
l = 1 and m = 1, 0, 1 are denoted as |1, 1, |1, 0 and |1, 1.
a) Write the 3 3 matrix of the operator Lx in this l,= 1 subspace.
b) Calculate the expectation value Lx for the state 12 |1, 1 + 2 |1, 0 + |1, 1 .
c) The same as the previous section for the state 12 [|1, 1 |1, 1].
d) Write the 3 3 matrix representation in this basis of the rotation
operator at angle around the z axis.
e) The same as in the previous section for an infinitesimal rotation with
angle d around the x axis.
29. Consider a particle of mass m in a 3D harmonic potential

1
V (x, y, z, ) = m 2 x2 + y 2 + z 2 .
2
Eyal Buks
(6.175)
169

The state vector | of the particle satisfy
ax | = x | ,
ay | = y | ,
az | = z | ,
(6.176)
(6.177)
(6.178)
where x , y and z are complex and ax , ay and az are annihilation

operators
0

m
ipx
ax =
x+
,
(6.179)
2
m
0

m
ipy
ay =
y+
,
(6.180)
2
m
0

m
ipz
az =
z+
,
(6.181)
2
m
Let L be the vector operator of the orbital angular momentum.

a) Calculate Lz .
b) Calculate Lz .
30. A rigid rotator is prepared in a state
| = A (|1, 1 |1, 1) ,
(6.182)
where A is a normalization constant, and where the symbol |l, m denotes

an angular momentum state with quantum numbers l and m. Calculate
a) '
Lx .
(
b) (Lx )2 .
31. The Hamiltonian of a top is given by

H=
L2x + L2y
L2
+ z ,
2I1
2I2
(6.183)
where L is the angular momentum vector operator. Let |0 be the

ground state of the system.
a) Calculate the quantity Az (), which is defined as

iLz
iLz
Az () = 0 | exp
(6.184)
H exp
|0 .

b) Calculate the quantity Ax (), which is defined as

iLx
iLx
H exp
|0 .
Ax () = 0 | exp

(6.185)
32. The wavefunction of a point particle is given by

(r) = (x + y + 2z) f (r) ,
(6.186)

where f (r) is a function of the radial coordinate r = x2 + y 2 + z 2 .
Eyal Buks
170
6.6. Problems
a) In a measurement of L2 what are the possible outcomes and the
corresponding probabilities.
b) The same for a measurement of Lz .
33. Consider a system comprising of two spin 1/2 particles.
a) Show that
, 2
(6.187)
S , Sz = 0 ,
where S = S1 + S2 , Sz = S1z + S2z and where S1 and S2 are the angular momentum vector operators of the first and second spin repetitively, i.e. S1 = (S1x , S1y , S1z ) and S2 = (S2x , S2y , S2z ).
b) Find an orthonormal basis of common eigenvectors of S2 and Sz
[recall that the existence
, of such
- a basis is guaranteed by the result
of the previous section S2 , Sz = 0, see Eqs. (2.147) and (2.148)].
34. Consider a system in a common eigenvector |j, m of the angular momentum operators J2 and Jz . A measurement of the operator Jn = n
J is
being performed, where n
= (cos sin , sin sin , cos )
is
a
unit
'
( vector.
2
Calculate the expectation value Jn and the variance (Jn ) .
35. Consider a harmonic oscillator having angular resonance frequency and
mass m. The operator S (, ) is defined by [compare with Eq. (5.83)]
i 2

e a ei a2
S (, ) = exp
,
(6.188)
2
where both and are real and
0

m
ip
a=
x+
2
m
(6.189)
is the annihilation operator [see Eq. (5.9)]. Show that S (, ) can be

factorized according to
i

e 2
S (, ) = exp
a tanh
2

log (cosh )
exp
aa + a a
2
i

e
2
exp
a tanh .
2
(6.190)
36. Show that the operator S (, ) (6.188) satisfies

S (, 0) = Q e ,
(6.191)
where Q, which is called the squeezing operator, is given by
Eyal Buks
171

Q () =
dx
|x / x | ,
(6.192)
where |x is an eigenvector of the position operator x having eigenvalue

x , i.e. x |x = x |x .
6.7 Solutions
1. By cyclic permutation of
cos sin 0
Rz = sin cos 0 ,
0
0
1
one has
(6.193)
1 0
0
Rx = 0 cos sin ,
0 sin cos
cos 0 sin
0 .
Ry = 0 1
sin 0 cos
(6.194)
(6.195)
On one hand
1 [Rx () , Ry ()]
1
1 + cos2 sin sin cos
1 cos2
1
sin cos sin
=
sin sin cos sin cos sin
1
2
1 0

= 2 1 0 + O 3 .
0 0 1
On the other hand
cos 2 sin 2
2
Rz = sin 2 cos 2
0
0
thus
0
1 2 0

0 = 2 1 0 + O 3 , (6.197)
1
0 0 1

1 [Rx () , Ry ()] = Rz 2 + O 3 .
Eyal Buks
(6.196)
(6.198)
172
6.7. Solutions
2. Using the identity (2.173), which is given by
eL AeL = A + [L, A] +
1
1
[L, [L, A]] + [L, [L, [L, A]]] + , (6.199)
2!
3!
and the commutation relations (6.22), which are given by

[Ji , Jj ] = iijk Jk ,
one has

iJz
Jx exp
exp

2
i
1 i
= Jx +
[Jz , Jx ] +
[Jz , [Jz , Jx ]]

2!
3
1 i
+
[Jz , [Jz , [Jz , Jx ]]] +
3!

1 2
1 3
= Jx 1 + Jy +
2!
3!
Jx cos Jy sin .
iJz
(6.200)
3. The components of the Pauli matrix vector are given by:

01
0 i
1 0
x =
, y =
, z =
.
10
i 0
0 1
(6.201)
(6.202)
a) The following
holds

ax iay
az
a=
,
(6.203)
ax + iay az

bz
bx iby
b =
,
(6.204)
bx + iby bz
thus

az bz + (ax iay ) (bx + iby ) az (bx iby ) (ax iay ) bz
( a) ( b) =
(ax + iay ) bz az (bx + iby ) az bz + (ax + iay ) (bx iby )

10
= ab
01

01
+i (ay bz az by )
10

0 i
+i (az bx ax bz )
i 0

1 0
+i (ax by ay bx )
0 1
= a b + i (a b) .
Eyal Buks
(6.205)
173

b) Using (a) one has
( n
)2 = 1 ,
(6.206)
thus with the help of the Taylor expansion of the functions cos (x)
and sin (x)
one finds

i n
exp
= cos
i sin
2
2
2
= 1 cos i n
sin .
2
2
(6.207)
4. In spherical coordinates the unit vectors n
is expressed as
n
= (cos sin , sin sin , cos ) ,
(6.208)
thus
n
=
cos sin ei
sin ei cos
(6.209)
The eigen values + and are found solving

+ + = Tr ( n
) = 0 ,
(6.210)
+ = Det ( n
) = 1 ,
(6.211)
and
thus
= 1 .
(6.212)
The normalized eigenvectors can be chosen to be given by

cos 2 exp i
2

|+ =
,
sin 2 exp i
2

sin 2 exp i
2

| =
.
cos 2 exp i
2
(6.213)
(6.214)
5. Using Eq. (6.213) one finds the probability p+ to measure +/2 is given
by

cos exp i
2
2
2
= cos2 ,

(6.215)
p+ =
1 0
sin 2 exp i
2
2
and the probability p to measure /2 is
p = 1 p+ = sin2
Eyal Buks
.
2
(6.216)
174
6.7. Solutions
The average value of

cos2 sin2
2
2
the component of the spin along z axis is thus

= cos .
(6.217)
2
2
6. In general, note that all smooth functions of the matrix ( n

) commute,
a fact that greatly simplifies the calculations.
a) The following holds
1
= 1 + i ( n
) + [i ( n
)]2 + ,
1 i ( n
)
thus
1
1 i ( n
)
(6.218)
= 1 i ( n
) + [(i) ( n
)]2 +
=
1
,
1 + i ( n
)
(6.219)
therefore
1 + i ( n
) 1 i ( n
)
=1,
1 i ( n
) 1 + i ( n
)
UU =
(6.220)
and similarly U U = 1.
b) Exploiting again the fact that all smooth functions of the matrix
( n
) commute and using Eq. (6.206) one has
[1 i ( n
)] ( n
) + [1 + i ( n
)] ( n
)
dU
=i
2
d
[1 i ( n
)]
2 ( n
)
[1 i ( n
)]2
2 ( n
)
1 + i ( n
)
=i
[1 i ( n
)] [1 + i ( n
)] 1 i ( n
)
2i ( n
)
=
U.
1 + 2
=i
(6.221)
c) By integration one
has

d
U = U0 exp 2i ( n
)
2
0 1+

1
U0 exp 2i ( n
) tan ,
(6.222)
where U0 is a the matrix U at = 0. With the help of Eq. (6.138)
one thus finds that
,

U = U0 1 cos 2 tan1 + i n
sin 2 tan1 ,
(6.223)
Eyal Buks
175

Using the identities

1 2
cos 2 tan1 =
,
1 + 2

2
sin 2 tan1 =
,
1 + 2
and assuming U0 = 1 one finds that
U=
(6.224)
(6.225)
2
1 2
+ i n
.
1 + 2
1 + 2
(6.226)
7. In terms of Pauli matrices

H=E
a 0 + x + Ed z ,
(6.227)
where
Ea =
EL + ER
EL ER
, Ed =
,
2
2
0 =
(6.228)
and
10
01
, x =
01
10
, z =
1 0
0 1
Using Eq. (6.138), which is given by

i n
exp
= cos i n
sin ,
2
2
2
(6.229)
(6.230)
the time evolution operator u (t) can be calculated

iHt
u (t) = exp

iEa 0 t
i ( x + Ed z ) t
= exp
exp

!
"

i n
2 + Ed2 t
iEa t
= exp
exp
,

(6.231)
where
(, 0, Ed )
n
=
,
2 + Ed2
(6.232)
thus
"

!
2 + Ed2 t
t 2 + Ed2
iEa t
x + Ed z
u (t) = exp
cos
i
sin
.

2 + Ed2
Eyal Buks
176
6.7. Solutions
(6.233)
The probability pR (t) is thus given by
2
pR (t) = |R| u (t) | (t = 0)|

= |R| u (t) |L|2
2
2
EL ER 2 sin
2
.

2
R
t 2 + EL E
2
.
(6.234)

H = 0 Sz + 1 (cos (t) Sx + sin (t) Sy ) ,
(6.235)
where
|e| B0
,
me c
|e| B1
1 =
.
me c
0 =
(6.236)
(6.237)
The matrix representation in the basis {|+ , |} (where |+ = |+;
z
and | = |;
z) is found using Eqs. (6.70), (6.75) and (6.76)

1 exp (it)
0
.
(6.238)
H=
0
2 1 exp (it)
The Schrdinger equation is given by
i
d
| = H | .
dt
It is convenient to express the general solution as

it
it
| (t) = b+ (t) exp
|+ + b (t) exp
| .
2
2
Substituting into the Schrdinger equation yields
it

d
e 2 0
b+
i
it
b
dt
0 e 2

it

it
1
0 1e
e 2 0
b+
=
,
it
b
2 1 eit 0
0 e 2
(6.239)
(6.240)
(6.241)
or
Eyal Buks
177

i
2
i
=
2
e
0
it
2
0
1 eit

it

b+
e 2 0
b +
+
it
it
b
b
e 2
0 e 2

it
e 2 0
b+
1 eit
.
it
0
2
b
0 e
(6.242)
By multiplying from the left by

it

e 2
0
it
0 e 2
one has

i 1 0
i 0 1
b+
b +
b+
+
=
,
0
1
b
b
b
2
2
1
0
(6.243)
or
i
b +
b
=
2
b+
b
(6.244)
where
=
1
1
= z + 1 x ,
(6.245)
and
= 0 .
(6.246)
At time t = 0

b+ (0)
1
=
.
b (0)
0
(6.247)
The time evolution is found using Eq. (6.138)

it
b+ (t)
b+ (0)
= exp
b (t)
b (0)
2
sin
cos i
i 21 sin 2
21 +()2
1 +()
=
sin
i 21 sin 2
cos + i
2
1 +()
1 +()2

1
,
0
(6.248)
where
=
Eyal Buks
.
21 + ()2 t
2
(6.249)
178
6.7. Solutions
The probability is thus given by
.
2
21 + ()2 t
1
2
P+ (t) = 2
.
2 sin
2
1 + ()
9. The transformation into the rotating frame reads

it
it
| (t) = b+ (t) exp
|+ + b (t) exp
| .
2
2
(6.250)
(6.251)
For time periods where g (t) is constant the time evolution is governed
by Eq. (6.248). Thus at time t = 2 p + 0 one has

1
b+ (2 p + 0 )
,
(6.252)
= Mp M0 Mp
b (2 p + 0 )
0
where
Mp =
M0 =
sin 2p
cos p i
1+
i sin p
1+
i sin p
cos p +
1+ 2
1+
ei 2 0
0
0 ei 2
i sin p
and where

1 + 2 p
p =
.
2
(6.253)
(6.254)
(6.255)
Thus the probability P++ (2 p + 0 ) is given by

P++ (2 p + 0 ) = |b+ (2 p + 0 )|2
2
"2
2
i0
i
sin
e
sin
p
p
=
cos p
1 + 2
1 + 2

1 ei0 sin2 p
i sin (2 p )
=
cos (2p )
+
1 + 2
1 + 2
(6.256)
For the case where p = /2 one has to lowest nonvanishing order in

P++ (2 p + 0 ) =

1 cos 0
+ sin 0 + O 2 .
2
(6.257)
Note that the probability P++ (t) is unchanged for t > 2 p + 0 .

Eyal Buks
179

10. The Schrdinger equation is given by
i
d |
= H | ,
dt
(6.258)
where
H = Sz ,
(6.259)
and
(t) =
|e| B (t)
.
me c
(6.260)
In the basis of the eigenvectors of Sz one has

| = c+ |+ + c | ,
(6.261)
i (c+ |+ + c |) = Sz (c+ |+ + c |) ,
(6.262)
and
where

Sz | = | ,
2
(6.263)
thus one gets 2 decoupled equations

i
c+ ,
2
i
c =
c .
2

i t
c (t) = c (0) exp

(t ) dt
2 0

t
i |e|
= c (0) exp
B (t ) dt .
2me c 0
c+ =
(6.264)
(6.265)
(6.266)
11. At time t = 0
1
| (t = 0) = (|+ + |) .
2
(6.267)
Using the result of the previous problem and the notation

c =
Eyal Buks
eB
,
mc
(6.268)
180
6.7. Solutions
one finds
t
t
1
i
i
c
c
| (t) = exp
cos (t ) dt |+ + exp
cos (t ) dt |
2
2
2
0
0

i c sin t
ic sin t
1
|+ + exp
| ,
= exp
2
2
2
(6.269)
thus
Sz (t) = (t)| Sz | (t) = 0 .
(6.270)
12. The Schrdinger equation for u is given by

i
du
= Hu ,
dt
(6.271)
thus
du
4iSz
1
u.
=
dt
1 + (t)2
By integration one finds
t
4iS
dt
z
u (t) = u (0) exp

1 + (t )2
0

4iSz
= u (0) exp
tan1 (t) .

Setting an initial condition u (t = 0) = 1 yields

4iSz
1
u (t) = exp
tan (t) .
(6.272)
(6.273)
(6.274)
The matrix elements of u (t) in the basis of the eigenstates | of Sz are

given by

1 + it
+| u (t) |+ = exp 2i tan1 (t) =
,
(6.275)
1 it

1 it
| u (t) | = exp 2i tan1 (t) =
,
(6.276)
1 + it
+| u (t) | = | u (t) |+ = 0 .
(6.277)
13. The eigenvector of S n
, where n
= (sin cos , sin sin , cos ) with
eigenvalue +/2 is given by [see Eq. (6.213)]
Eyal Buks
181

|+; S n
= cos ei 2 |+ + sin ei 2 | .
2
2
(6.278)
The operator Sz is written as

Sz =

(|+ +| | |) .
2
(6.279)
a) Thus

+; S n
| Sz |+; S n
=
2

2
2
cos
sin
= cos . (6.280)
2
2
2
b) Since Sz2 is the identity operator times 2 /4 one has

'
( ) *
2
2
1 cos2 =
sin2 . (6.281)
(Sz )2 = Sz2 Sz 2 =
4
4
14. We seek a unit vector n

such that
| = |+; S n
,
(6.282)
where |+; S n
is given by Eq. (6.213)

i+
i+
+
+
exp
|++sin
exp
| , (6.283)
|+; S n
= cos
2
2
2
2
thus the following hold
ctg
=

,
2
(6.284)
and
+ = arg () arg () .
(6.285)
Similarly, by requiring that

| = |; S n
,
(6.286)
where
|; S n
= sin

i
i
exp |++cos
exp
| , (6.287)
2
2
2
2
one finds
tan
=
,
2
= arg () arg () + .
Eyal Buks
(6.288)
(6.289)
182
6.7. Solutions
15. The Hamiltonian at the time interval 0 < t < is given by
) ,
H = B0 (S u
(6.290)
where is the gyromagnetic ratio and S is the angular momentum operator. The eigenvectors of S u
with eigenvalue /2 are given by [see
Eqs. (6.213) and (6.214)]

|+; S u
= cos ei 2 |+ + sin ei 2 | ,
2
2
(6.291)

|; S u
= sin ei 2 |+ + cos ei 2 | ,
(6.292)
2
2
Thus in the time interval 0 < t < the state vector is given by

iB0 t
iB0 t
| = |+; S u
+; S u
|+ exp
+ |; S u
; S u
|+ exp
2
2

iB0 t

iB0 t
= |+; S u
cos ei 2 exp
|; S u
sin ei 2 exp
2
2
2
2

iB0 t
iB0 t
= ei cos2 exp
+ sin2 exp
|+
2
2
2
2

iB0 t
iB0 t
+ sin cos
exp
exp
|
2
2
2
2

1 + cos
iB0 t
1 cos
iB0 t
= ei
exp
+
exp
|+
2
2
2
2

B0 t
+i sin sin
|
2

B0 t
B0 t
B0 t
= ei cos
+ i cos sin
|+ + i sin sin
| .
2
2
2
(6.293)
Thus for t >

2
P (t) = sin sin
B0
2
(6.294)
An alternative solution - The Hamiltonian in the basis of | states is

given by
H=
B0
( u
) ,
2
where is the Pauli matrix vector

01
0 i
1 0
1 =
, 2 =
, 3 =
.
10
i 0
0 1
Eyal Buks
(6.295)
(6.296)
183

The time evolution operator is given by

iHt
iB0 t
u (t) = exp
= exp
( u
) .

2
(6.297)
Using the identity (6.138) one finds

B0 t
B0 t
+ i
u sin
u (t) = I cos
2
2

cos B20 t + i cos sin B20 t

i sin ei sin B20 t

,
=
i sin ei sin B20 t
cos B20 t i cos sin B20 t
(6.298)
thus for t >

2

B0
1
2
2
P (t) =
0 1 u (t)
=
sin
sin
.
0
(6.299)
16. The matrix representation of the Hamiltonian in the basis of |; Sz

states is given by
H=

(
x ) ,
2
where is the Pauli matrix vector

01
0 i
1 0
x =
, y =
, z =
.
10
i 0
0 1
a) The time evolution operator is given by

iHt
it
u (t) = exp
=
exp
(
x ) .

2
(6.300)
(6.301)
(6.302)
Using the identity

exp (iu ) = 1 cos + i
u sin ,
(6.303)
where u =
u is a three-dimensional real vector and u
is a threedimensional real unit vector, one finds
t
t
u (t) =
1 cos
i1 sin
2
2

t
cos t
i
sin
2
2
=
.
i sin t
cos t
2
2
(6.304)
Eyal Buks
184
6.7. Solutions
b) Let P++ (t) be the probability to measure Sz = +/2 at time t > 0
given that at time t = 0 the spin was found to have Sz = +/2.
Similarly, P (t) is the probability to measure Sz = /2 at time
t > 0 given that at time t = 0 the spin was found to have Sz = /2.
These probabilities
are given by

2

t
1
P++ (t) =
1 0 u (t)
,
(6.305)
= cos2
0

2

t
0
P (t) =
0 1 u (t)
.
(6.306)
= cos2
1
2
Thus, assuming that the first measurement has yielded Sz = +/2
T -N
,
one finds psame = P++ N
, whereas assuming that the first mea,
T -N
.
surement has yielded Sz = /2 one finds psame = P N
Thus in general independently on the result of the first measurement
one has
psame = cos2N
c) Using
psame
T
.
2N
(6.307)

T
= exp 2N log cos
2N
!
!
4 ""

2
1 T
1
= exp 2N log 1
+O
2 2N
N
!
3 "
1
(T )2
= exp
+O
,
4N
N
(6.308)
one finds
lim psame = 1 .
(6.309)
This somewhat surprising result is called the quantum Zeno effect or

the watched pot never boils effect.
17. The eigenvectors of S u
with eigenvalues /2 are given by
(6.310a)
|+; u
= cos ei 2 |+ + sin ei 2 | ,
2
2

|; u
= sin ei 2 |+ + cos ei 2 | ,
(6.310b)
2
2
where the states | are eigenvectors of S
z. Let P (3 , 2 |1 ) be the
probability to measure S u
= 2 (/2) in the second measurement and to
measure S
z =3 (/2) in the third measurement given that the result of
the first measurement was S
z =1 (/2), and where n {+, }. The
following holds
Eyal Buks
185
,
(6.311a)
2
2
P (+, |+) = |+|; u
|2 |+|; u
(6.311b)
| = sin 4 ,
2
2
2
P (, |) = ||; u
| ||; u
| = cos4 ,
(6.311c)
2
2
2
P (, +|) = ||+; u
| ||+; u
| = sin 4 ,
(6.311d)
2
thus independently on what was the result of the first measurement one
has
2
P (+, +|+) = |+|+; u

| |+|+; u
| = cos4
1
+ sin 4 = 1 sin2 .
2
2
2
psame = cos4
(6.312)
H = B .
(6.313)
Using Eq. (4.38) for z one has

d z
1
=
[z , H]
dt
i
2
= Bx [Sz , Sx ] + By [Sz , Sy ]
i
2
= By Sx Bx Sy
= ( B)
z.
(6.314)
Similar expressions are obtained for x and y that together can be

written in a vector form as
d
(t) = (t) B (t) .
(6.315)
dt
19. Using Eq. (6.137), which is given by
( a) ( b) = a b + i (a b) ,
one has
3
q 42
p A
= p
c

= p
(6.316)
q 2
A + i ((p qA) (p qA))
c
q 2
q
A i (A p + p A) .
c
c
(6.317)
The z component of the term (A p + p A) can be expressed as

(A p + p A)
z = Ax py Ay px + px Ay py Ax
= [Ax , py ] [Ay , px ] ,
Eyal Buks
(6.318)
186
6.7. Solutions
thus, with the help of Eq. (3.76) one finds that

dAx dAy
(A p + p A)
z = i
= i ( A)
z . (6.319)
dy
dx
Similar results can be obtained for the x and y components, thus
3
q 2 q
q 42
p A
= p A B.
c
c
c
(6.320)
j, m| Jx |j, m = j, m| Jy |j, m = 0 ,
(6.321)
Jx2 + Jy2 = J2 Jz2 ,
(6.322)
20. Since
and
one finds that

3
4
j, m| (Jx )2 + (Jy )2 |j, m = j, m| J2 |j, m j, m| Jz2 |j, m

= 2 j 2 + j m2 .
(6.323)
21. The condition is
pA=Ap,
(6.324)
[px , Ax ] + [py , Ay ] + [pz , Az ] = 0 ,
(6.325)
or
or using Eq. (3.76)

Ax Ay
Az
+
+
=0,
x
y
z
(6.326)
A=0 .
(6.327)
or
22. The Hamiltonian is given by Eq. (1.62)

H=

2
p qc A
+ q ,
2m
(6.328)
thus, the coordinates representation of H | = E | is given by

r | H | = E r | .
Eyal Buks
(6.329)
187

Using the notation
r | = (r )
(6.330)
for the wavefunction together with Eqs. (3.23) and (3.29) one has

1
q 2
i A + q (r ) = E (r ) .
(6.331)
2m
c

2

2
p ec A
py eBx
p2x + p2z
c
H=
=
+
2m
2m
2m

2
2
p2x
p
1
cp
y
=
+ m 2c x
+ z ,
2m 2
eB
2m
(6.332)
where
c =
eB
.
mc
(6.333)
Using the clue

(x, y, z) = (x) exp (iky y) exp (ikz z)
(6.334)
one finds that the time independent Schrdinger equation for the wave
function (x) is thus given by

2

p2x
1
cky
2 kz2
2
+ m c x
(x) = E
(x) , (6.335)
2m 2
eB
2m
where px = i/x, thus the eigenenergies are given by

1
2 kz2
+
,
En,k = c n +
2
2m
(6.336)
where n is integer and k is real, and the ground state energy is

En=0,k=0 =
c
.
2
(6.337)
24. Using the gauge A = Bx

y the Hamiltonian is given by [see Eq. (1.62)]

2
p qc A
H=
qEx
2m

2
qBx
2
2
p
y
c
p + pz
= x
+
qEx .
2m
2m
(6.338)
Eyal Buks
188
6.7. Solutions
The last two terms can be written as
2

3
4
py qBx
p2y
c
1
2
qEx =
+ m 2c (x x0 ) x20 ,
2m
2m 2
(6.339)
where
c =
qB
,
mc
(6.340)
x0 =
mc2
qpy
qE
+
B .
q2 B2
mc
(6.341)
and
Substituting the trial wavefunction

(x, y, z) = (x) exp (iky y) exp (ikz z) ,
(6.342)
into the three dimensional Schrdinger equation yields a one dimensional

Schrdinger equation

2 ky2 + 2 kz2
1
p2x
1
2
2
2 2
0 ) m c x
0 +
+ m c (x x
(x) = E (x) ,
2m 2
2
2m
(6.343)
where px = i/x and where

mc2
qky
x
0 = 2 2 qE +
B .
q B
mc
(6.344)
This equation describes a harmonic oscillator with a minimum potential

at x = x
0 , with added constant terms that give rise to a shift in the
energy level, which are thus given by

2 ky2 + 2 kz2
1
1
En,ky ,kz = c n +
m 2c x20 +
2
2
2m

2 2
1
mc E
cky E 2 kz2
= c n +
+
,
2
2
2B
B
2m
(6.345)
where n = 0, 1, 2, and where the momentum variables ky and kz can
take any real value.
25. The Schrdinger equation reads

2
p
ec A
+ U (y) (x, y) = E (x, y) ,
(6.346)
2m
Eyal Buks
189

where
p
= i .
Employing the gauge A = By
x one has

2
p2y
px + ec By
+
+ U (y) (x, y) = E (x, y) ,
2m
2m
(6.347)
where px = i/x and py = i/y. By substituting the trial wavefunction

(x, y) = exp (ikx) (y) ,
one obtains a one dimensional Schrdinger equation for (y)

e
2
p2y
1
2 2
c By + k
+
+ m 0 y (y) = E (y) ,
2m
2m
2
(6.348)
(6.349)
or

p2y
eBk
2 k2 1
2 2
+
+ mc0 y
y (y) = E (y) ,
2m
2m
2
mc
(6.350)
where 2c0 2c + 20 and c = |e| B/mc. This can also be written as

2
p2y
1
eBk
2 k2 20
2
+ mc0 y 2 2
+
(y) = E (y) . (6.351)
2m 2
m c c0
2m 2c0
This is basically a one-dimensional Schrdinger equation with a parabolic
potential of a harmonic oscillator and the eigenenergies are thus given
by:

1
2 k2 20
E (n, k) = c0 n +
,
+
2
2m 2c0
where n = 0, 1, 2, and k is real.
26. It is convenient to choose a gauge having cylindrical symmetry, namely
1
A= rB.
2
(6.352)
For this gauge A = 0, thus according to Eq. (6.167) the Hamiltonian

is given by
H=
Eyal Buks
1 2
q
q2
p pA+
A2 .
2
c
2c2
(6.353)
190
6.7. Solutions
The Schrdinger equation in cylindrical coordinates (, z, ) is given by
(note that A = (B/2) )

2

2
iqB
B
1
1 2 2
q2
+ 2
+ 2 +
+
= E .

2
z
2c 2c2
2
(6.354)
The particle is constrained to move along the ring, which is located at

z = 0 and = a, thus the effective one dimensional Schrdinger equation
of the system is given by
2 2 iqB q 2 a2 B 2
+
+
= E .
2a2 2
2c
8c2
(6.355)
a) Consider a solution of the form

1
exp (im) ,
() =
2a
(6.356)
where the pre factor (2a)1/2 ensures normalization. The continuity requirement that (2) = (0) implies that m must be an integer. Substituting this solution into the Schrdinger equation (6.355)
yields
2 m2 qBm q 2 a2 B 2
Em =
+
2a2
2c
8c2
!

"
2
2 2
2
qBa
1
qBa
=
m+
m2
2a2
c
4
c

2
qBa2
2
m
=
2a2
2c

2
2

=
m
,
2a2
0
(6.357)
where
= Ba2 ,
(6.358)
is the magnetic flux threading the ring and

0 =
ch
.
q
(6.359)
b) In general the current density is given by Eq. (4.200). For a wavefunction having the form
(r) = (r) ei(r) ,
Eyal Buks
(6.360)
191

where both and are real, one has

q
J = Im [ ( () + (i))] (A)
c
2
q 2
=
() A
c
2
||
q
=
() A .
c
In the present case one has
B
A=
,
2
m
=
,
a
and the normalized wavefunctions are
(6.361)
(6.362)
(6.363)
1
exp (im) ,
m () =
2a
(6.364)
thus
Jm =
1
2a

m q aB

=
m
.
a c 2
2a2
0
(6.365)
Note that the following holds

|Jm | =
c Em
.
q
(6.366)
27. The Hamiltonian can be written as

L2
L2 L2z
H=
+ z
2Ixy
2Iz

L2
1
1
=
+
L2z ,
2Ixy
2Iz
2Ixy
(6.367)
Thus the states |l, m (the standard eigenstates of L2 and Lz ) are eigenstates of H and the following holds
H |l, m = El,m |l, m ,
(6.368)
where
El,m = 2
l (l + 1)
+
2Ixy
1
1
2Iz
2Ixy
m2
(6.369)
Using the expression
Eyal Buks
192
6.7. Solutions
Y11 (, )
3
sin ei ,
8
(6.370)
one finds that

sin cos =

2 1
Y1 Y11 ,
3
(6.371)
thus the normalized state at t = 0 can be written as

1
| (0) = (|1, 1 |1, 1) .
2
(6.372)
Since E1,1 = E1,1 the state | (0) is stationary. Moreover

(t)| Lz | (t) = (0)| Lz | (0)
1
= ((1, 1| 1, 1|)) Lz ((|1, 1 |1, 1))
2
1
= ((1, 1| 1, 1|)) (( |1, 1 |1, 1))
2
=0.
(6.373)
28. With the help of the relations
L+ + L
,
2
L+ |l, m = l (l + 1) m (m + 1) |l, m + 1 ,

L |l, m = l (l + 1) m (m 1) |l, m 1 .
Lx =
(6.374)
(6.375)
(6.376)
one finds
a)
b)
010

1 0 1 .
Lx =
2 010
1
Lx =
2
2
1 1
2 2
c)
1
2 0
Lx =
2
Eyal Buks
(6.377)
1
010
2
1 0 1 1 = .
2
1
010
2
1
2
1
2
010
1 0 1 0 = 0 .
1
010
2
(6.378)
(6.379)
193

d)

exp (i) 0
0
iLz
.
0
1
0
=

Dz () = exp

0
0 exp (i)
(6.380)
e) In general

i (d) L n
i (d) L n
=1
+ O (d)2 ,
Dn (d) = exp

(6.381)
thus
i(d)

Dx (d) =
2
0
29. Using
i(d)
2
i(d)
2

2
+ O (d)
i(d)
.
2
(6.382)
Lz = xpy ypx ,
0

x=
ax + ax ,
2m
0

ay + ay ,
y=
2m
0

m
px = i
ax + ax ,
2
0

m
py = i
ay + ay ,
2
one finds

i ,
Lz =
ax + ax ay + ay ay + ay ax + ax
2

= i ax ay ax ay .
(6.383)
(6.384)
(6.385)
(6.386)
(6.387)
(6.388)
a) Thus

Lz = i x y x y .
b) Using
relations
, thecommutation
ax , ax = 1 ,
,
ay , ay = 1 ,
one finds
Eyal Buks
(6.389)
(6.390)
(6.391)
194
6.7. Solutions
) 2*

Lz = 2 x , y , z | ax y x ay x ay ax y |x , y , z
3

4
2
2
2
2
2
2
= 2 |x | 1 + |y | + |y | 1 + |x | x y (x ay ) ,
(6.392)
thus
3

4
2
2
(Lz )2 = 2 |x |2 1 + |y |2 + |y |2 1 + |x |2 + x y x y x y (x ay )2
3

4
2
2
2
2
2
2
= 2 |x | 1 + |y | + |y | 1 + |x | 2 |x | |y |

2
2
= 2 |x | + |y | ,
(6.393)
and
.
Lz = |x |2 + |y |2 .
(6.394)
30. The normalization constant can be chosen to be A = 1/ 2. In general:

L+ + L
,
2
L+ |l, m = l (l + 1) m (m + 1) |l, m + 1 ,

L |l, m = l (l + 1) m (m 1) |l, m 1 .
Lx =
(6.395)
(6.396)
(6.397)

(L |1, 1 L+ |1, 1)
Lx | =
2 2
(|1, 0 |1, 0)
=
=0,
2
(6.398)
thus
Lx = 0 .
(6.399)
b) Using Lx | = 0 one finds

'
( ) *
(Lx )2 = L2x Lx 2 = 0 0 = 0 .
(6.400)
L2
L2
L2
L2
L2
+ z z =
+ z ,
2I1 2I2 2I1
2I1 2Ie
(6.401)
I1 I2
.
I1 I2
(6.402)
31. The Hamiltonian can be expressed as

H=
where
Ie =
Thus, the angular momentum states |l, m, which satisfy

Eyal Buks
195

L2 |l, m = l (l + 1) 2 |l, m ,
Lz |l, m = m |l, m ,
(6.403)
(6.404)
are eigenvector of H
H |l, m = El,m |l, m ,
(6.405)
where
l (l + 1) 2 m2 2
2
=
+
=
2I1
2Ie
2I1
El,m

2
2 I1
l (l + 1) m + m
. (6.406)
I2
a) Since [H, Lz ] = 0 one has

iLz
iLz
exp
H exp
=H,

(6.407)
thus for the ground state l = m = 0

Az () = 0 | H |0 = E0,0 = 0 .
(6.408)
b) The operator Lx can be expressed as

Lx =
L+ + L
.
2
In general

L+ |l, m = l (l + 1) m (m + 1) |l, m + 1 ,

L |l, m = l (l + 1) m (m 1) |l, m 1 ,
thus
L+ |0, 0 = L |0, 0 = 0 ,
and consequently

iLx
exp
|0 = |0 ,
(6.409)
(6.410)
(6.411)
(6.412)
(6.413)
thus
Ax () = 0 | H |0 = E0,0 = 0 .
(6.414)
32. The wavefunction of a point particle is given by

(r) = (x + y + 2z) f (r) ,
(6.415)

where f (r) is a function of the radial coordinate r = x2 + y 2 + z 2 . As
can be see from Eqs. (6.131) and (6.132), which are given by
Eyal Buks
196
6.7. Solutions
0
3
=
sin ei ,
8
0
3
0
Y1 (, ) =
cos .
4
the following holds
0

2
x=r
Y11 + Y11
3
0

2 1
Y1 + Y11
y = ir
3
0
4 0
z=r
Y .
3 1
and thus
0

1 + i 1 1 + i 1
0
Y1 + Y1 + 2Y1 rf (r) .
(r) = 2
3
2
2
Y11 (, )
(6.416)
(6.417)
(6.418)
(6.419)
(6.420)
(6.421)
a) In a measurement of L2 the only possible outcome is 22 .

b) In a measurement of Lz the outcome and have both probability
1/6, whereas the outcome 0 has probability 2/3.
33. The notation |1 , 2 is used to label the common eigenvectors of the
operator S1z , S2z , S21 and S22 , where 1 {+, } and 2 {+, }. The
following holds [see Eqs. (6.69) and (6.76)]

S1z |1 , 2 = 1 |1 , 2 ,
2

S2z |1 , 2 = 2 |1 , 2 ,
2
(6.422)
(6.423)
and
S21 |1 , 2 = S22 |1 , 2 =
32
|1 , 2 .
4
(6.424)

S2 = S21 + S22 + S1 S2 + S2 S1 .
(6.425)
Any operator of the first particle commutes with any operator of the
second one thus
S2 = S21 + S22 + 2S1 S2
= S21 + S22 + 2 (S1x S2x + S1y S2y + S1z S2z ) .
(6.426)
In terms of the operators S1 and S2 , which are related to S1x , S2x ,
S1y and S2y by
Eyal Buks
197

S1+ + S1
S1+ S1
, S1y =
,
2
2i
S2+ + S2
S2+ S2
S2x =
, S2y =
,
2
2i
2
S is given by
S1x =
S2 = S21 + S22 + S1+ S2 + S1 S2+ + 2S1z S2z .
(6.427)
(6.428)
(6.429)
With
, 2 the
- help
, Eqs. (6.24) and (6.41) one finds that
S , Sz = S21 + S22 + S1+ S2 + S1 S2+ + 2S1z S2z , S1z + S2z
= [S1+ S2 + S1 S2+ , S1z + S2z ]
= [S1+ , S1z ] S2 + [S1 , S1z ] S2+ + S1+ [S2 , S2z ] + S1 [S2+ , S2z ]
= (S1+ S2 + S1 S2+ + S1+ S2 S1 S2+ ) ,
(6.430)
thus
, 2
(6.431)
S , Sz = 0 .
b) The following holds [see Eqs. (6.71) and (6.72)]
|+, +
2000
|+, +
|+,
2 0 1 1 0 |+,
S2
|, + = 0 1 1 0 |, + ,
|,
0002
|,
(6.432)
and
|+, +
100 0
|+, +
|+,
0 0 0 0 |+,
Sz
|, + = 0 0 0 0 |, + .
|,
0 0 0 1
|,
(6.433)
198
It is thus easy to show that the following set of 4 ket vectors

|+, |, +
,
(6.434)
|S = 0, M = 0 =
2
|S = 1, M = 1 = |+, + ,
(6.435)
|+, + |, +
|S = 1, M = 0 =
,
(6.436)
2
|S = 1, M = 1 = |, ,
(6.437)
forms the desired complete and orthonormal basis of common eigenvectors of S2 and Sz , and the following holds
S2 |S, M = S (S + 1) 2 |S, M ,
(6.438)
Sz |S, M = M |S, M .
(6.439)
34. The following holds [see Eqs. (6.32) and (6.36)]
Eyal Buks
6.7. Solutions
Jn =

sin ei J+ + ei J
+ cos Jz ,
2
(6.440)
thus [see Eqs. (6.63), (6.64), (6.65) and (6.66)]

j, m| Jn |j, m = m cos ,
(6.441)
and [see Eqs. (6.43) and (6.44)]

sin2 (J+ J + J J+ )
+ cos2 Jz2 |j, m
4

sin2 J2 Jz2
= j, m|
+ cos2 Jz2 |j, m
2

2
2
2 sin j (j + 1) m
2
2
=
+ cos m ,
2
j, m| Jn2 |j, m = j, m|
(6.442)
thus the expectation value is given by Jn = m cos and the variance
is given by
'
(
j (j + 1) m2
(Jn )2 = 2
sin2 .
2
(6.443)
35. Define the vector of operators = (x , y , z ), where

a2 a2
,
2
a2 + a2
y = i
,
2
aa + a a
z =
.
2
x =
Using Eq. (5.13), which is given by

, a, a = 1 ,
(6.444)
(6.445)
(6.446)
(6.447)
one finds that
[x , y ] = 2iz ,
(6.448)
[y , z ] = 2ix ,
(6.449)
[z , x ] = 2iy ,
(6.450)
thus
[i , j ] = 2iijk k ,
(6.451)
where i, j, k {x, y, z}. The operator S (, ) (6.188) can be rewritten as

Eyal Buks
199

,
S (, ) = exp ei + + ei ,
(6.452)
where
1
a2
(x + iy ) =
,
2
2
1
a2
= (x iy ) = .
2
2
+ =
(6.453)
(6.454)
The vector of Pauli matrices = (x , y , z ) satisfies a similar set of commutation relations [ i , j ] = 2iijk k as the set (6.451). Thus, all identities that are derived for the vector of Pauli matrices = (x , y , z ) are
applicable for the vector = (x , y , z ) provided that the derivation
uses only the commutation relations [i , j ] = 2iijk k . With the help
of the identity (6.138) one finds that the 2 2 matrix s (, ), which is
defined by [compare with Eq. (6.452)]
,
s (, ) = exp ei + + ei ,
(6.455)
where

1
01
+ = ( x + iy ) =
,
00
2

1
00
= ( x iy ) =
,
10
2
(6.456)
(6.457)
is given by
s (, ) =
cosh ei sinh
i
e
sinh cosh
Furthermore, with the help of the following matrix identity

cosh ei sinh
ei sinh cosh

i
1 e tanh
=
0
1
log(cosh )

e
0
0
elog(cosh )

1
0
,
ei tanh 1
(6.458)
(6.459)
and the relations 2+ = 2 = 0 one has
Eyal Buks
200
6.7. Solutions
s (, )

= exp ei tanh +
exp ( log (cosh ) z )

exp ei tanh .
(6.460)
The above expression for s (, ) yields a similar identity for the operator
S (, )
i

e 2
S (, ) = exp
a tanh
2

log (cosh )
exp
aa + a a
2
i

e
exp
a2 tanh .
2
(6.461)

1
Q () =
dx
x0

x a2
x2
exp 2 2 + 2
a
2 x
x0
2
0
: exp a a :

x
x2
a2
exp 2 + 2 a
.
2x0
x0
2
(6.462)
Since the integrated function is in normal ordering the integration can

be performed while disregarding the nonvanishing commutation relation
between a and a , namely by treating these operators as if they where
c-numbers
1
Q () =
x0
: e

1+ 12 x2
2x2
0
dx

2 a+ a
x
x0
(a+a )2
2
: ,
(6.463)
or with the help of the identity (5.136) one finds that

Eyal Buks
201

0
2
:e
1 + 2

2
a+ a
1+ 12
(a+a )
Q () =
:
0
(1)2
12 a2 a2
2
1+2 aa
2
1+2
=
:
e
: .
1 + 2
(6.464)
Thus, using the notation

= e ,
(6.465)
and the identities

1 e2
= tanh ,
1 + e2
2e
1
,
=
1 + e2
cosh

2
1 e
1
=1
,
1 + e2
cosh
one has
(6.466)
(6.467)
(6.468)
2
2
1
2
tanh a a
+( cosh
2
1)aa :
:
e
Q () =
1 + 2
0
tanh 2
tanh 2
1
1
=
e 2 a : e( cosh 1)aa : e 2 a .
cosh
This can be further simplified with the help of Eq. (5.102)

0
tanh 2
tanh 2
1
Q () =
e 2 a e log(cosh )a a e 2 a .
cosh
(6.469)
(6.470)
Using also
0
1
1
= e 2 log(cosh )
cosh
and
a a +
1
aa + a a
=
2
2
one has
Q () = e
tanh 2
a
2
log(cosh )
2
2
a
(aa +a a) e tanh
2
.
The last result together with Eq. (6.190) leads to

S (, 0) = Q e .
Eyal Buks
(6.471)
(6.472)
202
7. Central Potential
Consider a particle having mass m in a central potential, namely a potential

V (r) that depends only on the distance

r = x2 + y 2 + z 2
(7.1)
from the origin. The Hamiltonian is given by
H=
p2
+ V (r) .
2m
(7.2)

[H, Lz ] = 0 ,
,
H, L2 = 0 .
(7.3)
(7.4)
[xi , pj ] = i ij ,
Lz = xpy ypx ,
(7.5)
(7.6)
Solution 7.0.1. Using
one has
, 2
- ,
- ,
- ,
p , Lz = p2x , Lz + p2y , Lz + p2z , Lz
,
- ,
= p2x , xpy p2y , ypx
= i (2px py + 2py px )
=0,
(7.7)
and
,
- ,
- ,
- ,
r2 , Lz = x2 , Lz + y 2 , Lz + z 2 , Lz
,
- ,
= y x2 , px + y 2 , py x
=0.
(7.8)
Thus Lz commutes with any smooth function of r2 , and consequently

[H, Lz ] = 0. In
way one can show that [H, Lx ] = [H, Ly ] = 0,
, a similar
and therefore H, L2 = 0.
Chapter 7. Central Potential

In classical physics the corresponding Poissons brackets relations hold
{H, Lx } = {H, Ly } = {H, Lz } = 0 ,
(7.9)
and

H, L2 = 0 .
(7.10)
These relations imply that classically the angular momentum is a constant of

the motion [see Eq. (1.40)]. On the other in quantum mechanics, as we have
seen in section 2.12 of chapter 2, the commutation relations
[H, Lz ] = 0 ,
,
H, L2 = 0 ,
(7.11)
(7.12)
imply that it is possible to find a basis for the vector space made of common
eigenvectors of the operators H, L2 and Lz .
7.1 Simultaneous Diagonalization of the Operators H,

L2 and Lz
We start by proving some useful relations:
L2 = r2 p2 (r p)2 + ir p .
(7.13)
Solution 7.1.1. The following holds

L2z = (xpy ypx )2
= x2 p2y + y 2 p2x xpy ypx ypx xpy
= x2 p2y + y 2 p2x xpx ([py , y] + ypy ) ypy ([px , x] + xpx )

= x2 p2y + y 2 p2x xpx ypy ypy xpx + i (xpx + ypy ) .
(7.14)
Using the relation

xpx xpx = x ([px , x] + xpx ) px = ixpx + x2 p2x ,
(7.15)
ixpx = x2 p2x xpx xpx ,
(7.16)
or
one has
Eyal Buks
204
7.1. Simultaneous Diagonalization of the Operators H, L2 and Lz

i
(xpx + ypy )
2

1
+ x2 p2x xpx xpx + y 2 p2y ypy ypy .
2
L2z = x2 p2y + y 2 p2x xpx ypy ypy xpx +
(7.17)
By cyclic permutation one obtains similar expression for L2x and for L2y . Combining these expressions lead to
L2 = L2x + L2y + L2z

i
1 2 2
(ypy + zpz ) +
y py ypy ypy + z 2 p2z zpz zpz
2
2

i
1 2 2
2 2
2 2
+z px + x pz zpz xpx xpx zpz + (zpz + xpx ) +
z pz zpz zpz + x2 p2x xpx xpx
2
2

i
1 2 2
2 2
2 2
+x py + y px xpx ypy ypy xpx + (xpx + ypy ) +
x px xpx xpx + y 2 p2y ypy ypy
2
2

= x2 + y2 + z 2 p2x + p2y + p2z (xpx + ypy + zpz )2 + i (xpx + ypy + zpz )
= y 2 p2z + z 2 p2y ypy zpz zpz ypy +
= r2 p2 (r p)2 + ir p .
(7.18)

1 2
1
2
2
2
r r | 2 2 r | L | .
r | p | =
r r2
r
(7.19)
Solution 7.1.2. Using the identities
and
L2 = r2 p2 (r p)2 + ir p ,
(7.20)
r | r | = r r | ,
(7.21)
r | p | =

r | ,
i
(7.22)
one finds that

r | L2 | = r | r2 p2 | r | (r p)2 | + i r | r p | .
(7.23)
The following hold

r | r p | = ir r | = ir
r | ,
r

2
r | (r p)2 | = 2 r r |
r

2
= 2 r2 2 + r r | ,
r
r
Eyal Buks
(7.24)
(7.25)
205

r | r2 p2 | = r2 r | p2 | ,
(7.26)
thus
r | p2 | = 2
2
2
+
r2 r r
r |

1
2
r
|
L
|
,
2 r2
(7.27)
(7.28)
or
r | p | =
1
1 2
r r | 2 2 r | L2 |
r r2
r
The time-independent Schrdinger equation in the coordinates representation

r | H | = E r | ,
(7.29)
where the Hamiltonian H is given by Eq. (7.2), can thus be written using the
above results as

2 1 2
1
2
r | H | =
r r | 2 2 r | L | +V (r ) r | . (7.30)
2m r r2
r
7.2 The Radial Equation


r | = (r ) = R (r ) Ylm , .
(7.31)

r | L2 | = 2 l (l + 1) (r ) .
Substituting into Eq. (7.30) yields an equation for R (r)

1
2 1 d2
rR (r) 2 l (l + 1) R (r) + V (r) R (r) = ER (r) .
2m r dr2
r
(7.32)
(7.33)
The above equation, which is called the radial equation, depends on the quantum number l, however, it is independent on the quantum number m. The
different solutions for a given l are labeled using the index k

2 1 d2
1
rR
l
(l
+
1)
R
+ V Rkl = ERkl .
(7.34)
kl
kl
2m r dr2
r2
It is convenient to introduce the function ukl (r), which is related to Rkl (r)
by the following relation
1
Rkl (r) = ukl (r) .
r
Eyal Buks
(7.35)
206
7.2. The Radial Equation

Substituting into Eq. (7.34) yields an equation for ukl (r)
2 2

d
+
V
(r)
ukl (r) = Ekl ukl (r) ,
eff
2m dr2
(7.36)
where the effective potential Veff (r) is given by

Veff (r) =
l (l + 1) 2
+ V (r) .
2mr2
(7.37)
The total wave function is thus given by

1
klm (r) = ukl (r) Ylm (, ) .
r
(7.38)
Since the spherical harmonic Ylm (, ) is assumed to be normalized [see Eq.

(6.114)], to ensure that klm (r) is normalized we require that
1=

0
drr |Rkl (r)| =

0
dr |ukl (r)|2 .
(7.39)
In addition solutions with different k are expected to be orthogonal, thus

druk l (r) ukl (r) = kk .
(7.40)
The wave functions klm (r) represent common eigenstates of the operators
H, Lz and L2 , which are denoted as |klm and which satisfy the following
relations
klm (r ) = r |klm ,
(7.41)
H |klm = Ekl |klm ,

L2 |klm = l (l + 1) 2 |klm ,
Lz |klm = m |klm .
(7.42)
(7.43)
(7.44)
and
The following claim reveals an important property of the radial wavefunction near the origin (r = 0):
Claim. If the potential energy V (r) does not diverge more rapidly than 1/r
near the origin then
lim u (r) = 0 .
(7.45)
r0
Eyal Buks
207

Proof. Consider the case where near the origin u (r) has a dominant power
term having the form rs (namely, all other terms are of order higher than s,
and thus become negligibly small for sufficiently small r). Substituting into
Eq. (7.36) and keeping only the dominant terms (of lowest order in r) lead
to
l (l + 1) 2 s2
2
s (s 1) rs2 +
r
=0,
2m
2m
(7.46)
thus s = l or s = l + 1. However, the solution s = l for l 1 must

be rejected since for this case the normalization condition (7.39) cannot be
satisfied as the integral diverges near r = 0. Moreover, also for l = 0 the
solution s = l must be rejected. For this case (r) 1/r near the origin,
however, such a solution contradicts Eq. (7.30), which can be written as
2 2
(r) + V (r) (r) = E (r) .
2m
(7.47)
since
2
1
= 4 (r) .
r
(7.48)
We thus conclude that only the solution s = l + 1 is acceptable, and consequently limr0 u (r) = 0.
7.3 Hydrogen Atom

The hydrogen atom is made of two particles, an electron and a proton. It
is convenient to employ the center of mass coordinates system. As is shown
below, in this reference frame the two body problem is reduced into a central
potential problem of effectively a single particle.
Exercise 7.3.1. Consider two point particles having mass m1 and m2 respectively. The potential energy V (r) depends only on the relative coordinate r = r1 r2 . Show that the Hamiltonian of the system in the center of
mass frame is given by
H=
p2
+ V (r) ,
2
(7.49)
where the reduced mass is given by

=
m1 m2
.
m1 + m2
Eyal Buks
(7.50)
208
7.3. Hydrogen Atom

Solution 7.3.1. The Lagrangian is given by
L=
m1 r 21 m2 r 22
+
V (r1 r2 ) .
2
2
(7.51)
In terms of center of mass r0 and relative r coordinates, which are given by

m1 r1 + m2 r2
r0 =
,
(7.52)
m1 + m2
r = r1 r2 ,
(7.53)
the Lagrangian is given by

2
2
m1
2
r
m1 r 0 + m1m+m
2
0
m1 +m2
2
+
V (r)
L=
2
2
M r 20 r2
=
+
V (r) ,
2
2
(7.54)
where the total mass M is given by
M = m1 + m2 ,
(7.55)
and the reduced mass by

=
m1 m2
m1 m22 + m2 m21
=
.
2
m1 + m2
(m1 + m2 )
(7.56)
Note that the Euler Lagrange equation for the coordinate r0 yields that
r0 = 0
(since the potential is independent on r0 ). In the center of mass frame r0 = 0.
The momentum canonically conjugate to r is given by
p=
L
.
r
(7.57)
Thus the Hamiltonian is given by

H = p r L =
p2
+ V (r) .
2
(7.58)
For the case of hydrogen atom the potential between the electron having
charge e and the proton having charge e is given by
V (r) =
e2
.
r
(7.59)
Since the protons mass mp is significantly larger than the electrons mass
me (mp 1800me ) the reduced mass is very close to me
=
me mp
me .
me + mp
Eyal Buks
(7.60)
209

The radial equation (7.36) for the present case is given by
2 2

d
+
V
(r)
eff
2 dr2
(7.61)
where
Veff (r) =
e2 l (l + 1) 2
+
.
r
2r2
(7.62)
In terms of the dimensionless radial coordinate

r
,
=
a0
(7.63)
where
a0 =
2
= 0.53 1010 m ,
e2
(7.64)
is the Bohrs radius, and in terms of the dimensionless parameter

0
Ekl
,
kl =
EI
(7.65)
where
e4
= 13.6 eV ,
22
is the ionization energy, the radial equation becomes

d2
2 + Vl () + 2kl ukl = 0
d
EI =
(7.66)
(7.67)
where
2 l (l + 1)
Vl () = +
.
(7.68)
10
8
6
4
2
0
-2
-4
-6
-8
-10
The function Vl () for l = 0 (solid line) and l = 1 (dashed line).
Eyal Buks
210
7.3. Hydrogen Atom

We seek solutions of Eq. (7.67) that represent bound states, for which Ekl
is negative, and thus kl is a nonvanishing real positive. In the limit
the potential Vl () 0, and thus it becomes negligibly small in comparison
with kl [see Eq. (7.67)]. Therefore, in this limit the solution is expected to
be asymptotically proportional to ekl . To ensure that the solution is normalizable the exponentially diverging solution e+kl is excluded. Moreover,
as we have seen above, for small the solution is expected to be proportional
to l+1 . Due to these considerations we express ukl (r) as
ukl (r) = y () l+1 ekl .
(7.69)
Substituting into Eq. (7.67) yields an equation for the function y ()

2

d
(l + 1)
2 (1 kl (l + 1))
d
+
y=0.
(7.70)
+
2
kl
d2
Consider a power series expansion of the function y ()

y () =
<
cq q .
(7.71)
q=0
Substituting into Eq. (7.70) yields
<
q=0
q (q 1) cq q2 + 2 (l + 1)
2kl
<
q=0
<
qcq q2
q=0
qcq q1 + 2 (1 kl (l + 1))
<
cq q1 = 0 ,
q=0
(7.72)
thus
2 [kl (q + l) 1]
cq
=
.
cq1
q (q + 2l + 1)
(7.73)
We argue below that for physically acceptable solutions y () must be a polynomial function [i.e. the series (7.71) needs to be finite]. To see this note that
for large q Eq. (7.73) implies that
lim
cq
q cq1
2kl
.
q
(7.74)
Similar recursion relation holds for the coefficients of the power series expansion of the function e2kl
e2kl =
<
q=0
where
Eyal Buks
cq q ,
(7.75)
211

q
(2kl )
,
q!
(7.76)
cq
2kl
.
=
cq1
q
(7.77)
cq =
thus
This observation suggests that for large the function ukl asymptotically
becomes proportional to ekl . However, such an exponentially diverging solution must be excluded since it cannot be normalized. Therefore, to avoid
such a discrepancy, we require that y () must be a polynomial function.
As can be see from Eq. (7.73), this requirement is satisfied provided that
kl (q + l) 1 = 0 for some q. A polynomial function of order k 1 is obtained when kl is taken to be given by
kl =
1
,
k+l
(7.78)
where k = 1, 2, 3, . With the help of Eq. (7.73) the polynomial function

can be evaluated. Some examples are given below
yk=1,l=0 () = c0 ,
yk=1,l=1 () = c0 ,

yk=2,l=0 () = c0 1

yk=2,l=1 () = c0 1
(7.79)
(7.80)
n=k+l ,
(7.83)

,
(7.81)
2
.
(7.82)
6
The coefficient c0 can be determined from the normalization condition.
As can be seen from Eqs. (7.65) and (7.78), all states having the same
sum k + l, which is denoted as
have the same energy. The index n is called the principle quantum number.
Due to this degeneracy, which is sometimes called accidental degeneracy, it
is common to label the states with the indices n, l and m, instead of k, l and
m. In such labeling the eigenenergies are given by
En =
EI
,
n2
(7.84)
where
n = 1, 2, .
(7.85)
For a given n the quantum number l can take any of the possible values
l = 0, 1, 2, , n 1 ,
Eyal Buks
(7.86)
212
7.4. Problems
and the quantum number m can take any of the possible values
m = l, l + 1, , l 1, l .
(7.87)
The level of degeneracy of the level En is given by

gn = 2
n1

(2l + 1) = 2
l=0

2 (n 1) n
+ n = 2n2 .
2
(7.88)
Note that the factor of 2 is due to spin. The normalized radial wave functions
of the states with n = 1 and n = 2 are found to be given by
3/2
1
er/a0 ,
(7.89)
R10 (r) = 2
a0

3/2
r
r
1
R20 (r) = 2
e 2a0 ,
(7.90)
a0
2a0

3/2
r
1
r
R21 (r) =
e 2a0 .
(7.91)
2a0
3a0
The wavefunction n,l,m (r) of an eigenstate with quantum numbers n, l and
m is given by
nlm (r, , ) = Rnl (r) Ylm (, ) .
(7.92)
While the index n labels the shell number, the index l labels the sub-shell.
In spectroscopy it is common to label different sub-shells with letters:
l=0 s
l=1p
l=2d
l=3 f
l=4g
7.4 Problems
1. Consider the wave function with quantum numbers n, l, and m of a
hydrogen atom n,l,m (r).
a) Show that the probability current in spherical coordinates r, , is
given by
2

n,l,m (r)

,
(7.93)
Jn,l,m (r) = m
r sin
is a unit vector orthogonal to

where is the reduced mass and
z
and
r.
Eyal Buks
213

b) Use the result of the previous section to show that the total angular
momentum expectation value is given by L = m
z.
2. Show that the average electrostatic potential in the neighborhood of an
hydrogen atom in its ground state is given by

1
1
2r
exp
,
(7.94)
=e
+
a0
r
a0
where a0 is the Bohr radius.
3. An hydrogen atom is in its ground state. The distance r between the
electron and the proton is measured. Calculate the expectation value r
and the most probable value r0 (at which the probability distribution
function obtains a maximum).
4. Tritium, which is labeled as 3 H, is a radioactive isotope of hydrogen.
The nucleus of tritium contains 1 proton and two neutrons. An atom
of tritium is in its ground state, when the nucleus suddenly decays into
a helium nucleus, with the emission of a fast electron, which leaves the
atom without perturbing the extra-nuclear electron. Find the probability
that the resulting He+ ion will be left in:
a) 1s state.
b) 2s state.
c) a state with l = 0.
5. At time t = 0 an hydrogen atom is in the state
| (t = 0) = A (|2, 1, 1 + |2, 1, 1) ,
where A is a normalization constant and where |n, l, m denotes the eigenstate with quantum numbers n, l and m. Calculate the expectation value
x at time t.
6. Find the ground state energy E0 of a particle having mass m in a central
potential V (r) given by
+
0 arb
V (r) =
,
(7.95)
else

where r = x2 + y2 + z 2 .
7. Consider a particle having mass m in a 3D potential given by
V (r) = A (r a) ,
(7.96)

where r =
x2 + y 2 + z 2 is the radial coordinate, the length a is a
constant and () is the delta function. For what range of values of the
constant A the particle has a bound state.
8. Consider a particle having mass m in a 3D central potential given by
+
U0 r r0
V (r) =
.
(7.97)
0 r > r0
Eyal Buks
214
7.5. Solutions

where r = x2 + y 2 + z 2 is the radial coordinate, U0 is real and r0 is
positive. For what range of values of the potential depth U0 the particle
has a bound state.
9. A spinless point particle is in state |. The state vector | is an eigenvector of the operators Lx , Ly and Lz (the x, y and z components of
the angular momentum vector operator). What can be said about the
wavefunction (r ) of the state |?
10. Consider two (non-identical) particles having the same mass m moving
under the influence of a potential U (r), which is given by
1
U (r) = m 2 r2 .
2
In addition, the particles interact with each other via a potential given
by
1
V (r1 , r2 ) = m 2 (r1 r2 )2 ,
2
where r1 and r2 are the (three dimensional) coordinate vectors of the first
and second particle respectively. Find the eigenenergies of the system.
7.5 Solutions
1. In general the current density is given by Eq. (4.200). For a wavefunction
having the form
(r) = (r) ei(r) ,
(7.98)
where both and are real, one has

Im [ ( + i)]
2
=
J=
||2
.
(7.99)
a) The wavefunction n,l,m (r) is given by

nlm (r, , ) = Rnl (r) Ylm (, ) = Rnl (r) Flm () eim , (7.100)
where both Rnl and Flm are real, thus
2

n,l,m (r)
Jn,l,m (r) =
(m) .
Eyal Buks
(7.101)
215

In spherical coordinates one has
=x
+y
+
z
x
y
z
1
=
r
+
+
,
r
r
r sin
(7.102)
thus
2

n,l,m (r)

.
Jn,l,m (r) = m
r sin
(7.103)
b) The contribution of the volume element d3 r to the angular momentum with respect to the origin is given by r Jn,l,m (r) d3 r. In
spherical coordinates the total angular momentum is given by
2

n,l,m (r)
3
L = rJn,l,m (r) d r = m
r
d3 r . (7.104)
r sin
By symmetry, only the component along
z of r
contributes, thus
L = m
z.
(7.105)
2. The charge density of the electron in the ground state is given by

2
e
2r
.
(7.106)
= e 1,0,0 (r) = 3 exp
a0
a0
The Poissons equation is given by
2 = 4 .
(7.107)
To verify that the electrostatic potential given by Eq. (7.94) solves this
equation we calculate
1 d2
(r)
r dr2

2
e d
r
2r
=
+ 1 exp
r dr2
a0
a0

2r
4e exp a0
=
a30
= 4 .
2 =
(7.108)
Note also that

lim (r) = 0 ,
(7.109)
as is required for a neutral atom.

Eyal Buks
216
7.5. Solutions
3. The radial wave function of the ground state is given by
R10 (r) = 2
1
a0
3/2

r
exp
a0
(7.110)
thus the probability distribution function of the variable r is given by

3

4 r
2r
2
.
(7.111)
f (r) = |rR10 (r)| =
exp
r a0
a0
Thus
r =
rf (r) dr = 4a0
3
x3 exp (2x) dx = a0 .
2
The most probable value r0 is found from the condition

8r0
2r0
df
0=
= 4 exp
(a0 r0 ) ,
dr
a0
a0
(7.112)
(7.113)
thus
r0 = a0 .
(7.114)
4. The radial wave function of a hydrogen-like atom with a nucleus having

charge Ze is found by substituting e2 by Ze2 in Eqs. (7.89), (7.90) and
(7.91), namely
3/2
Z
(Z)
R10 (r) = 2
eZr/a0 ,
(7.115)
a0

3/2
Zr
Z
(Z)
R20 (r) = (2 Zr/a0 )
e 2a0 ,
(7.116)
2a0

3/2
Zr
Z
Zr 2a
(Z)
e 0 .
(7.117)
R21 (r) =
2a0
3a0
The change in reduced mass is neglected. Therefore
a) For the 1s state
2

7 2a3 2
2
(Z=1)
(Z=2)
0
2
Pr (1s) = drr R10 R10 = 3
= 0.702 .
a0 33
0
b) For the 2s state
2

2

16 a30
(Z=1) (Z=2)
Pr (2s) = drr2 R10 R20 = 6
(2 3) = 0.25 .
a0 8
0
Eyal Buks
217

c) For this case the probability vanishes due to the orthogonality between spherical harmonics with different l.
5. The normalization constant is chosen to be A = 1/ 2. Since both states

|2, 1, 1 and |2, 1, 1 have the same energy the state | is stationary.
The following holds
nlm (r, , ) = Rnl (r) Ylm (, ) ,

3/2
r
1
r
R21 (r) =
e 2a0 ,
2a0
3a0
0
1
3
Y11 (, ) =
sin ei ,
2 2
0
1
3
1
Y1 (, ) =
sin ei ,
2 2
x = r sin cos .
(7.118a)
(7.118b)
(7.118c)
(7.118d)
(7.118e)
In general

n l m | x |nlm =
dr r R
n l
Rnl
1
d (cos )
2
0

d sin cos Ylm
Ylm .
(7.119)
thus
2, 1, 1|x|2, 1, 1
2
d cos = 0 ,
(7.120)
2, 1, 1|x|2, 1, 1
2
d cos = 0 ,
(7.121)
2, 1, 1|x|2, 1, 1
2
d cos e2i = 0 ,
(7.122)
2, 1, 1|x|2, 1, 1
2
d cos e2i = 0 ,
(7.123)
and therefore
x (t) = 0 .
(7.124)
6. The radial equation is given by [see Eq. (7.36)]

2 d2
l (l + 1) 2
+
+
V
(r)
uk,l (r) = Ek,l uk,l (r) .
2m dr2
2mr2
Eyal Buks
(7.125)
218
7.5. Solutions
Since the centrifugal term l (l + 1) 2 /2mr2 is non-negative the ground
state is obtained with l = 0. Thus the ground state energy is [see Eq.
(4.181)]
E0 =
2 2
.
2m (b a)2
(7.126)
7. The radial equation is given by

2 d2
l (l + 1) 2
+
+
V
(r)
uk,l (r) = Ek,l uk,l (r) .
2m dr2
2mr2
(7.127)
The boundary conditions imposed upon u (r) by the potential are

u (0) = 0 ,

u a+ = u a
(7.128)
(7.129)
du (a+ ) du (a )
2
= u (a) .
dr
dr
a0
where
a0 =
2
.
mA
(7.130)
(7.131)
Since the centrifugal term l (l + 1) 2 /2mr2 is non-negative the ground

state is obtained with l = 0. We seek a solution for that case having the
form
+
sinh (r)
r<a
u (r) =
,
(7.132)
sinh (a) exp ( (r a)) r > a
where
2mE
=
.
(7.133)
The condition (7.130) yields

sinh (a) cosh (a) =
2
sinh (a) ,
a0
(7.134)
or
a0
1
=
.
2
1 + coth (a)
A real solution exists only if
a0
<a,
2
or
A>
Eyal Buks
2
.
2ma
(7.135)
(7.136)
219

8. The radial equation is given by

2 d2
l (l + 1) 2
+
+ V (r) uk,l (r) = Ek,l uk,l (r) .
2m dr2
2mr2
(7.137)
The boundary condition that is imposed upon u (r) at the origin is u (0) =
0. Since the centrifugal term l (l + 1) 2 /2mr2 is non-negative the ground
state is obtained with l = 0. For that case the solution in the range r r0
has the form u (r) = sin kr, where k is related to the energy E by
h2 k2
= E + U0 .
2m
(7.138)
In the range r > r0 the general solution has the form u (r) = Aer +
Ber , where
h2 2
= E .
2m
(7.139)
A bound state can be obtained provided that E < 0 (to ensure that
is real) and B = 0 (to ensure that limr u (r) = 0; it is assumed that
is non-negative). The requirements that both u (r) and du/dr [see Eq.
(4.127)] are continuous at r = r0 yield (for the case B = 0)
sin kr0 = Aer0 ,
k
cos kr0 = Aer0 ,
thus the following must hold

tan kr0 =
k
.
(7.140)
(7.141)
(7.142)
Since both k and are required to be nonnegative, the above condition

can be satisfied only if tan kr0 0, which implies that
0
2m (E + U0 )
kr0 =
r0 .
(7.143)
h2
2
This together with the requirement that E < 0 yield
0
2mU0
r0 ,
h2
2
or
U0
Eyal Buks
2 h2
.
8mr02
(7.144)
(7.145)
220
7.5. Solutions
9. The state vector | is an eigenvector of the operators Lx , Ly , therefore it
is easy to see that it consequently must be an eigenvector of the operator
[Lx , Ly ] with a zero eigenvalue. Thus, since [Lx , Ly ] = iLz , one has
Lz | = 0. Similarly, one finds that Lx | = Ly | = 0. Therefore, |
is also an eigenvector of the operator L2 = L2x + L2y + L2z with a zero
eigenvalue. Therefore the wavefunction has the form
R (r )
m=0
(r ) = R (r ) Yl=0
, =
,
4
(7.146)
where the radial function R (r ) is an arbitrary normalized function.

10. The Lagrangian is given by

1
m r 21 + r 22
1
L=
m2 r21 + r22 m 2 (r1 r2 )2 .
(7.147)
2
2
2
In terms of center of mass r0 and relative r coordinates, which are given

by
r1 + r2
r0 =
,
(7.148)
2
r = r1 r2 ,
(7.149)
the Lagrangian is given by

3
2
2 4
m r 0 + 12 r + r 0 12 r
L=
2
2
2
1
1
1
1
2
m
r0 + r + r0 r
m 2 r2
2
2
2
2
2 1 2

m 2r0 + 2 r
1
1
1
=
m 2 2r20 + r2 m 2 r2
2
2
2
2
2
2

M r 0 1
r
1
=
M 2 r20 +
2 + 2 2 r2 ,
2
2
2
2
(7.150)
where the total mass M is given by

M = 2m ,
(7.151)
and the reduced mass [see also Eq. (7.50)] by

=
m
.
2
(7.152)
The Lagrangian L describes two decoupled three dimensional harmonic

oscillators. The first, which is associated with the center of mass motion,
has mass M = 2m and angular resonance frequency , whereas the
Eyal Buks
221

second one, which is associated with the relative
coordinate r, has mass
= m/2 and angular resonance frequency 2 + 2 2 . The quantum
energy eigenvectors are denoted by |n0x , n0y , n0z , nx , ny , nz , where all
six quantum numbers n0x , n0y , n0z , nx , ny and nz are integers, and the
corresponding eigenenergies are given by

3
En0x ,n0y ,n0z ,nx ,ny ,nz =
+ n0x + n0y + n0z
2

3
2
2
+ nx + ny + nz .
+ + 2
2
(7.153)
Eyal Buks
222
8. Density Operator
Consider an ensemble of N identical copies of a quantum system. The ensemble can be divided into subsets, where all systems belonging to the same
subset have the
state vector. Let N wi be the number of systems having
same
*
state vector
(i) , where
0 wi 1 ,
(8.1)
and where

wi = 1 .
(8.2)
The state vectors are all assumed to be normalized
'
(
(i)
(i) = 1 .
(8.3)
Consider a measurement of an observable A, having a set of eigenvalues

{an } and corresponding set of eigenvectors {|an }
A |an = an |an .
(8.4)
The set of eigenvectors {|an } is assumed to be orthonormal and complete

n
am |an = nm ,
(8.5)
|an an | = 1 .
(8.6)
Consider a measurement of the observable A done on a system that is randomly chosen
from
* the ensemble. The probability to choose

* a system having
state vector
(i)
is w
i . Given that the state vector is
(i) , the expectation
)
*
value of A is (i)
A
(i) [see Eq. (2.84)]. Thus, the expectation (average)
value of such a measurement done on a system that is randomly chosen from
the ensemble is given by
(
(
2
'
A =
wi (i)
A
(i) =
wi
an
(i)
an .
(8.7)
i
Chapter 8. Density Operator

Claim. The expectation value can be expressed as
A = Tr (A) ,
(8.8)
where
=

i
('
wi
(i) (i)
(8.9)
is the density operator.
Proof. Let {|bm } be an orthonormal and complete basis for the vector space

|bm bm | = 1 .
(8.10)
m
The following holds
(
'
wi (i)
A
(i)
A =
i

'
(
wi (i)
A |bm bm
(i)
bm |

i
= Tr (A) ,
('
wi
(i) (i)
A |bm
(8.11)
where
=

i
('
wi
(i) (i)
.
Below we discuss some basic properties of the density operator:

Exercise 8.0.1. Show that = .
Solution 8.0.1. Trivial by the definition (8.9).
Exercise 8.0.2. Show that Tr () = 1.
Solution 8.0.2. Using a complete orthonormal basis
has
Eyal Buks
<
m |bm bm |
= 1 one
224
Tr () =
bm |
'
(
wi (i)
(i)

i
"
('
(i)
(i)
wi

|bm
!
"
'
(
(i)
wi
|bm bm |
(i)

i
!

wi
=1.
(8.12)
Exercise 8.0.3. Show that for any normalized state | the following holds
0 | | 1 .
(8.13)
Solution 8.0.3. Clearly, 0 | | since
2
('

'

wi
(i) (i) | =
wi
(i) |
0 .
| | =
(8.14)
one has
.)
'
(i)
|
| (i)
(i) = 1 .
(8.16)
On the other hand, according to the Schwartz inequality [see Eq. (2.166)],
which is given by

|u |v| u |u v |v ,
(8.15)
Moreover,
<
i wi
| | =
= 1, thus

i
'
wi
(i) |
1 .
(8.17)

Exercise 8.0.4. Show that Tr 2 1.
Solution 8.0.4. The fact that is Hermitian (i.e., = ) guaranties the

existence of a complete orthonormal basis {|qm } of eigenvectors of , which
satisfy

m
and
qm |qm = mm ,
|qm qm | = 1 ,
Eyal Buks
(8.18)
(8.19)
225

|qm = qm |qm ,
(8.20)
where the eigenvalues qm are real. Using this basis one has

2
Tr 2 =
qm | 2 |qm =
qm
.
m
(8.21)
According to inequality (8.13)

0 qm = qm | |qm 1 ,
(8.22)
thus
2
qm
Tr 2 =
m
!

qm
"2
= (Tr ())2 = 1 .
(8.23)
Definition 8.0.1. An ensemble is said to be pure if its density operator can

be expressed as
= | | .

Exercise 8.0.5. Show that Tr 2 = 1 iff represents a pure ensemble.
Solution 8.0.5. (i)Assuming that represents a pure ensemble,

one has

2 = , thus Tr 2 = Tr () = 1. (ii) Assume that Tr 2 = 1. Since
is Hermitian (i.e., = ), there is a complete orthonormal basis {|qm } of
eigenvectors of , such that

(8.24)
|qm qm | = 1 ,
(8.25)
|qm = qm |qm ,
(8.26)
and
qm |qm = mm ,
where the eigenvalues qm are real. Moreover, according to inequality (8.22)

0 qm 1 .
(8.27)

For this basis the assumption Tr 2 = 1 yields
2
1 = Tr 2 =
qm .
(8.28)
Moreover, also Tr () = 1, thus

1=
qm .
(8.29)
Both equalities can be simultaneously satisfied only if

+
1 m = m0
qm =
.
0 m = m0
(8.30)
For this case = |qm0 qm0 |, thus represents a pure ensemble.

Eyal Buks
226
8.2. Quantum Statistical Mechanics
8.1 Time Evolution

Consider a density operator
('

(t) =
wi
(i) (t) (i) (t)
,
(8.31)
*
where the state vectors
(i) (t) evolve in time according to
(
d
(i)
i
= H
(i) ,
(8.32)
)dt(i)
'
i
= (i)
H ,
(8.33)
dt
where H is the Hamiltonian. Taking the time derivative yields
!
"
('
('
d
1
(i)
(i)
(i)
(i)
(t)

=
wi H
wi
(t)
H , (8.34)
dt
i
i
i
thus
d
1
= [, H] .
dt
i
(8.35)
This result resembles the equation of motion (4.37) of an observable in the

Heisenberg representation, however, instead of a minus sign on the right hand
side, Eq. (4.37) has a plus sign.
Alternatively, the time evolution of the operator can be expressed in
terms of
the time
u (t,
* evolution operator
* t0 ), which relates the state vector
at time
(i) (t0 ) with its value
(i) (t) at time t [see Eq. (4.4)]
(
(
(i)
(8.36)
(t) = u (t, t0 )
(i) (t0 ) .
With the help of this relation Eq. (8.31) becomes
(t) = u (t, t0 ) (t0 ) u (t, t0 ) .
(8.37)
8.2 Quantum Statistical Mechanics

Consider an ensemble of identical copies of a quantum system. Let H be the
Hamiltonian having a set of eigenenergies {Ei } and a corresponding set of
eigenstates {|i}, which forms an orthonormal and complete basis

i
H |i = Ei |i ,
|i i| = 1 .
Eyal Buks
(8.38)
(8.39)
227

Consider the case where the ensemble is assumed to be a canonical ensemble
in thermal equilibrium at temperature T . According to the laws of statistical
mechanics the probability wi to find an arbitrary system in the ensemble in
a state vector |i having energy Ei is given by
wi =
1 Ei
,
e
Z
(8.40)
where = 1/kB T , kB is Boltzmanns constant, and where

Z=
eEi
(8.41)
is the partition function.

Exercise 8.2.1. Show that the density operator can be written as
=
eH
.
Tr (eH )
(8.42)
Solution 8.2.1. According to the definition (8.9) one has

1 Ei
=
wi |i i| =
e
|i i| .
Z i
i
(8.43)
Moreover, the following hold

Z=
eEi =
i| eH |i = Tr eH ,
(8.44)
and

i
eEi |i i| =

i
eH |i i| = eH

i
|i i| = eH ,
(8.45)
thus
=
eH
.
Tr (eH )
(8.46)
As will be demonstrated below [see Eq. (8.446)], the last result for can
also be obtained from the principle of maximum entropy.
8.3 Problems
1. Consider a spin 1/2 in a magnetic field B = B
z and in thermal equilibrium at temperature T . Calculate S u
, where S is the vector operator
of the angular momentum and where u
is a unit vector, which can be
described using the angles and
u
= (sin cos , sin sin , cos ) .
Eyal Buks
(8.47)
228
8.3. Problems
2. A spin 1/2 particle is an eigenstate of the operator Sy with eigenvalue
+/2.
a) Write the density operator in the basis of eigenvectors of the operator
Sz .
b) Calculate n , where n is integer.
c) Calculate the density operator (in the same basis) of an ensemble of
particles, half of them in an eigenstate of Sy with eigenvalue +/2,
and half of them in an eigenstate of Sy with eigenvalue /2.
d) Calculate n for this case.
3. A spin 1/2 is at time t = 0 in an eigenstate of the operator S = Sx sin +
Sz cos with an eigenvalue +/2, where is real and Sx and Sz are
the x and z components, respectively, of the angular momentum vector
operator. A magnetic field B is applied in the x direction between time
t = 0 and time t = T .
a) The z component of the angular momentum is measured at time
t > T . Calculate the probability P+ to measure the value /2.
b) Calculate the density operator of the spin at times t = T .
4. A spin 1/2 electron is put in a constant magnetic field given by B =
B
z, where B is a constant. The system is in thermal equilibrium at
temperature T .
a) Calculate the correlation function
Cz (t) = Sz (t) Sz (0) .
(8.48)
b) Calculate the correlation function

Cx (t) = Sx (t) Sx (0) .
(8.49)
5. Consider a harmonic oscillator .

with frequency . Show that the variance
of the number operator N = N 2 N 2 (where N = a a ) is given

by
a) N =
0 for energy eigenstates.
b) N = N for coherent states.
c) N = N (N + 1) for thermal states.
6. Consider a harmonic oscillator having angular resonance frequency .
The oscillator is in)thermal
equilibrium at temperature T . Calculate the
*
expectation value x2 .
7. Consider a harmonic oscillator in thermal equilibrium at temperature T ,
whose Hamiltonian is given by
H=
p2
m2 x2
+
.
2m
2
(8.50)
Show that the density operator is given by
Eyal Buks
229

=

d2 | | P () ,
(8.51)
where | is a coherent state, d2 denotes infinitesimal area in the

complex plane,
!
"
2
1
||
P () =
exp
,
(8.52)
N
N
and where N is the expectation value of the number operator N .
H=
p2
m2 x2
+
.
2m
2
(8.53)
Calculate the probability distribution function f (x) of the random variable x.

9. An LC oscillator (see figure) made of a capacitor C in parallel with an
inductor L, is in thermal equilibrium at temperature T . The charge in
the capacitor q is being measured.
a) Calculate the expectation

' value
( q of q.
b) Calculate the variance (q)2 .
10. Consider an observable A having a set of eigenvalues {an }. Let Pn be

a projector operator onto the eigensubspace corresponding to the eigenvalue an . A given physical system is initially described by the density operator 0 . A measurement of the observable A is then performed. What
is the density operator 1 of the system immediately after the measurement?
11. A model that was proposed by von Neumann describes an indirect measurement process of a given observable A. The observable A is assumed
to be a function of the degrees of freedom of a subsystem, which we refer
to as the measured system (MS). The indirect measurement is performed
by first letting the MS to interact with a measuring device (MD), having its own degrees of freedom, and then in the final step, performing a
quantum measurement on the MD. The MS is assumed to initially be in
a pure state | (i.e. its density operator is assumed to initially be given
by 0 = | |). Let A be an observable operating on the Hilbert space
Eyal Buks
230
8.3. Problems
of the MS. The initial state of the MS can be expanded in the basis of
eigenvectors {|an } of the observable A

cn |an ,
(8.54)
| =
n
where cn = an | and where

A |an = an |an .
(8.55)
For simplicity, the Hamiltonian of the MS is taken to be zero. The MD is

assumed to be a one dimensional free particle, whose Hamiltonian vanishes, and whose initial state is labeled by |i . The position wavefunction
(x ) = x |i of this state is taken to be Gaussian having width x0
!
2 "
1
1 x
(x ) =
exp
.
(8.56)
1/2
2 x0
1/4 x
0
The interaction between the MS and the MD is taken to be given by

V (t) = f (t) xA ,
(8.57)
where f (t) is assumed to have compact support with a peak near the
time of the measurement.
a) Express the vector state of the entire system | (t) at time t in the
basis of states {|p |an }. This basis spans the Hilbert space of
the entire system (MS and MD). The state |p |an is both, an
eigenvector of A (with eigenvalue an ) and of the momentum p of the
MD (with eigenvalue p ).
b) In what follows the final state of the system after the measurement
will be evaluated by taking the limit t . The outcome of the
measurement of the observable A, which is labeled by A, is determined by performing a measurement of the momentum variable p of
the MD. The outcome, which is labeled by P, is related to A by
A=
P
,
pi
(8.58)
where
pi =
dt f (t ) .
(8.59)
Calculate the probability distribution g (A) of the random variable

A.
c) Consider another measurement that is performed after the entanglement between the MS and the MD has been fully created. The
Eyal Buks
231

additional measurement is associated with the observable B, which
is assumed to be a function of the degrees of freedom of the MS only.
of the observable B is given by
Show that the expectation value B

=
B
an | BR |an ,
(8.60)
n
where the operator R , which is called the reduced density operator,

is given by
R =
cn cn e
n ,n
a
n an
2
2
|an an | .
(8.61)
12. A particle having mass m moves in the xy plane under the influence of
a two dimensional potential V (x, y), which is given by
V (x, y) =

m 2 2
x + y 2 + m 2 xy ,
2
(8.62)
where both and are real constants. Calculate

) *in thermal equilibrium
at temperature T the expectation values x, x2 .
13. Consider a harmonic oscillator having angular resonance
*
) frequency and
mass m. Calculate the correlation function G (t) = x(H) (t) x(H) (0) ,
where x(H) (t) is the Heisenberg representation of the position operator,
for the cases where
a) the oscillator is in its ground state.
b) the oscillator is in thermal equilibrium at temperature T .
14. In general, the Wigner function of a point particle moving in one dimension is given by
1
W (x , p ) =
2
p x
dx exp i
5
6
x
x
x +
, (8.63)
2

where is the density operator of the system, and where |x represents

an eigenvector of the position operator x having eigenvalue x , i.e. x |x =
x |x . As can be seen from Eq. (4.227), the Wigner function is the inverse
Weyl transformation of the density operator divided by the factor of 2.
Consider the case of a point particle having mass m in a potential of a
harmonic oscillator having angular frequency . Calculate the Wigner
function W (x , p ) for the case where the system is in a coherent sate
|.
15. A particle having mass m is in the ground state of the one-dimensional
potential well V1 (x) = (1/2) m 2 (x x )2 for times t < 0 . At time
t = 0 the potential suddenly changes and becomes V2 (x) = (1/2) m 2 x2 .
Calculate the Wigner function of the system at times t > 0.
Eyal Buks
232
8.3. Problems
16. Consider a point particle having mass m in a potential of a harmonic
oscillator having angular frequency . Calculate the Wigner function
W (x , p ) for the case where the system is in thermal equilibrium at
temperature T .
oscillator having angular frequency . Calculate the Wigner function
W (x , p ) for the case where the system is in the number state |n = 1.
18. The Wigner function of a point particle moving in one dimension is
given by Eq. (8.63). Show that the marginal distributions x | |x and
p | |p of the position x and momentum p observables, respectively, are
given by

1
x | |x =
dp W (x , p ) ,
(8.64)

p | |p = 1
dx W (x , p ) .
(8.65)
19. Show that for a pure state the Wigner function is bounded by |W (x , p )|
1/2. Note that this bound together with Eqs. (8.64) and (8.65) can be
used to demonstrate the uncertainty principle (3.10).
20. Consider a particle having mass m moving along the x axis under the
influence of the potential V (x). Show that the time evolution of the
Wigner function W (8.63) is governed by
2l
<
dW
2l+1 V 2l+1 W
2i
,
= {H, W } +
2l+1
dt
(p )2l+1
l=1 (2l + 1)! (x )
(8.66)
where H is the Hamiltonian and {H, W } is the Poissons brackets of H

and W .
21. The function W (X , P ) is defined as the inverse Fourier transform of
(, )
the function W
W (X , P ) =
1
(2)

(, ) eiX +iP ,
dd W
(8.67)
(, ) is given by
where the function W
(, ) = Tr [exp (iX iP ) ] ,
W
(8.68)
X and P are dimensionless position and momentum operators, which are

given by
X=
Eyal Buks
a + a
a a
, P = ,
2
i 2
(8.69)
233

and which satisfy [X, P ] = i [see Eq. (5.13)] and is the density operator.
Show that
1
W (X , P ) =
2
dX
6

X
X
X
X +
eiX P , (8.70)
2

i.e. show that W (X , P ) is the Wigner function expressed in terms of

the dimensionless variables X and P [whereas Eq. (8.63) is the Wigner
function expressed in terms of the position x and momentum p variables].
22. Equation (8.67) can be rewritten as
W (X , P ) = Tr ( ) ,
(8.71)
where the operator is given by
(X , P ) =
1
(2)

dd ei(X
X )+i(P P )
(8.72)
Note that the operator given by Eq. (8.72) is the dimensionless version
of the Weyl kernel (4.46), which defines the Weyl transformation (4.45).
Show that

X + iP
X + iP
PD
,
(8.73)
(X , P ) = 1 D
2
2
where

D () = exp a a
(8.74)
is the displacement operator [see Eq. (5.36)], a is the annihilation operator and
P=
dX |X X |
(8.75)
is the parity operator [see Eq. (5.101)], where |X is an eigenvector of the

dimensionless position operator X having eigenvalue X , i.e. X |X =
X |X .
23. Homodyne Tomography - Consider a point particle having mass m
in a potential of a harmonic oscillator having angular frequency . The
normalized homodyne observable X with a real phase is defined by
X =
Eyal Buks
a ei + aei
,
2
(8.76)
234
8.3. Problems
where a and aare
annihilation and creation operators [see Eqs. (5.9) and
(5.10)]. Let w X be the normalized probability distribution function

of the observable
X
. The technique of homodyne detection can be used
to measure w X for any given value of the phase .
a) To generalize Eqs. (8.64) and (8.65) show that the following holds
for any real

w X =

dP W X cos P sin , X sin + P cos .
(8.77)
b) Show that the Wigner function

(8.63) can be extracted from the
measured distributions w X for all values of .

H=
p2
m2 x2
+
.
2m
2
(8.78)
Calculate the matrix elements x | |x of the density operator in the

basis of eigenvectors of the position operator x.
25. Consider a harmonic oscillator having angular resonance frequency .
The oscillator is in)thermal* equilibrium at temperature T . Calculate the
expectation value eiX , where X is given by [see Eq. (8.76)]
a ei + aei
,
2
X =
(8.79)
a and a are annihilation and creation operators [see Eqs. (5.9) and (5.10)]
and
both
* and are real. Use your result for the expectation value
) iX
to evaluate
e
the Wigner function of the system.
26. Show that when w X is independent the following holds

1
W (X , P ) =
2

0

d w
() J0 X 2 + P 2 ,
(8.80)

where w
() is the ( independent) Fourier transform of w X , i.e.
w
() =
Eyal Buks

dX w X eiX .
(8.81)
235

oscillator having angular frequency . Express the Wigner function
W (X , P ; t) at time t in terms of the Wigner function W (X , P ; 0)
at time t = 0.
28. Let W (X , P ) be the Wigner function of a system whose density operator is . Express the Wigner function W (X , P ) of a system whose
density operator
displaced
is
according to = D ( ) D ( ), where
D ( ) = exp a a is the displacement operator [see Eq. (5.36)],

and where is complex, in terms of the Wigner function of the undisplaced system W (X , P ).
29. Consider a weak measurement of the dimensionless position X [see Eq.
(8.69)] of a point particle moving in one dimension. In view of Eq. (8.61),
the reduced density operator of the system after the measurement, which
is labeled as R , is expected to be related to the density operator before
the measurement by the following relation
5
6 5
6 2
X
X
X
X
X
2
X
,
R
X +
= X

X +
e
2
2
2
2
(8.82)
where the dimensionless parameter characterizes the strength of the
measurement, and where |X represents an eigenvector of the dimensionless position operator X [see Eq. (8.69)] having eigenvalue X , i.e.
X |X = X |X . Express the reduced Wigner function WR (X , P ),
which is given by [see Eq. (8.70)]
1
WR (X , P ) =
2
dX
6
5

X
X
+
eiX P , (8.83)
X
2
R
in terms of the Wigner function W (X , P ) before the measurement.

30. A point particle having mass m is confined by the three dimensional
potential
1
(8.84)
V (r) = m2 r2 ,
2

2
2
2
where
) r2 *= x + y + z and where is a real constant. Calculate x
and x in thermal equilibrium at temperature T .
31. The entropy is defined by
= Tr ( log ) .
(8.85)
Show that is time independent.

32. Consider the 2 2 matrix , which is given by
=
Eyal Buks
1
(1 + k ) ,
2
(8.86)
236
8.3. Problems
where k = (kx , ky , kz ) is a three dimensional vector of complex numbers
and where = ( x , y , z ) is the Pauli matrix vector.
a) Under what conditions on k the matrix can represent a valid density
operator of a spin 1/2 particle?
b) Under what conditions on k the matrix can represent a valid density
operator of a spin 1/2 particle in a pure state?
c) Calculate the term Tr (
u ), where u
is a unit vector, i.e. u
u
= 1.
33. The matrix representation in the basis of eigenvectors of Sz of the density
operator of a spin 1/2 particle is given by
=
1
(1 + k ) ,
2
(8.87)
where k = (kx , ky , kz ) is a three dimensional vector of real numbers, and

= (x , y , z ) is the Pauli matrix vector. The entropy is defined by
= Tr ( log ) .
(8.88)
a) Calculate .
b) A measurement of Sz is performed. Calculate the entropy after the
measurement.
34. The maximum entropy principle - The entropy is defined by
() = Tr ( log ) .
(8.89)
Consider the case where the density matrix is assumed to satisfy a set of
contrarians, which are expressed as
gl () = 0 ,
(8.90)
where l = 0, 1, L. The functionals gl () maps the density operator

to a complex number, i,e. gl () C.
a) Find an expression for a density matrix that satisfy all these constrains, for which the entropy obtains a stationary point (maximum, minimum or a saddle point). Assume that the constrain l = 0
is the requirement that Tr () = 1, i.e. g0 () can be taken to be given
by
g0 () = Tr () 1 = 0 .
(8.91)
Moreover, assume that the other constrains l = 1, L are the requirements that the expectation values of the Hermitian operators
X1 , X2 , XL are the following real numbers X1 , X2 , ,XL respectively, i.e. gl () for l 1 can be taken to be given by
gl () = Tr (Xl ) Xl = 0 .
Eyal Buks
(8.92)
237

b) Express for the case of a microcanonical ensemble, for which the
only required constrain is (8.91).
c) Express for the case of a canonical ensemble, for which in addition
to the constrain is (8.91) the expectation value of the Hamiltonian
H is required to have a given value, which is labeled by H.
d) Express for the case of a grandcanonical ensemble, for which in addition to the constrain is (8.91) the expectation values of the Hamiltonian H and of the operator N are required to have given values,
which are labeled by H and N respectively. The operator N ,
which will be defined in chapter 16, is called the number of particles
operator.
35. Consider a point particle having mass m moving in one dimension under
the influence of the potential V (x). Calculate the canonical partition
function Zc [see Eq. (8.447)] in the classical limit, i.e. in the limit of high
temperature.
36. Let be the density operator of a given system. The system is composed
of two subsystems, each having its own degrees of freedom, which are
labeled as 1 and 2 (e.g. a system of two particles). Let {|n1 1 } ({|n2 2 })
be an orthonormal basis spanning the Hilbert space of subsystem 1
(2). The set of vectors {|n1 , n2 }, where |n1 , n2 = |n1 1 |n2 2 , forms an
orthonormal basis spanning the Hilbert space of the combined system.
For a general operator O the partial trace over subsystem 1 is defined
by the following relation

Tr1 (O)
(8.93)
1 n1 | O |n1 1 .
n1
Similarly, the partial trace over subsystem 2 is defined by

Tr2 (O)
2 n2 | O |n2 2 .
(8.94)
n2
The observable A1 is a given Hermitian operator on the Hilbert space of

subsystem 1. Show that the expectation value of a measurement of A1
that is performed on subsystem 1 is given by
A1 = Tr1 (1 A1 ) .
(8.95)
where the operator 1 , which is given by

1 = Tr2 ,
(8.96)
is called the reduced density operator of subsystem 1.

37. Let be the density operator of a given system. The total entropy of the
system is given by
= Tr ( log ) .
Eyal Buks
(8.97)
238
8.4. Solutions
As in the previous exercise, the system is composed of two subsystems,
which are labeled as 1 and 2. Let {|n1 1 } ({|n2 2 }) be an orthonormal
basis spanning the Hilbert space of subsystem 1 (2). The set of vectors {|n1 , n2 }, where |n1 , n2 = |n1 1 |n2 2 , forms an orthonormal basis
spanning the Hilbert space of the combined system. The reduced density
operators 1 and 2 of subsystems 1 and 2 respectively are giving by

1 =
(8.98)
2 n2 | |n2 2 = Tr2 ,
n2
2 =

n1
1 n1 | |n1 1
= Tr1 ,
(8.99)
and the subsystems entropies 1 and 2 are given by

1 = Tr1 (1 log 1 ) ,
2 = Tr2 (2 log 2 ) .
(8.100)
(8.101)
Show that
1 + 2 .
(8.102)
38. Consider a spin 1/2 in a magnetic field B = B

z and in thermal equilibrium at temperature T . Calculate the entropy , which is defined by
= Tr ( log ) ,
(8.103)
where is the density operator of the system.

39. A spin 1/2 is in a state |H, which satisfies the following relation

1
1 2 (Sx + Sz )
|H H| =
1+
,
(8.104)
2

2
where 1 is the identity operator, and where Sx and Sz are spin angular
momentum operators. In a measurement of Sz what is the probability
pz+ to obtain the value +/2?
40. Consider a harmonic oscillator of angular frequency and mass m in
thermal equilibrium at temperature T . Calculate the entropy , which is
defined by
= Tr ( log ) ,
(8.105)
where is the density operator of the system.
8.4 Solutions
Eyal Buks
239

H = Sz ,
(8.106)
where
=
|e| B
me c
(8.107)
is the Larmor frequency. In the basis of the eigenvectors of Sz

Sz | = | ,
2
one has

H | =
| ,
2
thus
eH
=
Tr (eH )
=

2
(8.108)
(8.109)
|+ +| + e

e 2
+e

2

2
| |
,
(8.110)
where = 1/kB T , and therefore with the help of Eqs. (2.102) and (2.103),
which are given by

Sx = (|+ | + | +|) ,
(8.111)
2

Sy = (i |+ | + i | +|) ,
(8.112)
2
one has
Sx = Tr (Sx ) = 0 ,
Sy = Tr (Sy ) = 0 ,
(8.113)
(8.114)
and with the help of Eq. (2.99), which is given by

Sz =

(|+ +| | |) ,
2
(8.115)
one has
Sz = Tr (Sz )
!
"

e 2 |+ +| + e 2 | |
= Tr
(|+ +| | |)

2
e 2 + e 2

e 2 e 2

2 e
2
+e 2

= tanh
,
2
2
=
Eyal Buks
(8.116)
240
8.4. Solutions
thus
S u
=
cos
tanh
2

2
(8.117)
2. Recall that
1
|; y
= (|+ i |) ,
2
(8.118)
a) thus
1
=

1 1 i
1
1 i =
.
i
2 i 1
b) For a pure state n = .

c) For this case

1
1 10
=
|+; y
+; y
| + |; y
; y
| =
,

2
2 01
(8.119)
(8.120)
=1
d) and
n =
1
2n
10
01
3. The state at time t = 0 is given by

cos 2
| (t = 0) =
,
sin 2
and the one at time t = T is

iT x
| (t = T ) = exp
| (t = 0) ,
2
(8.121)
(8.122)
(8.123)
where x is a Pauli matrix, and

=
eB
.
me c
(8.124)
Using the identity

i n
exp
= cos i n
sin ,
2
2
2
(8.125)
one finds

T
i n
T
T
cos T
2 i sin 2
exp
= cos
+ix sin
=
, (8.126)
T
i sin T
2
2
2
2 cos 2
Eyal Buks
241

thus
| (t = T ) =
T
cos T
2 i sin 2
T
T
i sin 2 cos 2

cos 2
sin 2
cos T
2 cos 2 + i sin 2 sin 2
T
T
i sin 2 cos 2 + cos 2 sin 2
.
(8.127)
a) The probabilities to measured /2 are thus given by

T
P+ = cos2
cos2 + sin2
sin2
2
2
2
2
1 + cos (T ) cos
=
,
(8.128)
2
and
T
P = cos2
sin 2 + sin2
cos 2
2
2
2
2
1 cos (T ) cos
=
.
(8.129)
2
b) The density operator is given by
11 = P+ ,
22 = P ,

T
21 = cos
cos + i sin
sin
i sin
cos + cos
sin
2
2
2
2
2
2
2
2
sin
i
=
sin T cos ,
2
2
12 = 21 .
H = Sz ,
(8.130)
where
=
eB
,
me c
(8.131)
thus, the density operator is given by

0
1 exp 2kB T

,
=
Z
0
exp 2k
BT
(8.132)
(8.133)
where
Z = exp
Eyal Buks

2kB T

+ exp
.
2kB T
242
8.4. Solutions
a) Using
Sz (t) = exp
iHt

iHt
Sz (0) exp
= Sz (0) ,
(8.134)
one finds
)
*

2
Cz (t) = Sz2 (0) = Tr Sz2 (0) =
.
4
(8.135)
b) The following holds

iSz t
iSz t
Sx (t) = exp
Sx (0) exp

= Sx cos (t) + Sy sin (t) ,
(8.136)
thus
)
*
Cx (t) = cos (t) Sx2 (0) + sin (t) Sy (0) Sx (0)
(8.137)
cos (t)
+ sin (t) Sy (0) Sx (0) .
4
In terms of Pauli matrices

exp 2k
0
2
0
i
0
1
T
B

Sy (0) Sx (0) =

Tr
i 0
10
4Z
0
exp 2k
BT

2
0
i
exp

2kB T

=
Tr
4Z
0
i exp 2k
BT

i2

=
tanh
,
4
2kB T
(8.138)
thus

2

Cx (t) =
cos (t) i sin (t) tanh
.
(8.139)
4
2kB T
5. The variance N is given by
a) For an energy eigenstate |n one has
N |n = n |n ,
(8.140)
thus
N = n| N |n = n ,
(8.141)
and
) 2*
N = n| N 2 |n = n2 ,
Eyal Buks
(8.142)
243

therefore
N = 0 .
(8.143)
b) For a coherent state | one has

a | = | ,
(8.144)
thus
2
N = | N | = | a a | = || ,
(8.145)
and
) 2*
,
N = | a aa a | = | a a, a + a a a | = ||2 + ||4 ,

=1
(8.146)
therefore
N =

N .
(8.147)
c) In general, for a thermal state one has

O = Tr (O) ,
(8.148)
where O is an operator,
1
= eH ,
(8.149)
Z

Z = Tr eH ,
(8.150)
and = 1/kB T and H is the Hamiltonian. For the present case

1
H = N +
,
(8.151)
2
thus
Eyal Buks
244
8.4. Solutions
N = Tr (N )
<
n| eH N |n
n=0
= <
n| eH |n
n=0
<
nen
n=0
<
en

< n
e
log
n=0
1
n=0
e
=
,
1 e
=
(8.152)
and ) *

N 2 = Tr N 2
<
n| eH N 2 |n
n=0
= <
n| eH |n
n=0
<
n2 en
n=0
<
n=0
en

<
1 2 2
en

2
n=0
<
en
n=0

e
+ 1 e
(1 e )
(8.153)
and therefore
) *
(N)2 = N 2 N2 =
e
2
(1 e )
= N (N + 1) . (8.154)
6. The density operator is given by

=
1 H
e
.
Z
(8.155)
where
Eyal Buks
245

1
Z = tr eH =
e(n+ 2 ) =
n=0
and = 1/kB T . Thus using

x2 =
1
e 2

, (8.156)
=
1 e
2 sinh
2

2 2
a + a
+ 2a a + 1 ,
2m
(8.157)
one finds
) 2*

x = Tr x2
1
=
n| x2 eH |n
Z n=0
1 (n+ 12 )
e
n| x2 |n
Z n=0

1 (n+ 12 )
1
n+
=
e
m Z n=0
2

1
1
d (n+ 12 )
=
e
.
m Z
d n=0
(8.158)
However
e(n+ 2 ) = Z ,
1
(8.159)
n=0
thus
) 2*
x =
1 d
log Z 1
m 2 d

d
sinh
d
2
1

=
m 2 sinh
2

1

=
coth
.
m 2 2
2
(8.160)
Note that at high temperatures 1

) 2*
kB T
x
,
m 2
(8.161)
as is required by the equipartition theorem of classical statistical mechanics.

Eyal Buks
246
8.4. Solutions
7. In the basis of number states the density operator is given by
<
eH |n n|
H
e
n=0
=
= <
Tr (eH )
n| eH |n
n=0
<
= n=0
<
=
e(N+ 2 ) |n n|
n=0
<
n=0
n| e(N+ 2 ) |n

1
en |n n|
<
n=0
en

= 1 e
en |n n| ,
n=0
(8.162)
where = 1/kB T . Thus, N is given by

N = Tr (N )

= 1 e
nen
n=0

= 1 e
en
n=0

1
= 1 e
1 e
e
=
.
1 e
(8.163)
Moreover, the following holds

N + 1 =
1
,
1 e
N
= e ,
N + 1
(8.164)
(8.165)
thus, can be rewritten as
Eyal Buks
247

= 1e
en |n n|
n=0
n

1
N
=
|n n| .
N + 1 n=0 N + 1
(8.166)
To verify the validity of Eq. (8.51) we calculate

n| |m =
d2 P () n | |m .
(8.167)

| = e
||2
2

n
|n ,
n!
n=0
(8.168)
one finds that

1
n| |m =
N

||2
d exp
N
2
"
n
2
m
.
e||
n! m!
(8.169)
Employing polar coordinates in the complex plane = rei , where r is

non-negative real and is real,
1
n| |m =
N n!m!
2 nm
=
N n!
2
1
(1+ N
n+m+1
)r
dre
2
0
dre(1+ N )r r2n+1 ,
1
dei(nm)

2nm
(8.170)
and the transformation of the integration variable

1
t = 1+
r2 ,
N

1
dt = 1 +
2rdr ,
N
(8.171)
(8.172)
lead to
Eyal Buks
248
8.4. Solutions
n| |m =
nm

N 1 +
1
N
n+1
nm
n+1

1
N 1 + N
N n nm
(1 + N )
n+1
n! 0

dtet tn

(n+1)=n!
in agreement with Eq. (8.166).

8. The density operator [see Eq. (8.51)] is given by

=
d2 | | P () ,
(8.173)
(8.174)

complex plane,
!
"
||2
1
P () =
exp
,
(8.175)
N
N
and where
N =
e
1 e
(8.176)
is the expectation value of the number operator N. Thus,

f (x ) = x | |x =
d2 P () x | |x .
By employing the expression for the wave function (x ) = x | of a
coherent state which is given by [see Eq. (5.51)]
(x ) = x |

2

x x 2
2 m 1/4
x
= exp
exp
+ i p
,
4

2x
(8.177)
where
2
Re () ,
m
0
.

2
x = | (x) | =
,
2m
x = | x | =
Eyal Buks
(8.178)
(8.179)
249

one finds that
1 m 1/2
N
"

!
2

x x
||2
2
exp 2
d exp
N
2x
1 m 1/2
=
N

2
!
"

2
||
x
d2 exp
exp 2 .
Re ()
N
2
f (x ) = x | |x =
m 1/2
=
e
1 + 2 N
1
=
x2
m
1+2N
1
e2
(1 + 2 N)

m
2
x
1
= e

x2
m
1+2N
where
=

(1 + 2 N) ,
m
and where
1 + 2 N = 1 + 2
e
= coth
1 e

2
9. Recall that the LC circuit is a harmonic oscillator.

a) In terms of the annihilation and creation operators
0

L
ip
a=
q+
,
2
L
0

L
ip
a =
q
,
2
L
one has 0

a + a ,
2L

1
H = a a +
.
2
q=
Eyal Buks
(8.180)
(8.181)
(8.182)
(8.183)
(8.184)
250
8.4. Solutions
The density operator is given by
=
1 H
e
,
Z
(8.185)
1
,
kB T
(8.186)
where
=
and
Z = Tr eH =
e(n+ 2 ) =
1
n=0
e 2
1
=
, (8.187)

1e
2 sinh
2
thus
1
q = Tr (q) =
Z

n| a + a eH |n = 0 . (8.188)
2L n=0
b) Similarly
) 2*

q = Tr q 2

2
1
=
n| a + a eH |n
2L Z n=0

1 (n+ 12 )
1 1
=

n
+
e
L2 Z n=0
2
1 1 dZ
L 2 Z d
C

=
coth
.
2
2kB T
(8.189)
10. In general, 0 can be expressed as
('

0 =
wi
(i) (i)
,
(8.190)
) (i)
(i) *
<
where 0 wi 1, i wi = 1, and where

= 1. Assume first
(i) *
that the system is initially in the state . The probability for this to
be the case is wi . In general, the possible results of a measurement of the
observable A are the eigenvalues {an }. The probability p
n to *measure the
eigenvalue an given that the system is initially in state
(i) is given by
'
(
pn = (i)
Pn
(i) .
(8.191)
After a measurement of A with an outcome an the state vector collapses
onto the corresponding eigensubspace and becomes
Eyal Buks
251
(
Pn
(i)
(i)
.
)
*.
(i)
Pn
(i)
(8.192)
*
Thus, given that the system is initially in state
(i) the final density
operator is given by
*
) (i)
(
'

Pn
Pn
(i)
(i)
(i)
(i)
.)
1 =

Pn
* .)
*
(i)
Pn
(i)
(i)
Pn
(i)
n
('

=
Pn
(i) (i)
Pn .
n
(8.193)
Averaging over all possible initial states thus yields
('

(i)
1 =
wi 1 =
Pn
wi
(i) (i)
Pn =
Pn 0 Pn . (8.194)
n
11. Since [V (t) , V (t )] = 0 the time evolution operator from initial time t0
to time t is given by

i t
u (t, t0 ) = exp
dt V (t )
t0

ipi A
= exp
x ,

(8.195)
where
pi =
dt f (t ) .
(8.196)
t0
While the initial state of the entire system at time t0 is given by | (t0 ) =
|i |, the final state at time t is given by
| (t) = u (t, t0 ) | (t0 )

=
cn Jn |i |an ,
n
where the operator Jn is given by

ipi an
Jn = exp
x .
Eyal Buks
(8.197)
(8.198)
252
8.4. Solutions
/
a) By introducing the identity operator dp |p p | = 1MD on the
Hilbert space of the MD, where |p are eigenvectors of the momentum operator p, which is canonically conjugate to x, the state | (t)
can be expressed as

| (t) =
cn dp p | Jn | i |p |an .
(8.199)
n
With the help of the general identity (3.76), which is given by

[p, A (x)] = i
dA
,
dx
(8.200)
where A (x) is a function of the operator x, one finds that

pJn |p = ([p, Jn ] + Jn p) |p
= (pi an + p ) Jn |p ,
(8.201)
thus the vector Jn |p is an eigenvector of p with eigenvalue pi an +p .

Moreover, note that this vector, which is labeled as |p + pi an
Jn |p , is normalized, provided that |p is normalized, since Jn is
unitary. The momentum wavefunction (p ) = p |i of the state
|i is related to the position wavefunction x |i by a Fourier transform [see Eq. (3.60)]
/
ip x
dx e x | i
(p ) =
2
!
2 "
1
1 p
=
exp
,
1/2
2 p0
1/4 p
0
(8.202)
where
p0 =

.
x0
(8.203)
In terms of (p ) the state | (t) thus can be expressed as

| (t) =
cn dp p pi an |i |p |an
n

n
Eyal Buks
cn
dp (p pi an ) |p |an .
(8.204)
253

b) The probability distribution g (A) of the random variable A can be
calculated using Eq. (8.204)

2
|(an | pi A|) | (t)|
g (A) = pi
n
= pi

n
1/2
|cn | | (pi (A an ))|

n
|cn |2 e
(Aan )2
,
(8.205)
where
pi
x0
=
=
p0
dt f (t ) .
(8.206)
t0
The expectation value of A is given by

A =
=

n

n
dA g (A ) A
2
|cn |
1/2
2
dA e(A ) (A + an )
|cn | an .
(8.207)
c) The density operator of the entire system is taken to be given by

f = | () ()| for this case. The additional measurement is
associated with the observable B, which is assumed to be a function
of the degrees of freedom of the MS only. This assumption allows
of the observable B as
expressing the expectation value B
= Tr (Bf )
B

=
dp an | p | Bf |p |an
n

n
an | BR |an ,
where R , which is given by

R = dp p | f |p ,
Eyal Buks
(8.208)
(8.209)
254
8.4. Solutions
is called the reduced density operator. Note that R is an operator
on the Hilbert space of the MS. With the help of the expressions

| () =
cn
dp (p pi an ) |p |an ,
(8.210)
n
()| =
cn
dp (p pi an ) an | p | , (8.211)
one finds that

cn cn dp
R =
n ,n
(p pi an ) (p pi an ) |an an | .
(8.212)
By employing the transformation of integration variable

x=
2p pi (an + an )
,
2p0
(8.213)
and its inverse

pi an + an
p = p0 x +
,
p0
2
(8.214)
one finds that

a a
2
2 n 2 n
dp (p pi an ) (p pi an ) = e
,
(8.215)
thus
R =
cn cn e
n ,n
a
n an
2
2
|an an | .
12. It is convenient to employ the coordinate transformation

x+y
x = ,
2
x
y
y = .
2
(8.216)
(8.217)
(8.218)
The Lagrangian of the system can be written using these coordinates [see
Eq. (9.189)] as
L = L+ + L ,
(8.219)
where
Eyal Buks
255

L+ =
mx 2 m2
(1 + ) x2 ,
2
2
(8.220)
L =
my 2 m2
(1 ) y 2 .
2
2
(8.221)
and
Thus, the system is composed of

two decoupled harmonic
oscillators with
angular resonance frequencies 1 + (for x ) and 1 (for y ). In
thermal equilibrium according to Eq. (8.160) one has

) 2 *

1 +
x =
,
(8.222)
coth
2
2m 1 +

) 2 *

1
,
(8.223)
y =
coth
2
2m 1
where = 1/kB T . Moreover, due to symmetry, the following holds
x = y = 0 ,
x y = 0 .
(8.224)
(8.225)
With the help of the inverse transformation, which is given by

x + y
,
2
x y
y=
,
2
one thus finds
(8.226)
x=
(8.227)
x = 0 ,
(8.228)
and
) 2 * 1 ) 2
*
x =
x + y2
2

1+
1
coth
coth
2
2

.
=
+
4m
1+
1
(8.229)
13. Using Eq. (5.152), which is given by

x(H) (t) = x(H) (0) cos (t) +
p(H) (0)
sin (t) ,
m
(8.230)
one finds that

'
( sin (t) )
*
G (t) = cos (t) x2(H) (0) +
p(H) (0) x(H) (0) .
m
Eyal Buks
(8.231)
256
8.4. Solutions
Using the relations
0

x=
a + a ,
2m
0

m
p=i
a + a ,
2
, a, a = 1 ,
one finds that

2 2
a + a
x2 =
+ 2a a + 1 ,
2m

px
2 2
=i
a + a
1 .
m
2m
a) Thus, for the ground state [see Eqs. (5.28) and (5.29)]
G (t) =

[cos (t) i sin (t)] =
exp (it) .
2m
2m
(8.232)
(8.233)
(8.234)
(8.235)
(8.236)
(8.237)
b) The density operator is given by Eq. (8.166)

=
n

1
N
|n n| ,
N + 1 n=0 N + 1
(8.238)
where
N = Tr (N ) =
e
,
1 e
(8.239)
N = a a, and where = 1/kB T . Using the fact that ' is diagonal

) *
2 (
in the basis of number states one finds that a2 = a
= 0.
Combining all these results leads to

G (t) =
[(2 N + 1) cos (t) i sin (t)]
2m

=
coth
cos (t) i sin (t) .
2m
2
(8.240)
14. The wave function of the coherent state | is given by Eq. (5.51)
(x ) = x |

2

x x 2
2 m 1/4
x
= exp
exp
+ i p
.
4

2x

(8.241)
where
Eyal Buks
257

0
2
,
m
p = | p | = 2m ,
= Re () ,
= Im () ,
0
.

2
x = | (x) | =
,
2m
Using the definition (8.63) and the identity
x = | x | =

one has
W (x , p ) =
=
(8.243)
(8.244)
(8.245)
(8.246)
2
14 4ca+b
a
e
,
a
(8.247)
5
6

1
px
x
x
exp i
x
|
x +
dx
2

2
2
m 1/2
dx
2
!
(8.242)
x x x
2
2x
e
m 1/2
"2 !
"2
+i
p p
dx
(x x )2 +( x2
2(x )2
x + x x
2
2x
+i
p p
,
(8.248)
thus
1 12
W (x , p ) = e
x x
x
2
12
p p
p
2
(8.249)
where [see Eq. (5.49)]

0
.
m

2
p = | (p) | =
=
.
(8.250)
2
2x
*
15. At time t > 0 the system is in a coherent state
= 0 eit , where [see
Eqs. (5.53) and (5.240)]
0
m
0 = x
.
(8.251)
2
Thus, with the help of Eq. (8.249) one finds that the Wigner function of
the system at time t is given by
Eyal Buks
258
8.4. Solutions

1 12
W (x , p ) = e
x x
x
2
12
p p
p
2
(8.252)
where
x = x cos (t) ,
p = mx sin (t) ,
0

x =
,
2m
0
m
.
p =
2
(8.253)
(8.254)
(8.255)
(8.256)

=
d2 | | P () ,
(8.257)

complex plane,
!
"
1
||2
P () =
exp
,
(8.258)
N
N
and where
N =
e
1 e
(8.259)
is the expectation value of the number operator N. Thus

W (x , p ) =
d2 P () W (x , p ) .
(8.260)
where
1
W (x , p ) =
2
p x
exp i
5
6
x
x
x
|
x +
dx ,
2
2
(8.261)
which is the Wigner function of a coherent state |, was found to be

given by [see Eq. (8.249)]
1 12
W (x , p ) = e
x x
x
2
12
p p
p
2
(8.262)
where
Eyal Buks
259

2
= 2x ,
m
= | p | = 2m = 2p ,
= Re () ,
= Im () ,
0
.

2
,
= | (x) | =
2m
0
.
m

2
= | (p) | =
=
.
2
2x
x = | x | =
p
x
p
(8.263)
(8.264)
(8.265)
(8.266)
(8.267)
(8.268)
Thus W (x , p ) is given by

2

2

x x
p p
2
12
12
1
x
p
||

2
N

e
W (x , p ) = 2
d e
N

2
2
2
2
1 x 2x
1
1 p 2x
N 2
x
p
d e
= 2
d e N 2
.
N
(8.269)
With the help of the identity (5.136) one thus finds that
1
12 2N +1
1
1
W (x , p ) =
e
2 N + 1
2 2
p
+ p
x
x
(8.270)
where
2 N + 1 = 1 + 2
e
= coth
1 e

2
(8.271)
and where = 1/kB T .

17. With the help of Eq. (5.123) one finds that the wavefunction of the number state |n = 1 is given by
2
x
x
2 x0 exp 2x
2
0
n=1 (x ) = x |n = 1 =
,
(8.272)
1/2
1/4 x0
where
x0 =

.
m
(8.273)
thus
W (x , p ) =
Eyal Buks
ei
p x
x x2
x0
(x x2 )
x + x2
x0
1/2 x0
2x2
0
(x + x2 )
2x2
0
dx ,
260
8.4. Solutions
(8.274)
or
W (x , p ) =
2
xx
1
1/2
!
x
x0
2
X2
"
ip
e p0 X
X2
4
dX , (8.275)
where X = x /x0 and where p0 = /x0 . The integration, which is performed with the help of Eq. (5.136), yields

2

2
2 xx0 2 pp0 2
x
p
1

W (x , p ) = e
.
(8.276)
+
x0
p0
2
Note that near the origin the Wigner function W (x , p ) becomes negative.
18. The relation (8.64) is proven by

1

dp W (x , p )
6
5

p x
x
1
=

x +
dx
dp ei
x
2
2
2

(x )
= x | |x .
(8.277)
With the help of the identities (3.45) and (3.52) W (x , p ) can be expressed as
W (x , p )

p x +p (x x )p (x + x )
2
2
1
i

=
dx
dp
dp e
p | |p
(2)2

p
p

x p
x (p p )
2
2
1
1
i

dp
dp e
p | |p
dx e
.
=
2
2

p p2 p2
(8.278)
The above result easily leads to (8.65)
Eyal Buks
261

1
dx W (x , p )

p p
dp
dp p | |p p
=
2
2

x (p p )
1

dx ei
2

(p p )
= p | |p .
(8.279)
19. For a pure state = | | one finds that the Wigner function is given
by [see Eq. (8.63)]
1
W (x , p ) =
2
dx ei
p x

x
x
x
x +
, (8.280)
2
2
where (x ) = x | is the position wavefunction of |. Thus with

the help of Schwartz inequality one finds that

2
1
x
2

i p x
|W (x , p )|
e
dx
(2)2

2
dx
x +
,
2
(8.281)
thus
|W (x , p )|
1
.
2
(8.282)
20. The Hamiltonian operator of the system is given by

H=
p2
+ V (x) ,
2m
(8.283)
where p is the canonical conjugate operator to the position operator x.

With the help of Eq. (8.35), which reads
d
1
= [, H] ,
dt
i
(8.284)
one finds that
Eyal Buks
262
8.4. Solutions
dW
1
=
dt
2
dx e
ip x
= Sk + Sp ,
5
6
x
x
x
x
+
2
dt
2
(8.285)
where the term

1
Sk =
2i
dx e
ip x
6
x
p2
x
x
,
x +
2
2m
(8.286)
represents the contribution of the kinetic energy, and where the term
1
Sp =
2i
dx e
ip x
5
6
x
x
x +
x
[,
V
(x)]
(8.287)
represents the contribution of the potential energy. To evaluate the term

Sk , the closure relation (2.23) is employed

1=
|n n | ,
(8.288)
n
where {|n } is an arbitrary orthonormal basis, in order to obtain the

following relation

5

6
p2
x
x
x
,
x +
2
2m
6
5
1 x
x
2

=
|n n | p
x +
x
2m n
2
5
6

1
x
2
x
x
p |n n |
x +
.
2m n
2
(8.289)
By introducing the variables

x
,
2
x
,
x = x +
2
and employing the relations [see Eq. (3.29)]
5
x
2
d2
x
p |n = 2 2 x |n ,
2
dx
x
d2
n | p2
x +
= 2 2 x |n ,
2
dx
x = x
Eyal Buks
(8.290)
(8.291)
(8.292)
(8.293)
263

<
one finds that (after removing the factor n |n n |)

5

6
x
p2
x
,
x +
2
2m
2

2
d2
d2
=
x | |x .
2m dx2 dx2
(8.294)
Using the relations

d
x d
x d
1 d
d
=
+
=
,
dx
x dx
x dx
2 dx
dx
d
x d
x d
1 d
d
=
+ =
+ ,
dx
x dx
x dx
2 dx
dx
one finds that
d2
d2
d2
=
2
,
dx2 dx2
dx dx
and thus

5

6
x
p2
x
,
x
+
2
2m
5
6

2 d2
x

x + x
=
x
.
m dx dx
2

(8.295)
(8.296)
(8.297)
(8.298)
Inserting into Eq. (8.286) and integrating by parts with respect to x

lead to
5
6

d2

x
x
ip x
Sk =
dx e
x
x +
2im
dx dx
2

p 1
=
m 2
dx e
ip x
d
dx
6
5
x
x +
,
x
2

(8.299)
or [see Eq. (8.63)]

p W
.
(8.300)
m x
To evaluate Sp the following relation is employed
5
6
x
x
x +
x
[,
V
(x)]

5
6
x
x
x
x
= V x +
V x
x
x
+
.
2
2
2

2
(8.301)
Sk =
Eyal Buks
264
8.4. Solutions
The term V (x + x /2)V (x x /2), which represents an odd function
of x , can be Taylor expanded as
V

2l+1
<
x
x
1
2l+1 V
x
V x
=2
x +
.
2l+1
2
2
2
l=0 (2l + 1)! (x )
(8.302)
As can be seen from Eq. (8.63), the following holds

1
2
dx (x )
2l+1
ip x
5
6 2l+1 2l+1
x
W

x
x
+
=
.
2

2
i
(p )2l+1
(8.303)
Employing the above results to evaluate Eq. (8.287) and separating the
first term from all higher order terms yield
2l
<
V W
2l+1 V 2l+1 W
2i
Sp =
+
.
2l+1
x p
(p )2l+1
l=1 (2l + 1)! (x )
(8.304)
Combining Eqs. (8.300) and (8.304) yields
2l
p W V W <
2l+1 V 2l+1 W
dW
2i
=
+
+
, (8.305)
dt
m x x p l=1 (2l + 1)! (x )2l+1 (p )2l+1
or in terms of the Poissons brackets [see Eqs. (1.37) and (8.283)]

2l
<
dW
2l+1 V 2l+1 W
2i
= {H, W } +
.
2l+1
dt
(p )2l+1
l=1 (2l + 1)! (x )
(8.306)
Note that when 2l+1 V/ (x )2l+1 = 0 for l 1 the above result coincides with the classical equation of motion dW/dt = {H, W }. Thus one
concludes that the quantum time evolution of W of a harmonic oscillator
is identical to the classical one.
21. With the help of Eq. (2.175) one finds that
exp (iX iP ) = e
i
2
eiP eiX .
(8.307)
By evaluating the trace in Eq. (8.68) using an orthonormal basis of dimensionless position eigenstates (i.e. X |X = X |X ) one finds that
[see Eq. (3.19) and recall that Tr (AB) = Tr (BA)]
Eyal Buks
265

i
2
(, ) = e
W
= e
i
2
dX X | eiP eiX |X
dX eiX X | |X +
dX eiX
'

(
X

X +
,
2
2
(8.308)
thus
W (X , P ) =
1
(2)2
1
2

(, ) eiX +iP
dd W
dX
'

( iP
e
X

X +
2
2
1
dei(X X )
2

1
=
2
(X X )
dX
5
6

X
X
X +
eiX P .
2

(8.309)
22. For the case X = P = 0 the operator is given by

1
(0, 0) =
(2)2

dd eiXiP ,
(8.310)
or [see Eq. (2.175) and recall that [X, P ] = i]

(0, 0) =
1
2
(2)

dd e
i
2
eiP eiX ,
(8.311)
and thus the following holds
Eyal Buks
266
8.4. Solutions
(0, 0) |X =
1
(2)2
1
=
2

dd e
iP
d e
= 1 e2iX
i
2
eiP eiX |X
1
|X
d ei( 2 +X )
2

( 2 +X )
|X ,
(8.312)
or [see Eq. (3.19)]

(0, 0) |X = 1 |X ,
(8.313)
which implies that (0, 0) is related to the parity operator P (8.75) by

(0, 0) = 1 P .
(8.314)
By employing the relations

X=
a + a
a a
, P =
,
2
2i
(8.315)
together with Eq. (8.310) one finds that

i a+a
i aa
2
2i
P=
dd e
4
=
1
4

i+
a i
a
2
2
dd e
(8.316)
The above result together with Eq. (5.38) yield

X + iP
X + iP
1 D
PD
2
2

+iP
iP
1
i+
a X
i
a X
2
2
2
2
=
dd
e
(2)2
=
Eyal Buks
1
(2)2
1
(2)2

i+
2
dd e
dd ei(X
X+iP
+iP
X
2
X )+i (P P )
i
2
XiP
iP
(8.317)
267

thus [see Eq. (8.72)]

X + iP
X + iP
1
D
PD
= (X , P ) .
2
2
(8.318)
23. The normalized homodyne observable X can be expressed in terms of

the dimensionless position and momentum operators X and P , which are
given by
X=
a + a
a a
, P = ,
2
i 2
(8.319)
and which satisfy [X, P ] = i [see Eq. (5.13)], as

X iP i X + iP i
e +
e
2
2
= X cos + P sin .
X =
(8.320)
The associated dimensionless momentum operator P is defined as P =
X sin + P cos , i.e.

X
cos sin
X
=
,
(8.321)
P
sin cos
P
and the inverse transformation is given by

X
cos sin
X
=
.
P
sin cos
P
a) The expectation value
) iX *
e
= Tr [exp (iX ) ]
(8.322)
(8.323)
is the
function of the probability distribution
function
characteristic

Pr X of X , which is denoted below as w X , and thus it is

related
() of the probability distribution
to the Fourier transform w
w X by
) iX *
e
=

dX w X eiX = w
() .
(8.324)
(, ) of the Wigner
On the other hand, the Fourier transform W
function W (X , P ) is given by [see Eq. (8.68)]

(, ) = Tr [exp (iX iP ) ] .
W
Eyal Buks
(8.325)
268
8.4. Solutions
The comparison between Eq. (8.323) and Eq. (8.325) yields the following relation (recall that X = X cos + P sin )
( cos , sin ) = w
W
() .
(8.326)
Applying the inverse Fourier transform to Eq. (8.324) leads to [see

Eq. (8.67)]

1
w X =
d eiX w
()
2
=
1
2
( cos , sin ) ,
d eiX W
(8.327)
and thus by employing the inverse Fourier transform to Eq. (8.67)

, =
W

dX dP ei X
i P
W (X , P ) ,
(8.328)
one finds that

1
w X =
2

ddX dP ei (X
cos +P sin X
)
W (X , P ) .
(8.329)
The variable transformation (8.322) leads to

dX dP W X cos P sin , X sin + P cos
w X =
d ei (X X )
2

X
(X
)

=
dP W X cos P sin , X sin + P cos .
(8.330)
b) The relation (8.326) allows evaluating the Wigner function, which is
(, ) by Eq. (8.67),
related to its Fourier transformed function W
using the so-called inverse Radon transform. In polar coordinates
(8.67) becomes
Eyal Buks
269
W (X , P ) =
1
(2)2

0
( cos , sin ) ei (X
d || W
cos +P sin )
(8.331)
W (X , P ) =
1
2
(2)

0
d || w
() ei (X
cos +P sin )
(8.332)

1
W (X , P ) =
(2)2

0

dX || w X ei (X cos +P sin X ) .

=
d2 | | P () ,
(8.333)
(8.334)

complex plane,
!
"
1
||2
P () =
exp
,
(8.335)
N
N
and where
N =
e
1 e
(8.336)
is the expectation value of the number operator N . By employing the

expression for the wave function (x ) = x | of a coherent state
(x ) = x |

2

2
x x
2 m 1/4
x
= exp
exp
+ i p
,
4

2x
(8.337)
where
Eyal Buks
270
8.4. Solutions
0
2
,
(8.338)
m
(8.339)
p = | p | = 2m ,
= Re () ,
(8.340)
= Im () ,
(8.341)
0
.

2
x = | (x) | =
,
(8.342)
2m
one finds that

x | |x =
d2 P () x | |x
!
"
m 1/2
||2
2

=
d exp
N
N

2
x x 2
x x
(x x )
+ i p
exp
2x
2x

m 1/2

2 + 2
=
d2 exp
N
N

2
2
X 2
X 2
exp
+ i (X X )
2
2
m 1/2

2 N + 1 2
X 2 + X 2

=
d exp
+ (X + X )
N
N
4

2
d exp
+ i (X X ) .
N
(8.343)
x = | x | =
where
2m
x ,

0
2m
X =
x .

With the help of the identity (5.136) one finds that
X =
x | |x =
m 1/2
(X
1
e
2 N + 1
(8.344)
(8.345)
2
2 +X 2 N (X X )2 + N (X +X )
2N +1
4
(8.346)
or using the identity

X 2 + X 2 =
Eyal Buks
(X + X )2 + (X X )2
,
2
(8.347)
271

one has
x | |x =
m 1/2
e
2 N + 1

X +X 2
2
2(2N +1)

2 2N +1
X X
2
2
(8.348)
In terms of x and x this result can be written as

1
x | |x = e

x +x
2
2
(2N+1)2
x x
2
2
(8.349)
where
=

(2 N + 1) ,
m
(8.350)
e
2 N + 1 = 1 + 2
= coth
1 e

2
(8.351)
and where = 1/kB T . Alternatively, the result can be expressed as

x | |x =
2
2
+x
x
tanh( 2 ) x 2x
coth( 2 ) x 2x
0

x0 coth
2
(8.352)
where
x0 =

.
m
(8.353)
25. The following holds [see Eqs. (8.8), (8.42)]

) iX * Tr eH eiX
e
=
,
(8.354)
Tr (eH )

where H = a a + 1/2 [see
, Eq.- (5.16)]. Using the identity (2.173)
and the commutation relation a, a = 1 [see Eq. (5.13)] one finds that
the following holds
= aei ,
(8.355)
= a ei ,
(8.356)
eia a aeia
eia a a eia
and thus X can be expressed as
X = eia a X0 eia
(8.357)
where X0 , which is given by

Eyal Buks
272
8.4. Solutions
X0 =
a + a
,
2
(8.358)
is related to the position operator x by [see Eq. (5.11)]

x = x0 X0 ,
(8.359)
where
x0 =

.
m
(8.360)
The last result implies that
eiX = eia a eiX0 eia
(8.361)
and thus [see Eq. (8.354) and recall that Tr (AB) = Tr (BA)]
) iX * ) iX0 * ' i m x (

= e
= e
.
e
(8.362)
) iX *
In other words,
e
is found to be independent on . The expecta'
m (
i
x

tion value e
can be calculated by employing the expression
for the matrix elements x | |x of the density operator given by Eq.
(8.352)
) iX *
e
=
=
dx x | ei
dx ei
m
m
|x
2
tanh( 2 ) xx
0

,

x0 coth 2
(8.363)
and where = 1/kB T . Using the the identity (5.136) one finds that
2 coth( )
) iX *
2
4
e
=e
.
In view of Eq. (8.160), according to which

1

1 ) *
coth
= N + = X02 ,
2
2
2
(8.364)
(8.365)

N =
Eyal Buks
e
,
1 e
(8.366)
273

the above result can be rewritten as
2
2 X0

) iX *
e
= e 2
.
(8.367)
) iX *
The factor e
is thecharacteristic function of the probability

distribution function Pr X of X , which is denoted below as w X ,
and thus it is related
() of the probability
to the Fourier transform w
distribution w X by [see Eq. (8.324)]

) iX *
=
e

dX w X eiX = w
() .
(8.368)
With the help of Eqs. (8.324), (8.326) and (8.367) one finds that the
(, ) of the Wigner function W (X , P ) satisfy the
Fourier transform W
following relation for any real
( cos , sin ) = e
W

2
2 X0
2
(8.369)
and thus
(, ) = e
W
(2 +2 )X02
2
(8.370)
The inverse Fourier transformation [see Eqs. (5.136) and (8.309)] yields

2
2
2 X0
2 X0
iX 1
iP
1
2
2
W (X , P ) =
d e
e
d e
e
2
2
2
X +P
1
2
2X0
e
2 X02
X 2 +P 2
1
=
e 2N +1 .
(2 N + 1)
(8.371)
It is easy to see that the above result for the Wigner function W (X , P )
for the dimensionless variables X and P is consistent with Eq. (8.270)
for the Wigner function W (x , p ) for the displacement x and momentum
p variables.
26. By using Eqs. (8.67) and employing cylindrical coordinates
= cos , = sin ,
X = R cos , P = R sin ,
(8.372)
one finds that
Eyal Buks
274
8.4. Solutions
W (X , P ) =
1
(2)2
( cos , sin ) eiR

d W
cos()
(8.373)

thus since w X is independent [see Eqs. (8.324) and (8.326)] one
has [note that contrary to Eq. (8.333), integration over below is taken
to be over positive values only]
1
W (X , P ) =
(2)2
d w
()
d eiR
cos()
(8.374)
where
w
() =

dX w X eiX .
(8.375)
With the help of Jacobi-Anger expansion
exp (iz cos x) =
in Jn (z) einx ,
(8.376)
n=
one finds that

iR cos()
d e
in
i Jn (R ) e
n=
d ein
= 2J0 (R ) ,
(8.377)
thus
1
W (X , P ) =
2

0

d w
() J0 X 2 + P 2 ,
(8.378)
or
W (X , P ) =
/
0
dz w
z
X 2 +P 2
zJ0 (z)
2 (X 2 + P 2 )
(8.379)
As an example of the usage of Eq. (8.378), consider the case of a harmonic

oscillator having angular resonance frequency in thermal equilibrium
at temperature T . For this case [see Eq. (8.367)]
Eyal Buks
275

w
() = e

2
2 X0
2
(8.380)
and thus with the help of the identity

ze
z2 A2
2
J0 (zR) dz =
1 R22
e 2A ,
A2
(8.381)
one finds that Eq. (8.378) yields

W (X , P ) =
X +P
1
2X 2
0
e
,
2 X02
(8.382)

27. The density operator evolves in time according to [see Eq. (8.37)]
(t) = u (t) 0 u (t) ,
(8.383)
where u (t) is the time evolution operator for the system. The Wigner
function W (X , P ; t) can be expressed according to Eq. (8.333) as
W (X , P ; t) =
1
(2)

0

dX || w X ei (X cos +P sin X ) ,
(8.384)

where w X is the probability distribution function of the observable
X , which is given by
X =
a ei + aei
.
2
(8.385)
In terms of the density operator (t) one has [recall that Tr (AB) =
Tr (BA)]

* )

w X = Tr
X X
(t)
*)

= Tr u (t)
X X
u (t) 0 ,
(8.386)
where
X is an eigenvector of X with eigenvalue X , i.e. X
X =
X
X . With the help of Eqs. (2.173) and (4.9) one finds that
Eyal Buks
276
8.4. Solutions
+ aei i(a a+ 12 )t
e
2
i
a e
+ aei ita a
= eita a
e
2
a ei(+t) + aei(+t)
=
2
= X+t ,
u (t) X u (t) = ei(a
i
a+ 12 )t a e
(8.387)
thus the following holds
*
X+t u (t)
X = X u (t)
X ,
(8.388)
i.e. u (t)
X is an eigenvector of X+t with eigenvalue X . This eigen

vector is labeled below as
X+t
. Using these results and the trigonometric identities
cos (x + y) = cos x cos y sin x sin y ,
sin (x + y) = cos x sin y + sin x cos y ,
(8.389)
(8.390)
the Wigner function W (X , P ; t) becomes

1
W (X , P ; t) =
d d
dX ||
(2)2

*)
Tr
X+t X+t
0 ei (X cos +P sin X )

1
=
d d
dX ||
(2)2
0

* )
i (X cos( t)+P sin( t)X )
Tr
X X
0 e
,
(8.391)
thus
W (X , P ; t) = W (X cos (t) P sin (t) , X sin (t) + P cos (t) ; 0) ,

(8.392)
i.e. the time evolution of the Wigner function represents rigid rotation in
phase space at angular velocity .
28. The Wigner function W (X , P ) can be expressed as [see Eq. (8.67)]
1
W (X , P ) =
(2)2
Eyal Buks

(, ) eiX +iP ,
dd W
(8.393)
277

where
(, ) = Tr (D () ) ,
W
(8.394)
and where
+ i
=
,
2i
(8.395)
thus for the displaced system one has [see Eq. (5.41) and recall that
Tr (AB) = Tr (BA)]

W (X , P ) =
dd Tr D () D ( ) D ( ) eiX +iP
2
(2)
=
1
(2)2

dd e
Tr (D () ) eiX
+iP
(8.396)
thus using
e
iX +iP

i X +
+i P
2
2i
=e
(8.397)
one finds that

W (X , P ) = W (X X , P P ) ,
(8.398)
where
+
,
(8.399)
2

.
(8.400)
P =
i 2
29. In general, the convolution theorem states that

1
iX P
dX f (X ) g (X ) e
=
dP F (P ) G (P P ) ,
2
X =
(8.401)
where F (P ) (G (P )) is the Fourier transform of f (X ) (g (X )), i.e.

1
F (P ) =
dX f (X ) eiX P ,
(8.402)
2
G (P ) =
Eyal Buks
1
2
dX g (X ) eiX
(8.403)
278
8.4. Solutions
thus with the help of the identity
P 2
e( )
1
=
2

2
X
iX P
2
dX e
(8.404)
one finds that the reduced Wigner function WR (X , P ) is related to

W (X , P ) by [see Eq. (8.82)]
WR (X , P ) =
P P 2
e(
dP W (X , P )
(8.405)
As an example, for the case where W (X , P ) is normally distributed

according to

2
2
exp X +P
2
,
(8.406)
W (X , P ) =
2
where is a constant, Eq. (8.405) yields
WR (X , P ) =
2
X
P
2 +2
2
e
.
2
+ 2
(8.407)
30. The dynamics along the x direction is governed by the Hamiltonian Hx

of a harmonic oscillator
Hx =
p2x
1
+ m 2 x2 .
2m 2
(8.408)
) *
By symmetry x = 0. The expectation value x2 was calculated in Eq.
(8.160) and found to be given by

) 2*
1

x =
coth
.
(8.409)
m 2 2
2
31. In general, for any smooth function f () of the following holds

f ( (t)) = u (t, t0 ) f ( (t0 )) u (t, t0 ) .
(8.410)
This can be shown by Taylor expanding the function f () as a power

series
f ( (t)) =
an ( (t))n ,
(8.411)
n=0
using Eq. (8.37) and the fact that u (t, t0 ) u (t, t0 ) = 1, i.e. the unitarity
of the time evolution operator. By using this result for the function log
Eyal Buks
279

together with the general identity Tr (XY ) = Tr (Y X) [see Eq. (2.129)]
one easily finds that is time independent. This somewhat surprising
result can be attributed to the fact that the unitary time evolution that is
governed by the Schrdinger equation is symmetric under time reversal.
In the language of statistical mechanics it corresponds to a reversible
process, for which entropy is preserved.
32. Using the definition of the Pauli matrices (6.136) one finds that

1
1 + kz kx iky
=
,
(8.412)
2 kx + iky 1 kz
and
2 =
1
4
1 + 2kz + k2 2 (kx iky )

2 (kx + iky ) 1 2kz + k2
where k2 = kx2 + ky2 + kz2 .

a) Note that for any k the following holds Tr () = 1. The requirement
that is Hermitian, i.e. the requirement
that = , implies that

kz = kz and kx kx + i ky ky = 0, thus kx , ky and kz are all real.

Moreover, the requirement that Tr 2 = 1 + k2 /2 1 implies
that k2 1.

b) For this case Tr 2 = 1, thus k2 = 1.
c) With the help of Eq. (6.137), which is given by
( a) ( b) = a b + i (a b) ,
(8.413)
and the fact that all three Pauli matrices have a vanishing trace, one
finds that
1
1
Tr (
u ) = Tr (
u ) + Tr ((
u ) (k ))
2
2
1
= Tr ((
u ) (k ))
2
1
i
= Tr (
u k) + Tr ( (
u k))
2
2
1
= Tr (
u k)
2
=u
k .
(8.414)
33. Using the definition of the Pauli matrices (6.136) one finds that

1
1 + kz kx iky
.
(8.415)
=
2 kx + iky 1 kz
a) Let be the two eigenvalues of . The following holds
Tr () = + + = 1 ,
Eyal Buks
(8.416)
280
8.4. Solutions
and

Det () = + = 1 k2 /4 ,
(8.417)
where k2 = kx2 + ky2 + kz2 . Thus

=
1 |k|
,
2
(8.418)
and therefore
= f (|k|) ,
(8.419)
where
f (x) =
1x
1x 1+x
1+x
log
log
.
2
2
2
2
(8.420)
0.7
0.6
0.5
0.4
y
0.3
0.2
0.1
0.2
0.4
The function f (x) =
1x
2
0.6
log
0.8
1x
2
1+x
2
log 1+x
2 .
b) As can be seen from Eq. (8.194), after the measurement becomes

diagonal in the basis of eigenvectors of the measured observable,
namely, after the measurement the density matrix is given by

1 1 + kz 0
c =
,
(8.421)
0 1 kz
2
and thus the entropy after the measurement is
c = f (kz ) =
1 kz
1 kz
1 + kz
1 + kz
log
log
.
2
2
2
2
(8.422)
34. First, consider a general functional g () of the density operator having

the form
g () = Tr (f ()) ,
Eyal Buks
(8.423)
281

where the function f () can be Taylor expanded as a power series
f () =
ak k ,
(8.424)
k=0
and where ak are complex constants. Consider an infinitesimal change in

the density operator + d. To first order in d the corresponding
change dg in the functional g () can be expressed as
dg = g ( + d) g ()
!
"
4
3
k
k
= Tr
ak ( + d)
= Tr
k=0

k=0

2
.
ak k1 d + k2 (d) + k3 (d) 2 + + O (d)

k terms
(8.425)
By exploiting the general identity Tr (XY ) = Tr (Y X) the above result

can be simplified (note that generally needs not to commute with d)
!
"

dg = Tr
ak kk1 d + O (d)2 ,
(8.426)
k=0
thus to first order in d the following holds

df
dg = Tr
d .
d
(8.427)
In the above expression the term df /d is calculated by simply taking the

derivative of the function f (x) (where x is considered to be a number)
and substituting x = . Alternatively, the change dg can be expressed in
terms of the infinitesimal change dnm in the matrix elements nm of .
To first order in the infinitesimal variables dnm one has
dg =
g
d
.
nm nm
n,m
(8.428)
It is convenient to rewrite the above expression as

d ,
dg = g
(8.429)
and of d are given

where the vector elements of the nabla operator
by

n,m
Eyal Buks
,
nm
(8.430)
282
8.4. Solutions
and

d n,m = dnm .
(8.431)
Thus, to first order one has

d ,
d =
(8.432)
l d ,
dgl = g
(8.433)
and
where l = 0, 1, 2, ...L.
a) In general, the technique of Lagrange multipliers is very useful for
finding stationary points of a function, when constrains are applied.
A stationary point of occurs iff for every small change d, which is
0 , g
1 , g
2 , ..., g
L (i.e. a change which
orthogonal to all vectors g
does not violate the constrains) one has
d .
0 = d =
(8.434)
This condition is fulfilled only when

the vector belongs to the
L . In other
subspace spanned by the vectors g0 , g1 , g2 , ..., g

words, only when
= 0 g
0 + 1 g
1 + 2 g
2 + ... + L g
L,
(8.435)
where the numbers 0 , 1 , ..., L , which are called Lagrange multipliers, are constants. By multiplying by d the last result becomes
d = 0 dg0 + 1 dg1 + 2 dg2 + ... + L dgL .
Using Eqs. (8.427), (8.89, (8.91) and (8.92) one finds that
d = Tr ((1 + log ) d) ,
dg0 = Tr (d) ,
dgl = Tr (Xl d) ,
thus
!
"
L

0 = Tr
1 + log + 0 +
l Xl d .
(8.436)
(8.437)
(8.438)
(8.439)
(8.440)
l=1
The requirement that the last identity holds for any d implies that
1 + log + 0 +
L

l Xl = 0 ,
(8.441)
l=1
thus
Eyal Buks
283

10
=e
exp
L

l Xl
l=1
"
(8.442)
The Lagrange multipliers 0 , 1 , ..., L can be determined from Eqs.

(8.91) and (8.92). The first constrain (8.91) is satisfy by replacing
the factor e10 by the inverse of the partition function Z
"
! L

1
= exp
l Xl .
(8.443)
Z
l=1
where
!
Z = Tr
L

l=1
l Xl
"
(8.444)
As can be seen from the above expression for Z, the following holds
Xl =

log Z
1
Tr Xl eH =
.
Z
l
(8.445)
b) For the case of a microcanonical ensemble Eq. (8.443) yields = 1/Z,

i.e. is proportional to the identity operator.
c) For the case of a canonical ensemble Eq. (8.443) yields
c =
1 H
e
,
Zc
where the canonical partition function Zc is given by

Zc = Tr eH ,
(8.446)
(8.447)
and where labels the Lagrange multiplier associated with the given
expectation value H. By solving Eq. (8.92), which for this case is
given by [see also Eq. (8.445)]
H =

log Zc
1
Tr HeH =
.
Zc
(8.448)
the Lagrange multiplier can be determined. Note that the temperature T is defined by the relation = 1/kB T , where kB is the
Boltzmanns constant.
d) For the case of a grandcanonical ensemble Eq. (8.443) yields
gc =
1 H+N
e
,
Zgc
(8.449)
where the grandcanonical partition function Zgc is given by

Eyal Buks
284
8.4. Solutions

Zgc = Tr eH+N .
(8.450)
Here the Lagrange multiplier associated with the given expectation

value N is given by , where is known as the chemical potential. The average
H is given by
energy

H = Tr Hgc

Tr He(HN)

=
Tr e(HN)

Tr (H N ) e(HN)
Tr N eH+N

=
+
,
Tr (eH+N )
Tr e(HN)
thus
H =
log Zgc
log Zgc
(8.451)
Similarly, the average number of particles N is given by

Tr N eH+N

N = Tr Ngc =
.
Tr (eH+N )
(8.452)
In terms of the fugacity , which is defined by

= e ,
(8.453)
N can be expressed as
N =
log Zgc
.
(8.454)
35. The Hamiltonian can be expressed as a function of the operators p and

x as
H (p, x) =
p2
+ V (x) .
2m
(8.455)
Evaluating Zc according to Eq. (8.447) by tracing over momentum states

yields

Zc = Tr eH

= dp p | eH |p

= dx dp p |x x | eH |p .
(8.456)
In the classical limit the parameter , which is inversely proportional to

the temperature, is considered as small. Using Eq. (12.120) from chapter
12, which states that for general operators A and B the following holds
Eyal Buks
285

e(A+B) = eA eB + O 2 ,
(8.457)
one finds that

p2
eH = eV (x) e 2m + O 2 .
(8.458)
This result together with Eq. (3.52), which is given by

1
ip x
x |p =
exp
,

2
yield in the classical limit

p2
Zc = dx dp p |x x | eV (x) e 2m |p

p2
= dx
dp eV (x ) e 2m p |x x |p

1
dx dp eH(p ,x ) .
=
2
(8.459)
(8.460)
Note that the this result can be also obtained by taking the limit 0,
for which the operator x and p can be considered as commuting operators
(recall that [x, p] = i), and consequently in this limit eH can be
factored in the same way [see Eq. (8.458)].
36. The measurement of the observable A1 is describe by the its extension,
which is given by A1 12 , where 12 is the identity operator on subsystem
2. Thus with the help of Eq. (8.8) one finds that
A1 = Tr (A1 12 )

=
n1 , n2 | A1 12 |n1 , n2
n1 ,n2

n1
1 n1 |
!

n2
2 n2 | |n2 2
"
A1 |n1 1
= Tr1 (1 A1 ) .
(8.461)
37. In terms of the matrix elements n1 ,n2 ,m1 ,m2 of the operator , which are
given by
(n1 ,n2 ),(m1 ,m2 ) = n1 , n2 | |m1 , m2 ,
(8.462)
the matrix elements of 1 and 2 are given by

(1 )n1 ,m1 =
(n1 ,n2 ),(m1 ,n2 ) ,
(8.463)
n2
Eyal Buks
286
8.4. Solutions
and
(2 )n2 ,m2 =
(n1 ,n2 ),(n1 ,m2 ) .
(8.464)
n1
In general is Hermitian, i.e.

(n1 ,n2 ),(m1 ,n2 ) = (m1 ,n2 ),(n1 ,n2 ) ,
(8.465)
and therefore

(n1 ,n2 ),(m1 ,n2 )
(1 )n1 ,m1 =
n2
(m1 ,n2 ),(n1 ,n2 )
n2
= (1 )m1 ,n1 ,
(8.466)
i.e. 1 is also Hermitian, and similarly 2 is also Hermitian. Thus the
eigenvalues of 1 and 2 are all real. Moreover, these eigenvalues represent
probabilities, and therefore they are expected to be all nonnegative and
smaller than unity. In what follows it is assumed that the set of vectors
{|n1 1 } ({|n2 2 }) are chosen to be eigenvectors of the operator 1 (2 ).
Thus 1 and 2 can be expressed as

1 =
wn(1)
|n1 1 1 n1 | ,
(8.467)
1
n1
and
2 =

n2
wn(2)
|n2 2
1
2 n2 |
(8.468)
(1)
(2)
where the eigenvalues satisfy 0 wn1 1 and 0 wn1 1. Similarly,

can be diagonalized as

=
wk |k k| ,
(8.469)
k
where 0 wk 1. In terms of these eigenvalues the entropies are given

by

1 = Tr1 (1 log 1 ) =
wn(1)
log wn(1)
,
(8.470)
1
1
n1
2 = Tr2 (2 log 2 ) =
and
Eyal Buks
wn(2)
log wn(2)
,
2
2
(8.471)
n2
287

= Tr ( log ) =
wk log wk .
(8.472)
As can be seen from Eqs. (8.462), (8.463) and (8.467), the following holds
wn(1)
= (1 )n1 ,n1
1

=
(n1 ,n2 ),(n1 ,n2 )
n2

n2
n1 , n2 | |n1 , n2

n2
n1 , n2 |k wk k |n1 , n2 ,

(8.473)
thus
wn(1)
=
1
wn1 ,n2 ,
(8.474)
and similarly

=
wn1 ,n2 ,
wn(2)
2
(8.475)
n2
n1
where
wn1 ,n2 =

k
n1 , n2 |k wk k |n1 , n2 .
(8.476)
Note that
!
"

wn1 ,n2 =
wk k|
|n1 , n2 n1 , n2 | |k
n1 ,n2

k
n1 ,n2
wk k |k ,
(8.477)
thus the normalization condition k |k = 1 together with the requirement that

Tr =
wk = 1 ,
(8.478)
k
imply that

wn1 ,n2 = 1 ,
(8.479)
n1 ,n2
Eyal Buks
288
8.4. Solutions
i.e.
Tr1 1 =
wn(1)
=1,
1
(8.480)
wn(2)
=1.
2
(8.481)
n1
and
Tr2 2 =
n2

(1) (2)
Consider the quantity y wn1 ,n2 /wn1 wn2 , where the function y (x) is
given by
y (x) = x log x x + 1 .
(8.482)
The following holds

dy
= log x ,
dx
(8.483)
1
d2 y
= ,
dx2
x
(8.484)
and
thus the function y (x) has a single stationary point at x = 1, which is a

minima point. Moreover y (1) = 0, thus one concludes that
y (x) 0
(8.485)
(1)
(2)
for x 0. For x = wn1 ,n2 /wn1 wn2 the inequality (8.485) implies that
wn1 ,n2
(1) (2)
wn1 wn2
log
wn1 ,n2
(1) (2)
wn1 wn2
(1)
wn1 ,n2
(1)
(2)
wn1 wn2
+10.
(8.486)
(2)
Multiplying by wn1 wn2 and summing over n1 and n2 yields

wn ,n2
wn1 ,n2 log (1)1 (2)
wn1 ,n2 +
wn(1)
wn(2)
0 , (8.487)
1
2
wn1 wn2 n1 ,n2
n1 ,n2
n1
n2
thus with the help of Eqs. (8.479), (8.480) and (8.481) one finds that

wn ,n2
wn1 ,n2 log (1)1 (2)
0,
(8.488)
wn1 wn2
n1 ,n2
and with the help of Eqs. (8.470) and (8.471) that

1 + 2
wn1 ,n2 log wn1 ,n2 .
(8.489)
n1 ,n2
Eyal Buks
289

Using Eq. (8.476) one obtains

wn1 ,n2 log wn1 ,n2
n1 ,n2

n1 ,n2
|n1 , n2 |k| wk log
|n1 , n2 |k| wk log wk
n1 ,n2 k
!

wn1 ,n2
wk
|n1 , n2 |k| wk log

n1 ,n2
wn1 ,n2
n1 ,n2
+ .
"
|n1 , n2 |k|2 ,

=1
(8.490)
wn1 ,n2 log wn1 ,n2
n1 ,n2
"
wk log wk
thus
wk

n1 ,n2
1
wn1 ,n2
!

n1 ,n2
|n1 , n2 |k| wk log
!

k
wk
|n1 , n2 |k| wk log
wn1 ,n2
"
(8.491)
Furthermore, according to inequality (8.485) the following holds

wk
|n1 , n2 |k|2 wk log
wn1 ,n2
k

wk
wk
=
|n1 , n2 |k|2 wn1 ,n2
log
wn1 ,n2
wn1 ,n2
k

wk
|n1 , n2 |k|2 wn1 ,n2

1 ,
wn1 ,n2
k

|n1 , n2 |k|2 wk wn1 ,n2
|n1 , n2 |k|2
k
=0.
(8.492)
These results together with inequality (8.489) yield
1 + 2 .
Eyal Buks
(8.493)
290
8.4. Solutions
38. With the help of Eq. (8.110), which is given by
=
eH
= p+ |+ +| + p | | ,
Tr (eH )
(8.494)
where the probabilities p+ and p are given by

p =
e
e

2

2
+e

2
(8.495)
= |e| B/me c is the Larmor frequency [see Eq. (4.22)] and where =
1/kB T , one finds that
= p+ log p+ p log p
=
1 tanh
1 tanh
1 + tanh
1 + tanh
2
2
2
2
log
log
.
2
2
2
2
(8.496)

H| Sz |H = ,
2 2
(8.497)
thus
1
pz+ (1 pz+ ) = ,
2
(8.498)
and therefore
pz+ = cos2
.
8
(8.499)
40. The density operator is given by [see Eq. (8.162)]

= 1 e
en |n n| ,
(8.500)
n=0
where = 1/kB T , thus

= Tr ( log )

=
1 e en log 1 e en
n=0

= 1 e log 1 e
en
n=0

+ 1 e
nen .
n=0
Eyal Buks
(8.501)
291

By using the relations
en =
n=0
1
,
1 e
(8.502)
and [see Eq. (8.152)]
n
ne
n=0
1 n
e
=
e
=
2 ,
n=0
(1 e )
(8.503)
one finds that

= log 1 e +
Eyal Buks

.
e 1
(8.504)
292
9. Time Independent Perturbation Theory
Consider a Hamiltonian H0 having a set of eigenenergies {Ek }. Let gk be

the degree of degeneracy of eigenenergy Ek , namely gk is the dimension of
the corresponding eigensubspace , which is denoted by Fk . The set {|k, i}
of eigenvectors of H0 is assumed to form an orthonormal basis for the vector
space, namely
H0 |k, i = Ek |k, i ,
(9.1)
k , i |k, i = kk ii .
(9.2)
and
For a given k the degeneracy index i can take the values 1, 2, , gk . The
set of vectors {|k, 1 , |k, 2 , , |k, gk } forms an orthonormal basis for the
eigensubspace Fk . The closure relation can be written as
1=
gk

k
i=1
|k, i k, i| =
Pk ,
(9.3)
where
Pk =
gk

i=1
|k, i k, i|
(9.4)
is a projector onto eigen subspace Fk . The orthogonality condition (9.2)

implies that
Pk Pk = Pk kk .
(9.5)
A perturbation V = V is being added to the Hamiltonian

H = H0 + V ,
(9.6)
where R. We wish to find the eigenvalues and the eigenvectors of the

Hamiltonian H
H | = E | .
(9.7)
Chapter 9. Time Independent Perturbation Theory

In many cases finding an analytical solution to the above equation is either
very hard or impossible. In such cases one possibility is to employ numerical
methods. However, another possibility arrises provided that the eigenvalues
and eigenvectors of H0 are known and provided that the perturbation V can
be considered as small, namely, provided the eigenvalues and eigenvectors of
H do not significantly differ from those of H0 . In such a case an approximate
solution can be obtained by the time independent perturbation theory.
9.1 The Level En

Consider the level En of the unperturbed Hamiltonian H0 . Let Pn be the
projector onto the eigensubspace Fn , and let

Pk .
(9.8)
Qn = 1 Pn =
k
=n
Equation (9.7) reads

V | = (E H0 ) | .
(9.9)
It is useful to introduce the operator R, which is defined as

R=
Pk
.
E
Ek
k
=n
(9.10)
Claim. The eigenvector | of the Hamiltonian H is given by

1
| = 1 RV
Pn | .
(9.11)
Proof. Using Eq. (9.5) it is easy to show that

Pn R = RPn = 0 .
(9.12)
Moreover, the following holds

Qn R =
Pk Pk
Pk
=
=R,
E Ek
E Ek
k
=n k
=n
(9.13)
k
=n
and similarly
RQn = R .
(9.14)
Furthermore, by expressing H0 as
H0 =
gk

Eyal Buks
i=1
Ek |k, i k, i| = En Pn +
Ek P k ,
(9.15)
k
=n
294
9.1. The Level En

one finds that
R (E H0 ) =
=

<
Pk E En Pn k
=n Ek Pk
k
=n
E Ek
Pk (E Ek )
E Ek
k
=n
= Qn ,
(9.16)
and similarly
(E H0 ) R = Qn .
(9.17)
The last two results suggest that the operator R can be considered as the
inverse of E H0 in the subspace of eigenvalue zero of the projector Pn
(which is the subspace of eigenvalue unity of the projector Qn ). Multiplying
Eq. (9.9) from the left by R yields
RV | = R (E H0 ) | .
(9.18)

RV | = Qn | .
Since Pn = 1 Qn [see Eq. (9.8)] the last result implies that

Pn | = | RV | = 1 RV | ,
(9.19)
(9.20)
which leads to Eq. (9.11)

1
Pn | .
| = 1 RV
Note that Eq. (9.11) can be expanded as power series in

| = 1 + RV + 2 RV RV + Pn | .
(9.21)
(9.22)
9.1.1 Nondegenerate Case

In this case gn = 1 and
Pn = |n n| .
(9.23)
In general the eigenvector | is determined up to multiplication by a constant. For simplicity we choose that constant to be such that
Eyal Buks
295

Pn | = |n ,
(9.24)
namely
n | = 1 .
(9.25)
Multiplying Eq. (9.9), which is given by

V | = (E H0 ) | ,
(9.26)
from the left by n| yields

n| V | = n| (E H0 ) | ,
(9.27)
n| E | = n| H0 | + n| V | ,
(9.28)
or
thus
E = En + n| V | .
(9.29)
Equation (9.22) together with Eq. (9.24) yield

| = 1 + RV + 2 RV RV + |n
= |n +
+2
|k, i k, i| V |n

E Ek
k
=n
i
|k, i k, i| V |k , i k , i| V |n

(E Ek ) (E Ek )
k
=n k =
n
i
i
+ .
(9.30)
Substituting Eq. (9.30) into Eq. (9.29) yields

E = En + n| V |n
n| V |k, i k, i| V |n
+2
E Ek
k
=n
i
+3
n| V |k, i k, i| V |k , i k , i| V |n

(E Ek ) (E Ek )
k
=n k =
n
i
i
+ .
Eyal Buks
(9.31)
296
9.2. Example
Note that the right hand side of Eq. (9.31) contains terms that depend on E.
To second order in one finds
E = En + n| V |n +
|k, i| V |n|2

+ O 3 .
En Ek
(9.32)
k
=n
i
Furthermore, to first order in Eq. (9.30) yields

| = |n +
|k, i k, i| V |n

+ O 2 .
En Ek
(9.33)
k
=n
i
9.1.2 Degenerate Case

The set of vectors {|n, 1 , |n, 2 , , |n, gn } forms an orthonormal basis for
the eigensubspace Fn . Multiplying Eq. (9.9) from the left by Pn yields
Pn V | = Pn (E H0 ) | ,
(9.34)
thus with the help of Eq. (9.15) one has

Pn V | = (E En ) Pn | .
Substituting Eq. (9.22), which is given by

| = Pn + RV Pn + 2 RV RV Pn + | ,
(9.35)
(9.36)
into this and noting that Pn2 = Pn yield
Pn V Pn | + 2 Pn V RV Pn | + = (E En ) Pn | .
(9.37)
Thus, to first order in the energy correction E En is found by solving

Pn V Pn | = (E En ) Pn | .
(9.38)
The solutions are the eigenvalues of the gn gn matrix representation of the

operator V in the subspace Fn .
9.2 Example
Consider a point particle having mass m whose Hamiltonian is given by
H = H0 + V ,
(9.39)
where
Eyal Buks
297

H0 =
p2
m2 x2
+
.
2m
2
(9.40)
and where
V =
m
x.
(9.41)
The eigenvectors and eigenvalues of the Hamiltonian H0 , which describes a

one dimensional harmonic oscillator, are given by
H0 |n = En |n ,
where n = 0, 1, , 2 , and where

1
.
En ( = 0) = n +
2
(9.42)
(9.43)
Note that, as was shown in chapter 5 [see Eq. (5.161)], the eigenvectors
and eigenvalues of H can be found analytically for this particular case. For
the sake of comparison we first derive this exact solution. Writing H as
0
m 2 x2
p2
m
H=
+
+
x
2m
2

!
"
0
2
p2
m 2

1
=
+
x+
2 ,
2m
2
m
2
(9.44)
one sees that H describes a one dimensional harmonic oscillator (as H0 also
does). The exact eigenenergies are given by
1
En () = En ( = 0) 2 ,
2
and the corresponding exact wavefunctions are
=
0
x |n () = x +
|n .
m
(9.45)
(9.46)
Using identity (3.19), which is given by

J (x) |x = |x + x ,
(9.47)
where J (x) is the translation operator, the exact solution (9.46) can be
rewritten as
! 0
"

x |n () = x | J
|n ,
(9.48)
m
Eyal Buks
298
9.2. Example
or simply as
|n () = J

m
"
|n .
(9.49)
Next we calculate an approximate eigenvalues and eigenvectors using perturbation theory. Using the identity
0

a + a ,
x=
(9.50)
2m
one has

V = a + a .
2
(9.51)
Furthermore, using the identities
a |n = n |n 1 ,
a |n = n + 1 |n + 1 ,
(9.52)
(9.53)
one has

m| V |n = m| a |n + m| a |n
2

n m,n1 + n + 1 m,n+1 .
=
2
(9.54)
Thus En () can be expanded using Eq. (9.32) as

En () = En + n| V |n +

=0
|k, i| V |n|2

+ O 3
En Ek
k
=n
i
2

|n 1| V |n|
|n + 1| V |n|2
= En +
+
+ O 3
En En1
En En+1

2

1
n
(n + 1) 2
= n +
+

+ O 3
2
2
2

2

1
= n +

+ O 3 ,
2
2
(9.55)
in agreement (to second order) with the exact result (9.45), and |n () can
be expanded using Eq. (9.30) as
Eyal Buks
299

|n () = |n +
|k, i k, i| V |n

+ O 2
En Ek
k
=n
i

|n 1 n 1| V |n |n + 1 n + 1| V |n
+
+ O 2
En En1
En En+1
|n 1
n |n + 1
n+1

2
2
+ O 2
= |n +

2

= |n + a |n a |n + O .
2
2
(9.56)
= |n +
Note that with the help of the following identify

0

m
a + a ,
p=i
2
the last result can be written as
!
"
0

ip
|n () = 1 +
|n + O 2 .
m
(9.57)
(9.58)
Alternatively, in terms of the translation operator J (x), which is given by

ipx
J (x) = exp
,
(9.59)

one has
|n () = J

m
"

|n + O 2 ,
(9.60)
in agreement (to second order) with the exact result (9.49).
9.3 Problems
1. The volume effect: The energy spectrum of the hydrogen atom was
calculated in chapter 8 by considering the proton to be a point particle.
Consider a model in which the proton is instead assumed to be a sphere
of radius 0 where 0 << a0 (a0 is Bohrs radius), and the charge of
the proton +e is assumed to be uniformly distributed in that sphere.
Show that the energy shift due to such perturbation to lowest order in
perturbation theory is given by
En,l =
e2 2
|Rn,l (0)|2 ,
10 0
(9.61)
where Rn,l (r) is the radial wave function.

Eyal Buks
300
9.3. Problems
2. Consider a particle having mass m in a 3D central potential given by
m 2 r2
+ gr4 .
(9.62)
2

where r = x2 + y2 + z 2 is the radial coordinate, and where and g
are both positive. Calculate to lowest nonvanishing order in g the energy
of the ground state.
3. Consider an hydrogen atom. A perturbation given by
V (r) =
V = Ar ,
(9.63)

where r = x2 + y 2 + z 2 is the radial coordinate and A is a constant is
added.
a) Calculate to first order in A the energy of the ground state.
b) Calculate to first order in A the energy of the first excited state.
4. A weak uniform electric field E = E
z, where E is a constant, is applied
to a hydrogen atom. Calculate to 1st order in perturbation theory the
correction to the energy of the
a) level n = 1 (n is the principle quantum number).
b) level n = 2.
5. A particle having mass m and charge q is confined in a 3D infinite potential well of width l, which is given by
+
0 if |x| l/2 and |y| l/2 and |z| l/2
V (x, y, z) =
. (9.64)
+
elsewhere
A weak uniform electric field E = E
z, where E is a constant, is applied.
Calculate the eigenenergies to first order in E.
6. Consider two particles, both having the same mass m, moving in a onedimensional potential with coordinates x1 and x2 respectively. The potential energy is given by
1
1
V (x1 , x2 ) = m 2 (x1 a)2 + m 2 (x2 + a)2 + m2 (x1 x2 )2 ,
2
2
(9.65)
where is real. Find the energy of the ground state to lowest nonvanishing order in .
7. A particle having mass m is confined in a potential well of width l, which
is given by
+
0 for 0 x l
V (x) =
.
(9.66)
+ elsewhere
Find to lowest order in perturbation theory the correction to the ground
state energy due to a perturbation given by
Eyal Buks
301

l
W (x) = w0 x
,
2
(9.67)
where w0 is a real constant.

8. Consider a particle having mass m in a two dimensional potential well of
width a that is given by
+
0 if 0 x a and 0 y a
V (x, y) =
.
(9.68)
+
elsewhere
A perturbation given by
+
w0 if 0 x a2 and 0 y
W (x, y) =
0
elsewhere
a
2
(9.69)
is added.
a) Calculate to lowest non-vanishing order in w0 the energy of the
ground state.
b) The same for the first excite state.
9. Consider a particle having mass m moving in a potential energy given by
V (x, y) =

m 2 2
x + y 2 + m 2 xy ,
2
(9.70)
where the angular frequency is a constant and where the dimensionless

real constant is assumed to be small.
a) Calculate to first order in the energy of the ground state.
b) Calculate to first order in the energy of the first excited state.
10. Consider a harmonic oscillator having angular resonance frequency . A
perturbation given by

V = a a + aa
(9.71)
is added, where a is the annihilation operator and is a real constant.

Calculate the energies of the system to second order in .
11. The Hamiltonian of a spin S = 1 is given by

H = Sz2 + Sx2 Sy2 ,
(9.72)
where and are both constants.

a) Write the matrix representation of H in the basis {|S = 1, m = 1 , |S = 1, m = 0 |S = 1, m = 1}.
b) Calculate (exactly) the eigenenergies and the corresponding eigenvectors.
c) For the case << use perturbation theory to calculate to lowest
order in and the eigenenergies of the system.
Eyal Buks
302
9.3. Problems
12. Jaynes-Cummings model - Consider a system composed of a harmonic
oscillator having angular resonance frequency r > 0 and a two-level
system. The Hamiltonian of the system is assumed to be given by
H = Hr + Ha + V .
(9.73)
The term Hr is the Hamiltonian for the harmonic oscillator [see Eq.
(5.16)]

1
Hr = r a a +
,
(9.74)
2
where a and a are annihilation and creation operators respectively. The
term Ha is the Hamiltonian for the two-level system
Ha =
a
z ,
2
(9.75)
where
z = |+ +| | | ,
the ket vectors | represent the two levels and where a > 0. The
coupling term between the oscillator and the two-level system is given by

V = g a + a+ ,
(9.76)
where
+ = |+ | ,
= | +| .
(9.77)
(9.78)
a) Calculate to lowest non-vanishing order in g the eigenenergies of the

system for the case r = a .
b) The same for the case r = a .
c) Consider the unitary transformation
H = U HU ,
where
U = exp
g
S ,
the operator S is given by

S = a a+ ,
(9.79)
(9.80)
(9.81)
and where
= a r .
(9.82)
Calculate H to second order in g/.

Eyal Buks
303

d) Find the exact energy eigenvectors and eigenenergies of H.
e) Find a unitary operator U that diagonalizes H.
f) Use the result of the previous exercise and calculate H = U HU to
forth order in g/.
13. Consider a particle having mass m in a two-dimensional potential given
by

1
V0 = m 2 x2 + y 2 .
(9.83)
2
The following perturbation is added
2
L ,
(9.84)
z
where Lz is the z component of the angular momentum operator.
a) Find to second orders in the energy of the ground state.
b) Find to first order in the energy of the first excited level.
14. A particle having mass m moves in a one dimensional potential
+
V0 sin 2x
l 0xl .
(9.85)
V (x) =
else
V1 =
Consider the constant V0 to be small. Calculate the systems eigenenergies

En to first order in V0 .
15. Consider a particle having mass m confined by the one-dimensional potential well, which is given by
x < 0
V (x) = x
0xL .
L
x>L
Find to first order in the energy of the ground state.

16. A particle of mass m is trapped in an infinite 2 dimensional well of width
l
+
0 0 x l and 0 y l
V (x, y) =
.
(9.86)
else
A perturbation given by
W (x, y) =
2 2
(x lx ) (y ly ) .
m
(9.87)
is added, where
0 lx l ,
(9.88)
0 ly l .
(9.89)
and
Calculate to 1st order in perturbation theory the correction to the energy

of the:
Eyal Buks
304
9.3. Problems
a) ground state.
b) first excited state.
17. Consider a rigid rotator whose Hamiltonian is given by
H=
L2x + L2y
L2x L2y
L2
+ z +
,
2Ixy
2Iz
2Ixy
(9.90)
where L is the angular momentum vector operator. Use perturbation

theory to calculate the energy of the ground state to second order in .
18. Consider two particles having the same mass m moving along the x axis.
The Hamiltonian of the system is given by
H=
p21
p2
+ 2 (x1 ) (x2 ) + (x1 x2 ) ,
2m 2m
(9.91)
where x1 and x2 are the coordinates of the first and second particle
respectively, p1 and p2 are the corresponding canonically conjugate momentums, and are both real positive constants and () denotes the
delta function. Calculate to first order in the energy of the ground state
of the system.
19. In this problem the main results of time independent perturbation theory
are derived using an alternative approach. Consider a general square
matrix
W = D + V ,
(9.92)
where R, D is diagonal
D |n0 = n0 |n0 ,
n0 | D = n0 n0 | ,
(9.93a)
(9.93b)
and we assume that none of the eigenvalues of D is degenerate. The set

of eigenvectors of D is assumed to be orthonormal
n0 |m0 = nm ,
and complete (the dimensionality is assumed to be finite)

1=
|n0 n0 | .
(9.94)
(9.95)
Calculate the eigenvalues of W

W |n = |n
(9.96)
to second order in .
Eyal Buks
305

20. Consider the Hamiltonian H, which is given by
H = H0 + V ,
(9.97)
where R. The set {|k} of eigenvectors of H0 with corresponding

eigenvalues {Ek }, which satisfy
H0 |k = Ek |k ,
(9.98)
is assumed to form an orthonormal basis for the vector space, i.e.

k |k = kk .
(9.99)
Consider the transformation

HR = eL HeL ,
(9.100)
where the operator L is assumed to be anti Hermitian, i.e. L = L, in

order to ensure that eL is unitary. Show that to second order in the
matrix elements k| HR |k are given by

2
1
1
k| HR |k = Ek k,k +
k| V |k k | V |k
,
2
Ek Ek
Ek Ek
k
(9.101)
provided that the following condition is satisfied

V + [L, H0 ] = 0 .
(9.102)
Note that to first order in the following holds k| HR |k = k| H0 |k ,

thus, in spite of the fact that the perturbation V is first order in , the
transformed Hamiltonian HR depends on only to second order.
21. Calculate the expectation values of the kinetic energy nlm| T |nlm and
the potential energy nlm| V |nlm of a hydrogen atom in an energy eigenstate |nlm.
22. Calculate the expectation values nlm| r2 |nlm, where r is the radial
position coordinate and where |nlm is an energy eigenstate |nlm of a
hydrogen atom.
9.4 Solutions
1. The radial force acting on the electron is found using Gauss theorem
e2
r 2 r > 0

Fr (r) = e2 r 3
.
(9.103)
r2
r 0
0
Eyal Buks
306
9.4. Solutions
The potential energy V (r) is found by integrating Fr (r) and by requiring
that V (r) is continuous at r = 0
e2
r
r > 0
2
V (r) =
.
(9.104)
e2
r
2
3 r 0
Thus, the perturbation term in the Hamiltonian is given by
2
0
r > 0
e
2
=
. (9.105)
Vp (r) = V (r)
20
e2
r
+ r 3 r 0
20
r
0
To first order one has
En,l = nlm| Vp |nlm .
(9.106)
The wavefunctions for the unperturbed case are given by

nlm (r, , ) = Rnl (r) Ylm (, ) ,
(9.107)
Since Vp depends on r only, one finds that

En,l =
drr2 |Rnl (r)|2 Vp (r)
0
e2
drr |Rnl (r)|
20
!
r
0
2
2
+ 0 3
r
"
.
(9.108)
In the limit where 0 << a0 the term |Rnl (r)| can approximately be
replaced by |Rnl (0)|2 , thus
! 2
"
0
2
r
20
2
2 e
En,l = |Rnl (0)|
drr
+
3
20
0
r
0
e2 20
=
|Rnl (0)|2 .
10
(9.109)
2. For the unperturbed case, i.e. when g = 0, the energy eigenvectors are
denoted by |nx , ny , nz , where the quantum numbers nx , ny and nz are
non-negative integers, and the corresponding eigenenergies are given by

3
Enx ,ny ,nz =
+ nx + ny + nz .
(9.110)
2
Eyal Buks
307

With the help of Eqs. (5.11), (5.13), (5.28), (5.29) and (9.31) together
with the relation

2
r4 = x2 + y 2 + z 2
= x4 + y 4 + z 4 + 2x2 y2 + 2y2 z 2 + 2z 2 x2 ,
(9.111)
one finds that the energy of the ground state Egs is given by

3
Egs =
+ g 0, 0, 0| r4 |0, 0, 0 + O g2
2

2

3
=
+ 3g 0, 0, 0| x4 |0, 0, 0 + 6g 0, 0, 0| x2 |0, 0, 0 + O g 2
2

2

3

=
+ 15g
+ O g2 .
2
2m
(9.112)
3. The wavefunctions for the unperturbed case are given by
nlm (r, , ) = Rnl (r) Ylm (, ) ,
where for the states relevant to this problem
3/2
1
R10 (r) = 2
er/a0 ,
a0

3/2
r
1
R20 (r) = (2 r/a0 )
e 2a0 ,
2a0

3/2
r
1
r
R21 (r) =
e 2a0 ,
2a0
3a0
0
1
Y00 (, ) =
,
4
0
1
3
Y11 (, ) =
sin ei ,
2 2
0
1 3
0
Y1 (, ) =
cos ,
2
0
1
3
Y11 (, ) =
sin ei ,
2 2
and the corresponding eigenenergies are given by
En(0) =
EI
,
n2
(9.113)
(9.114a)
(9.114b)
(9.114c)
(9.114d)
(9.114e)
(9.114f)
(9.114g)
(9.115)
where
EI =
Eyal Buks
me e4
.
22
(9.116)
308
9.4. Solutions
The perturbation term V in the Hamiltonian is given by V = Ar. The
matrix elements of V are expressed as
n l m | V |nlm = A
dr r3 Rn l Rnl
= A l,l m,m
1
d (cos )
2

Ylm
Ylm
dr r3 Rn l Rnl .
(9.117)
a) Thus, to first order
(0)
E1 = E1 + 100| V |100 + O(A2 ) ,
(9.118)
where
100| V |100 = A
2
dr r3 R10
(r) =
3Aa0
.
2
(9.119)
b) The first excited state is degenerate, however, as can be seen from

Eq. (9.117) all off-diagonal elements are zero. The diagonal elements
are given by

(9.120a)
200| V |200 = A dr r3 R220 = 6Aa0 ,
0
21m| V |21m = A
dr r3 R21 = 5Aa0 .
(9.120b)
Thus, the degeneracy is lifted

(0)
(9.121)
(0)
(9.122)
E2,l=0 = E2 + 6Aa0 + O(A2 ) ,

E2,l=1 = E2 + 5Aa0 + O(A2 ) .
4. The wavefunctions for the unperturbed case are given by
nlm (r, , ) = Rnl (r) Ylm (, ) ,
(9.123)
where for the states relevant to this problem
Eyal Buks
309

R10 (r) = 2
1
a0
3/2
R20 (r) = (2 r/a0 )
1
2a0
(9.124)
3/2
e 2a0 ,
(9.125)
3/2
1
r
r
e 2a0 ,
2a0
3a0
0
1
Y00 (, ) =
,
4
0
1
3
sin ei ,
Y11 (, ) =
2 2
0
1 3
0
cos ,
Y1 (, ) =
2
0
1
3
Y11 (, ) =
sin ei ,
2 2
and the corresponding eigenenergies are given by
R21 (r) =
En(0) =
er/a0 ,
(9.126)
(9.127)
(9.128)
(9.129)
(9.130)
EI
,
n2
(9.131)
where
EI =
me e4
.
22
(9.132)
The perturbation term V in the Hamiltonian is given by

V = eEz = eEr cos .
(9.133)
The matrix elements of V are expressed as

n l m | V |nlm = eE
dr r Rn l Rnl
1
d (cos )
2
0

d cos Ylm
Ylm .
(9.134)
a) Disregarding spin this level is non degenerate. To 1st order

(0)
(0)
E1 = E1 + 1, 0, 0| V |1, 0, 0 = E1 ,
since
1
d (cos ) cos = 0 ,
thus the energy of the ground state is unchanged to 1st order.

Eyal Buks
310
9.4. Solutions
b) The level n = 2 has degeneracy 4. The matrix of the perturbationV
in thedegenerate subspace is given by
2, 0, 0| V |2, 0, 0 2, 0, 0| V |2, 1, 1 2, 0, 0| V |2, 1, 0 2, 0, 0| V |2, 1, 1

2, 1, 1| V |2, 0, 0 2, 1, 1| V |2, 1, 1 2, 1, 1| V |2, 1, 0 2, 1, 1| V |2, 1, 1
M =
2, 1, 0| V |2, 0, 0 2, 1, 0| V |2, 1, 1 2, 1, 0| V |2, 1, 0 2, 1, 0| V |2, 1, 1 .
2, 1, 1| V |2, 0, 0 2, 1, 1| V |2, 1, 1 2, 1, 1| V |2, 1, 0 2, 1, 1| V |2, 1, 1
(9.135)
Using
1
d (cos ) cos = 0 ,
(9.136)
1
d (cos ) cos sin = 0 ,
(9.137)
d (cos ) cos sin2 = 0 ,
(9.138)
1
1
d (cos ) cos3 = 0 ,
2
(9.139)
d ei = 0 ,
(9.140)
0
0
,
0
0
(9.141)
one finds
0
0
M =

0
0
00
00
00
where
= 2, 0, 0| V |2, 1, 0

1
2

3
= eE dr r R2,0 R2,1 d (cos ) d cos Y00 Y10
0
eE
8

0
Eyal Buks

4
r
r
r
dr 2
e a0
a0
a0
1
d (cos ) cos
2
d .
(9.142)
311

Using
1
d (cos ) cos2 =
2
,
3
(9.143)
and
x4 ex dx = 24
(9.144)
x5 ex dx = 120
(9.145)
one finds
= 2, 0, 0| V |2, 1, 0
4

r
r
eE
r
dr 2
e a0
=
24
a0
a0
0
a0 eE
=
24

0
dx (2 x) x4 ex
= 3a0 eE .
(9.146)
The eigenvalues of the matrix M are 0, 0, 3a0 eE and 3a0 eE. Thus
to 1st order the degeneracy is partially lifted with subspace of dimen(0)
sion 2 having energy E2 , and another 2 nondegenerate subspaces
(0)
having energies E2 3a0 eE.

5. For E = 0 the normalized wavefunctions (0)
nx ,ny ,nz (x , y , z ) are given by

(0)
nx ,ny ,nz (x , y , z )
= x , y , z |nx , ny , , nz

3/2
nx x + 2l
ny y + 2l
nz z + 2l
2
=
sin
sin
sin
,
l
l
l
l
(9.147)

2 2 n2x + n2y + n2z
(0)
Enx ,ny ,nz =
,
2ml2
(9.148)
where nx , ny , nz {1, 2, }. The matrix elements of the perturbation

V = qEz are given by
)

* 2qE
nx , ny , nz
V
nx , ny , nz =
nx ,nx ny ,ny Inz ,nz ,
l
(9.149)
where
Eyal Buks
312
9.4. Solutions
I
nz ,n
z

nz z + 2l
nz z + 2l
=
dz sin
sin
z .
l
l
l/2
l/2
(9.150)
Note that Inz ,nz = 0 if nz = nz (since for that case the integrand is an
odd function of z ), and thus
)

*

nx , ny , nz
V
nx , ny , nz nx ,nx ny ,ny 1 nz ,nz .
(9.151)
With the )help of the

result* it is easy to see that all the matrix
above
elements nx , ny , nz
V
nx , ny , nz that are needed for first order perturbation theory, for both non-degenerate and degenerate energy levels,
vanish, and consequently, to first order in E the energy eigenstates remain
unchanged.
6. To lowest order in perturbation theory the ground state energy is given
by
Egs = +m
dx1

dx2 20 (x1 a) 20 (x2 + a) (x1 x2 )2 +O 2 ,
(9.152)
where 0 (x) is the ground state wavefunction of a particle having mass

m confined by a potential (1/2) m2 x2 centered at x = 0. Employing the
transformation
x1 = x1 a ,
x2 = x2 + a ,
(9.153)
(9.154)

Egs =
+m
+m2
dx1 20 (x1 ) (x1 + a)
2m

+O 2
dx2 20 (x2 ) (x2 + a)
dx1 20 (x1 ) (x1
+ a)
dx2 20 (x2 ) (x2 a)

+ a2 + 2m 2 a2 + O 2
= + 2m2
2m

= + + 4m 2 a2 + O 2 .
Eyal Buks
(9.155)
313

Note that this problem can be also solved exactly by employing the coordinate transformation
x1 + x2
x+ =
,
(9.156)
2
x1 x2
x =
.
(9.157)
2
x+ + x
x1 =
,
(9.158)
2
x+ x
x2 =
.
(9.159)
2
The following holds
x21 + x22 = x2+ + x2 ,
(9.160)
x 21 + x 22 = x 2+ + x 2 .
(9.161)
and
Thus, the Lagrangian of the system can be written as

m x 21 + x 22
L=
V (x1 , x2 )

22

m x + + x 2
1
=
m 2 x2+ + x2 2a 2x + 2a2 + 4x2
2
2
= L+ + L ,
(9.162)
where
L+ =
mx 2+ 1
m2 x2+ ,
2
2
(9.163)
and
!
"2
2
mx 2 1
a
2
8a
. (9.164)
L =
m 2 (1 + 4) x
+
2
2
1 + 4
1 + 4
Thus, the system is composed of two decoupled harmonic oscillators, and

therefore, the exact eigenenergies are given by

1
1
4m 2 a2
En+ ,n = n+ +
+ 1 + 4 n +
+
, (9.165)
2
2
1 + 4
where n+ , n = 0, 1, 2, . To first order in one thus has

,

1
1
En+ ,n = n+ +
+ n +
+ (2n + 1) + 4m 2 a2 +O 2 .
2
2
(9.166)
Eyal Buks
314
9.4. Solutions
7. For w0 = 0 the normalized wavefunctions (0)
n (x) are given by
0
2
nx
(x)
=
x
|n
=
(0)
sin
,
n
l
l
(9.167)

En(0) =
2 2 n2
.
2ml2
(9.168)
The matrix elements of the perturbation are given by

2w0
n|W |m =
l
l
sin

nx
mx
l
sin
x
dx
l
l
2
2w0
n
m
=
sin
sin
.
l
2
2
(9.169)
For the ground state
1|V |1 =
2w0
,
l
(9.170)
thus
E1 =

2 2
2w0
+
+ O w02 .
2
2ml
l
(9.171)
8. For w0 = 0 the normalized wavefunctions (0)

nx ,ny (x, y) are given by

(0)
nx ,ny (x , y ) = x , y |nx , ny =
2
nx x
ny y
sin
sin
,
a
a
a

2 2 n2x + n2y
(0)
Enx ,ny =
,
2ma2
(9.172)
(9.173)
where nx = 1, 2, and ny = 1, 2, .
a) The ground state (nx , ny ) = (1, 1) is nondegenerate, thus to first
order in w0
2 2
E0 =
+ 1, 1|W |1, 1
ma2
a/2
a/2
2 2 4w0
x
y
=
+ 2
sin2
dx
sin2
dy
ma2
a
a
a
0
2 2

w0
+
,
ma2
4
(9.174)
Eyal Buks
315

b) The first excite state is doubly degenerate. The matrix of the perturbation
in the correspondingsubspace is given by

1, 2|W |1, 2 1, 2|W |2, 1
2, 1|W |1, 2 2, 1|W |2, 1
a/2
a/2
a/2
a/2
/
/
/
/
2y
y
2 x
2 2y
x
2x
sin a dx
sin a dy
sin a sin a dx
sin a sin a dy
4w0
0
0
0
0
= 2 a/2
a/2
a/2
a/2
/
/
/
a /
y
2y
2 2x
2 y
2x
1x
sin a sin a dx
sin a sin a dy
sin a dx
sin a dy
0
0
0
0
1 16
4 92
,
= w0 16
1
92
(9.175)
To first order in perturbation theory the eigenenergies are found

by adding the eigenvalues of the above matrix to the unperturbed
(0)
(0)
eigenenergy E1,2 = E2,1 . Thus, to first order in w0
E1, =

52 2 w0 16w0
+
+ O w02 .
2
2
2ma
4
9
(9.176)
9. For the unperturbed case = 0 one has
H0 |nx , ny = (nx + ny + 1) |nx , ny ,

where nx , ny = 0, 1, 2, . Using the identities
0

x=
ax + ax ,
2m
0

y=
ay + ay ,
2m
the perturbation term V1 = m 2 xy can be expressed as
V1 =
(9.177)
(9.178)
(9.179)

ax + ax ay + ay .
2
a) For the ground state |0, 0, which is nondegenerate, one has
|nx , ny | V1 |0, 0|2
E0,0 () = + 0, 0| V1 |0, 0 +

E0,0 (0) Enx ,ny
nx ,ny
=0,0
=0
= +
|1, 1| V1 |0, 0|

2

2

2
=
2

2
= 1
.
8
Eyal Buks
(9.180)
316
9.4. Solutions
b) The first excited state is doubly degenerate, thus the eigenenergies
are found by diagonalizing the matrix of V1 in the corresponding
subspace

0 1
1, 0| V1 |1, 0 1, 0| V1 |0, 1
.
(9.181)
=
0, 1| V1 |1, 0 0, 1| V1 |0, 1
10
2
Thus the degeneracy is lifted and the energies are given by 2 (1 /4).
Note that this problem can be also solved exactly by employing the
coordinate transformation
x+y
x = ,
(9.182)
2
xy
y = .
(9.183)
2
x + y
x=
,
(9.184)
2
x y
.
(9.185)
y=
2
The following hold
x2 + y 2 = x2 + y 2 ,
(9.186)
2
2
2
2
x + y = x + y ,
(9.187)

1 2
(9.188)
xy =
x y 2 .
2
Thus, the Lagrangian

of the system can be written as
m x 2 + y 2
L=
V (x1 , x2 )

22
m 2 2

m x + y 2
m 2 2
=
x + y 2
x y 2
2
2
2
= L+ + L ,
(9.189)
where
L+ =
mx 2 m 2
(1 + ) x2 ,
2
2
(9.190)
L =
my 2 m 2
(1 ) y2 .
2
2
(9.191)
and
Thus, the system is composed of two decoupled harmonic oscillators,

and therefore, the exact eigenenergies are given by

1
1
En+ ,n =
1 + nx +
+ 1 ny +
, (9.192)
2
2
where nx , ny = 0, 1, 2, . To second order in one thus has

Eyal Buks
317

nx ny
nx + ny + 1 2
En+ ,n = nx + ny + 1 +
+ O 3 .
2
8
(9.193)
10. Using Eqs. (5.28) and (5.29) one finds that

m| V |n = n (n 1) m,n2 + (n + 1) (n + 2) m,n+2 , (9.194)
thus
1
En () = n +
2
+ n| V |n +

=0
2
m
=n

|m| V |n|2
+ O 3
En (0) Em (0)

+
[n (n 1) (n + 1) (n + 2)] + O 3
2

2

1
+ O 3 .
= n +
12
2
= n +
1
2
(9.195)
11. In general the subspace of angular momentum states with J = 1 is

spanned by the basis
{|j = 1, m = 1 , |j = 1, m = 0 , |j = 1, m = 1} ,
(9.196)

j , m | Jz |j, m = m j ,j m ,m ,
j , m | J2 |j, m = j (j + 1) 2 j ,j m ,m ,

j , m | J |j, m = (j m) (j m + 1) j ,j m ,m1 ,
J = Jx iJy .
In matrix form
10 0
0 0 0 ,
Jz =
0 0 1
100
J2 =
22 0 1 0 ,
001
010
20 0 1 ,
J+ =
000
000
J =
21 0 0 .
010
Eyal Buks
(9.197)
(9.198)
(9.199)
(9.200)
(9.201)
(9.202)
(9.203)
(9.204)
318
9.4. Solutions
a) The Hamiltonian is given by

H = Sz2 + Sx2 Sy2
4
3
2
2
= Sz2 +
(S+ + S ) + (S+ S )
4

2
2
2
= Sz +
S+ + S
.
2
Thus, in matrix
form
100
001
000
H=
2 0 0 0 + 2 0 0 0 + 0 0 0
001
000
100
0
= 2 0 0 0 .
0
b) The eigenvalues
and
are given
eigenvectors
by
0
1
1
2
2
0
0
0
0

= ( + ) 0 ,
0
1
1
0
1
1
2 0 0 0 0 = 2 ( ) 0 ,
0
1
1

0
0
0
2 0 0 0 1 = 2 0 1 .
0
0
0
(9.205)
(9.206)
(9.207)
(9.208)
(9.209)
c) The Hamiltonian
as H = H0 + V where in matrix form
is written
100
H0 =
2 0 0 0 ,
(9.210)
001
001
V =
2 0 0 0 .
(9.211)
100
0
= 0 on has to second order
For the nondegenerate eigenenergy Em=0
in perturbation expansion
0
Em=0 = Em=0
+1, 0| V |1, 0+
|1, m | V |1, 0|2

= 0 . (9.212)
0
0
Em=0
Em
m =1
0
For the degenerate eigenenergy Em=1
= 2 the perturbation in
the subspace spanned by {|1, 1 , |1, 1} is given in matrix form by

01
Vm=1 =
2
,
(9.213)
10
Eyal Buks
319

thus to first order in perturbation expansion
Em=1 = 2 ( ) .
(9.214)
12. For the unperturbed case V = 0, the eigenvectors and eigenenergies are
related by
0
(Hr + Ha ) |n, = En,
|n, ,
(9.215)
where n = 0, 1, 2, is the quantum number of the harmonic oscillator,

and {1, +1} is the quantum number associated with the two-level
particle, and

1
a
0
En, = r n +
+
.
(9.216)
2
2
a) To second order in perturbation theorem [see Eq. (9.32)]
0
+ n, | V |n, +
En, = En,
|n , | V |n, |2

. (9.217)
0 E0
En,
n ,
n ,
=n,

Using
V |n, + = ga |n, = g n + 1 |n + 1, ,
V |n, = ga |n, + = g n |n 1, + ,

one finds for =
+1

1
a g 2 (n + 1)
En,+1 = r n +
+
+
2
2
a r

2

2
a + g
g
1
= r +
n+
+
,
2
2
(9.218)
(9.219)
(9.220)
and for = 1

1
a
En,1 = r n +
2
2

2
g
= r
n+
g2 n
a r

a +
1
+
2
2
g2
,
(9.221)
where
= a r .
For the general case this can be written as

g2
1
1
g2
En, = r +
n+
+
a +
.
2
2
Eyal Buks
(9.222)
(9.223)
320
9.4. Solutions
Thus, according to the above result (9.223), the energies of the states
(n, +1) and (n + 1, ), which are degenerate for the case where r =
a and where g = 0, are given to second order in g by

g2
(n + 1) +
,
(9.224)
En,+1 = r +
2
and
En+1,1

g2
(n + 1)
.
= r
(9.225)
b) In the degenerate case r = a the eigenenergies for the case

V = 0 are given by

1
0
En,
= n + +
,
(9.226)
2
2
thus the pairs of states |n, + and |n + 1, are degenerate. In the
subset of such a pair the perturbation is given by

n, +| V |n, +
n, +| V |n + 1,
n + 1, | V |n, + n + 1, | V |n + 1,
g n + 1
0
,
g n + 1
0
(9.227)
thus to first order in g the eigenenergies are given by
,
E = (n + 1) g n + 1 .
(9.228)
c) Using Eq. (2.173) one finds that (note that S = S)

H = H + [L, H] +
1
[L, [L, H]] + ,
2!
(9.229)
where
L=

g
a+ a .
(9.230)
Using the commutation relations

[z , + ] = 2+ ,
[z , ] = 2 ,
[+ , ] = z ,
, a, a a = a ,
, a , a a = a ,
(9.231)
(9.232)
(9.233)
(9.234)
(9.235)
one finds that

Eyal Buks
321

g2
[L, H] = g a + a+ + 2
and thus
1 + z
+ a az
2
, (9.236)

g2

g2
H = r + z a a +
a +
z

2
3
r + g
g
+
+O
.
2
(9.237)
Note that to second order in g/ the states |n, are eigenvalues of
H , and the following holds
H |n, = En, |n, ,
(9.238)
where
En, =

g2
1
1
g2
r +
n+
+
a +
.
2
2
(9.239)
The above result agrees with Eq. (9.223).

d) Consider the pair of states |n, + and |n + 1, . The following folds
[see Eq. (9.73)]

H |n, + = r (n + 1) |n, + +
|n, + + g n + 1 |n + 1, ,
2
(9.240)
and

|n + 1, + g n + 1 |n, + ,
H |n + 1, = r (n + 1) |n + 1,
2
(9.241)
where
= a r ,
(9.242)
or in a matrix
form

|n, +
H
|n + 1,

n cos sin
10
|n, +
+
,
= r (n + 1)
01
sin cos
|n + 1,
2
(9.243)
where

n = 2 + 4g 2 (n + 1) ,
(9.244)
2g n + 1
tan =
.
(9.245)
Thus, the states |n+ and |n , which are given by [see Eqs. (6.213)
and (6.214)]
Eyal Buks
322
9.4. Solutions
|n, + + sin |n + 1, ,
2
2
|n = sin |n, + + cos |n + 1, ,

2
2
are eigenstates of H and the following holds
|n+ = cos
H |n = En |n ,
(9.246)
(9.247)
(9.248)
where
3
n 4
En = r (n + 1)
2

0
2
2
= r (n + 1)
+ g (n + 1) .
4
(9.249)
The ground state is the state |0,
H |0, = Eg |n, ,
(9.250)
and the ground state energy is

Eg =

.
2
(9.251)
e) The desired unitary operator U is required to satisfy [see Eq. (9.250)]

U |0, = |0, ,
(9.252)
and [see Eqs. (9.246) and (9.247)]
U |n, + = |n+ = cos |n, + + sin |n + 1, ,

(9.253)
2
2
U |n + 1, = |n = sin |n, + + cos |n + 1, , (9.254)

2
2
where
2g n + 1
tan =
.
(9.255)
The required transformation can be constructed using the operators

S and N , which are defined by
S = a a+ ,
(9.256)
N = a a + |+ +| .
(9.257)
and
The following holds

S |0, = 0 ,
S |n, + = n + 1 |n + 1, ,
S |n + 1, = n + 1 |n, + ,
Eyal Buks
(9.258)
(9.259)
(9.260)
323

and
N |n + 1, = (n + 1) |n, ,
N |n, + = (n + 1) |n, + .
Thus, the operator I, which is defined by
I = N 1/2 S ,
(9.261)
(9.262)
(9.263)
satisfies
I |n, + = |n + 1, ,
I |n + 1, = |n, + .
Therefore, Eqs. (9.246) and (9.247) can be rewritten as
|n+ = U |n, + ,
|n = U |n + 1, ,
where
U = cos
+ sin I .
2
2
(9.264)
(9.265)
(9.266)
(9.267)
(9.268)
Furthermore, with the help of Eq. (9.255) one finds that [note that
I 2 |n, + = |n, + and I 2 |n + 1, = |n + 1, ]

1/2
I
1 2gN
U = exp
tan
.
(9.269)
2
To first order in g/ the following holds [compare with Eq. (9.80)]

3
g
g
U = eS + O
.
(9.270)
f) With the help of Eqs. (2.173) and (9.269) one finds that
H = H + [L, H] +
1
1
[L, [L, H]] + [L, [L, [L, H]]] + , (9.271)
2!
3!
where
L = Sf (N ) ,
(9.272)
and where the function f is given by

x1/2
2gx1/2
f (x) =
tan1
2

g 5
g
4x g 3
=
+O
.
(9.273)
The following holds

8x g 4
g 6
+O
,
3
(9.274)
324
f 2 (x) =
Eyal Buks
g 2
9.4. Solutions
and
f 3 (x) =
g 3
+O

g 5
(9.275)
Using the commutation relations

,
,
a + a+ , a a = a + a+ , |+ +| = a + a+ ,
(9.276)
one finds that

[H, N ] = 0 ,
and using the,commutation relations
a a, a a+ = a + a+ ,

z
, a a+ = a a+ ,
2

,
1
a + a+ , a a+ = 1 + 2 a a +
z ,
2
one finds that

1
[H, S] = a + a+ +g 1 + 2 a a +
z .
2
(9.277)
(9.278)
(9.279)
(9.280)
Thus, the following holds
[L, H] = [H, S] f (N ) ,
(9.281)
and (note that [S, N ] = 0 and [[H, S] , N ] = 0)

[L, [L, H]] = [[H, S] , S] f 2 (N ) ,
(9.282)
where

1
[[H, S] , S] = 1 + 2 a a +
z
2

1
4g a a +
a + a+ + 2g a + a+ z ,
2
(9.283)
and therefore (note that [[[H, S] , S] , N ] = 0)
[L, [L, [L, H]]]
,
= [[H, S] , S] f 2 (N ) , Sf (N )
= [[[H, S] , S] , S] f 3 (N ) ,
Eyal Buks
(9.284)
325

where
[[[H, S] , S] , S]

1
= 4 a a +
a + a+ 2 a + a+ z
2

1
1
1 + 2 a a +
z
4g a a +
2
2
2

8g a + a+

1
+2g 1 + 2 a a +
z z .
2
(9.285)
By combining the above results one finds that
4g 4
4g 4
1 H = r
+ z
a az a a
3
3
33
r
1
+
+ (a + ) z +
2
2
2
g 5
+O
.
(9.286)
where

g2
4g 2
=
1
.
(9.287)
32
13. Using creation and annihilation operators one has
H0 =

p2x + p2y
1
+ m 2 x2 + y 2 = (Nx + Ny + 1) ,
2m
2
(9.288)
where Nx = ax ax , Ny = ay ay , and
2
L
z
=
(xpy ypx )2

-2
,
=
i ax ay ax ay

4
3 2 2
= a2x ay + ax a2y ax ax ay ay ax ax ay ay
3 2 2
4
= a2x ay + ax a2y (1 + Nx ) Ny Nx (1 + Ny ) .
V =
(9.289)
a) For the case = 0 the ground state |0, 0 is nondegenerate and has
energy E0,0 = . Since V |0, 0 = 0 one finds to second order in
Eyal Buks
326
9.4. Solutions
E0,0 = +0, 0| V |0, 0
1

|nx , ny | V |0, 0|2
= +O 3 .
n
+
n
x
y
=0,0
nx ,ny
(9.290)
b) For the case = 0 the first excited state is doubly degenerate

(9.291)
H0 |1, 0 = 2 |1, 0 ,
H0 |0, 1 = 2 |0, 1 .
(9.292)
Thematrix of V in the basis {|1,
0
,
|0,
1}
is
given
by

1, 0| V |1, 0 1, 0| V |0, 1
0, 1| V |1, 0 0, 1| V |0, 1

1, 0| [(1 + Nx ) Ny + Nx (1 + Ny )] |1, 0
0
=
0
0, 1| [(1 + Nx ) Ny + Nx (1 + Ny )] |0, 1

10
=
.
01
(9.293)
Thus to first order in the first excited state remains doubly degenerate with energy 2 (1 + ). Note - The exact solution can be
found using the transformation
1
ad = (ax iay ) ,
(9.294)
2
1
ag = (ax + iay ) .
(9.295)
2
The following holds
3
4 ,
ad , ad = ag , ag = 1 ,

1
1
ax + iay (ax iay ) +
ax iay (ax + iay )
2
2
= ax ax + ay ay ,
(9.296)
ad ad + ag ag =
and

1
1
ax + iay (ax iay )
ax iay (ax + iay )
2
2

= i ax ay ax ay ,
(9.297)
ad ad ag ag =
thus
H0 = (Nd + Ng + 1) ,
(9.298)
V = (Nd Ng ) ,
(9.299)
and the exact eigen vectors and eigenenergies are given by
3
4
(H0 + V ) |nd , ng = nd + ng + 1 + (nd ng )2 |nd , ng .
(9.300)
Eyal Buks
327

14. For V0 = 0 the wavefunctions (0)
n (x) are given by
0
2
nx
(0)
sin
,
n (x) = x |n =
l
l
(9.301)

En(0) =
2 2 n2
.
2ml2
(9.302)
The matrix elements of the perturbation are given by

2V0
n|V |m =
l
l
nx
mx
2x
sin
sin
dx .
l
l
l
sin
(9.303)
For the diagonal case n = m

2V0
n|V |n =
l
2V0
=
l
l
sin2
nx
2x
sin
dx
l
l
(9.304)
l/2
sin2
l/2
2V0
=
l
=0,
l/2
ny
l
1 cos
l/2
n
sin
2
2ny
l
+ n

2y
+ dy
l
(9.305)
2y
dy
l
(9.306)
sin
(9.307)
(9.308)
since the integrand is clearly an odd function of y. Thus to first order in

V0 the energies are unchanged
En =

2 2 n2
+ O V02 .
2ml2
15. For the case = 0 the exact wave functions are given by
0
nx
2
(0)
n (x) =
sin
,
L
L
(9.309)
(9.310)

En(0) =
2 2 n2
,
2mL2
(9.311)
where n is integer. To first order in the energy of the ground state n = 1

is given by
Eyal Buks
328
9.4. Solutions
(0)
E1 = E1 +
dx
0

2

(0)
1 (x) x + O 2

x

2 L
x + O 2
=
+ 2
dx sin2
L 0
L

(0)
= E1 + + O 2
2
(0)
E1
(9.312)
16. For the case = 0 the exact wave functions of the eigenstates are given
by
(0)
nx ,ny (x, y) =
2
nx x
ny y
sin
sin
,
l
l
l

2 2 n2x + n2y
(0)
Enx ,ny =
,
2ml2
(9.313)
(9.314)
where nx and ny are non-zero integers.

a) The ground state is non degenerate thus to 1st order the energy is
given by
l l
2
(0)
(0)
E0 = E1,1 +
1,1 W dxdy
0
2 2
=
ml2
l l
x 2 y
2 2
+ 2 4
sin2
sin
(x lx ) (y ly ) dxdy
ml
l
l
0 0

2 2
lx
ly
=
1 + 4 sin2
sin2
.
ml2
l
l
(9.315)
b) The first excited state is doubly degenerate. The matrix of the perturbation
W in the eigen subspaceis given by
2, 1| W |2, 1 2, 1| W |1, 2
W =
1, 2| W |2, 1 1, 2| W |1, 2

!
"
2 2
sin 2ll x sin ll x sin ll y sin 2ll y
sin2 2ll x sin2 lly
= 4
ml2 sin llx sin 2ll x sin 2ll y sin ll y
sin2 llx sin2 2ll y
!
"
2 2
4 sin2 llx cos2 llx sin2 ll y
4 cos llx sin2 llx cos ll y sin2 ll y
= 4
ml2 4 cos ll x sin2 llx cos ll y sin2 ll y
4 sin2 llx sin2 ll y cos2 lly
!
"
l
162 2 sin2 llx sin2 ly
cos2 llx
cos llx cos lly
=
.
ml2
cos llx cos ll y
cos2 ll y
(9.316)
Eyal Buks
329

The eigenvalues of W are
w1 = 0 ,
(9.317)
and
162 2 sin2
w2 =
lx
l
sin2
ly
l

cos2
lx
l
+ cos2
ly
l
ml2
(9.318)
17. The unperturbed Hamiltonian ( = 0) can be written as

L2 L2z
L2
+ z
2Ixy
2Iz

2
L
1
1
L2z ,
=
+
2Ixy
2Iz
2Ixy
H=
(9.319)
thus the states |l, m (the standard eigenstates of L2 and Lz ) are eigenstates of H and the following holds
H |l, m = El,m |l, m ,
(9.320)
where
2
El,m =
l (l + 1)
+
2Ixy
1
1
2Iz
2Ixy
(9.321)
Since the unperturbed Hamiltonian is positive-definite, it is clear that

the state |l = 0, m = 0 is the (nondegenerate) ground state of the system
since its energy vanishes E0,0 = 0. Using
L+ + L
,
2
L+ L
Ly =
,
2i
one finds that the perturbation term V can be written as
Lx =
V =
L2+ + L2
.
4Ixy
(9.322)
(9.323)
(9.324)
To second order in the energy of the ground state is found using Eq.
(9.32)
E0 = E0,0 +0, 0| V |0, 0+
l ,m
=0,0

|l , m | V |0, 0|2
+O 3 . (9.325)
E0,0 El ,m
Using the relations

Eyal Buks
330
9.4. Solutions

l (l + 1) m (m + 1) |l, m + 1 ,

L |l, m = l (l + 1) m (m 1) |l, m 1 ,
L+ |l, m =
(9.326)
(9.327)
it is easy to see that all terms to second order in vanish, thus

(9.328)
E0 = 0 + O 3 .
18. The Hamiltonian can be written as

H = H1 + H2 + V ,
(9.329)
where
H1 =
p21
(x1 ) ,
2m
(9.330)
H2 =
p22
(x2 ) ,
2m
(9.331)
and
V = (x1 x2 ) .
(9.332)
First consider H1 only. A wavefunction (1) (x1 ) of an eigenstate of H1

must satisfy the following Schrdinger equation

2
d
2m
+
(E
+
(x
))
(1) (x1 ) = 0 .
(9.333)
1
dx21
2
Integrating around x1 = 0 yields the condition
d(1) (0+ ) d(1) (0 ) 2m (1)
+ 2 (0) = 0 .
dx1
dx1
(9.334)
Requiring also that the wavefunction is normalizable leads to

0
m

m
(1)
(x1 ) =
exp
|x
|
.
1
2
2
The corresponding eigenenergy is
(1)
E0 =
m2
.
22
The ground state of H2 can be found in a similar way. Thus, the normalized wavefunction of the only bound state of H1 + H2 , which is obviously
the ground state, is given by
m

m

m
0 (x1 , x2 ) = 2 exp 2 |x1 | exp 2 |x2 | ,
(9.335)

Eyal Buks
331

and the corresponding energy is given by
E0 =
m2
.
2
(9.336)
Therefore, to first order in the energy of the ground state of H is given

by Eq. (9.32)
m2
2

+
dx1
dx2 0 (x1 , x2 ) (x1 x2 ) 0 (x1 , x2 ) + O 2
Egs =
m 2
m
= 2 +

2
2

4m
dx1 exp 2 |x1 | + O 2

m2 m
= 2 +
+ O 2 .

22
(9.337)
19. Substituting the expansions
and

|n = |n0 + |n1 + 2 |n2 + O 3 ,

= n0 + n1 + 2 n2 + O 3 ,
(9.338)
(9.339)
into Eq. (9.96) and collecting terms having the same order in (up to
second order) yield
(D n0 ) |n0 = 0 ,
(D n0 ) |n1 + (V n1 ) |n0 = 0 ,
(D n0 ) |n2 + (V n1 ) |n1 n2 |n0 = 0 .
(9.340)
(9.341)
(9.342)
We further require normalization

n|n = 1 ,
(9.343)
and choose the phase of n0 |n such that

n0 |n R .
(9.344)
Expressing the normalization condition using Eq. (9.338) and collecting

terms having the same order in yield
n0 |n0 = 1 ,
n0 |n1 + n1 |n0 = 0 ,
n0 |n2 + n2 |n0 + n1 |n1 = 0 .
Eyal Buks
(9.345)
(9.346)
(9.347)
332
9.4. Solutions
These results together with Eq. (9.344) yield
n0 |n1 = n1 |n0 = 0 ,
1
n0 |n2 = n2 |n0 = n1 |n1 .
2
(9.348)
(9.349)
Multiplying Eq. (9.341) by m0 | yields

n1 m0 |n0 = (m0 n0 ) m0 |n1 + m0 | V |n0 ,
(9.350)
thus for m = n
n1 = n0 | V |n0 .
(9.351)
Using this result for m = n yields

m0 |n1 =
m0 | V |n0
,
n0 m0
(9.352)
thus with the help of Eq. (9.95) one has

|n1 =
m0 | V |n0
|m0 .
n0 m0
m
(9.353)
Multiplying Eq. (9.342) by n0 | yields

n2 = n0 | V |n1 n1 n0 |n1 ,
(9.354)
or using Eq. (9.353)

n2 =
n0 | V |m0 m0 | V |n0

.
n0 m0
m
(9.355)
Thus, using this result together with Eq. (9.351) one finds
= n0 + n0 | V |n0
n0 | V |m0 m0 | V |n0

+ 2
+ O 3 .
n0 m0
m
(9.356)
20. The condition [9.102] together with Eq. (9.98) can be used to evaluate
the matrix elements of L
k| L |k =
k| V |k
.
Ek Ek
(9.357)

Eyal Buks
333

3
4 1 3 3
44
L, L, H0 + V + . (9.358)
HR = H0 +V +[L, H0 ]+ L, V +
2!
Thus, for the case where condition (9.102) is satisfied the following holds
[note that according to Eq. (9.357) L = O ()]
3
4 1

HR = H0 + L, V + [L, [L, H0 ]] + O 3
2!
4

13
L, V + O 3 .
= H0 +
2
(9.359)
where Eq. (9.102) has been employed in the last step.

<The last result
together with Eq. (9.357) and the closure relation 1 = k |k k | [see
Eq. (2.23)] lead to
3
4

1
k| HR |k = k| H0 |k + k| L, V |k + O 3
2

<
1
1
2 k k| V |k k | V |k Ek E
E E

k
k
k
+ O 3 .
= Ek k,k +
2
(9.360)
21. Consider the Hamiltonian
H=
p2 (1 + ) e2
,
2
r
(9.361)
where is the reduced mass and e is the electron charge. The parameter
is a positive constant. The exact eigenenergies are given by Eq. (7.84)
(1 + )2 e4
.
(9.362)
22 n2
On the other hand, perturbation theory yields the following expansion
[see Eq. (9.32)]
En =

e4
e2
nlm|
|nlm + O 2 .
2
2
2 n
r
By comparing the above results for En one finds that
En =
nlm|
e2
e4
|nlm = 2 2 2 ,
r
2 n
(9.363)
(9.364)
thus (recall that V = e2 /r)

nlm| V |nlm = 2
e4
,
22 n2
(9.365)
and (recall that T + V = H)

nlm| T |nlm =
Eyal Buks
e4
.
22 n2
(9.366)
334
9.4. Solutions
22. The energy eigenvalues Ekl of the radial equation of the hydrogen atom,
2 2

d
e2 l (l + 1) 2
+

(9.367)
2 dr2
r
2r2
where is the reduced mass and e is the electron charge, are [see Eq.
(7.78)]
Ekl =
e4
22 (k + l)2
(9.368)
where k = 1, 2, 3, . The quantum number l can formally be treated as a

real number, which is not restricted to take integer values only. Consider
the case where the integer l is replaced by l + , where 0 1. While
the exact eigenenergies can still be evaluated by Eq. (7.78)
Ekl =
e4
22 (k + l + )2
(9.369)
perturbation theory yields the following expansion [see Eq. (9.32)]

Ekl =
e4
+ klm| VH |klm + ,
22 (k + l)2
(9.370)
where the perturbation VH is given by

[(l + ) (l + + 1) l (l + 1)] 2
2r2
2

(2l + 1)
=
+ O 2 ,
2
2r
VH =
(9.371)
By comparing both results for Ekl one finds that

klm| r2 |klm =
22 e4
4 (2l + 1) (k + l)3
(9.372)
or in terms of the quantum number n = k + l

nlm| r2 |nlm =
2
,
a20 (2l + 1) n3
(9.373)
where
a0 =
Eyal Buks
2
.
e2
(9.374)
335
10. Time-Dependent Perturbation Theory
Recall that the time evolution of a state vector | is governed by the

Schrdinger equation (4.1)
i
d |
= H | ,
dt
(10.1)
where the Hermitian operator H = H is the Hamiltonian of the system. The

time evolution operator u (t, t0 ) [see Eq. (4.4)] relates the state vector | (t0 )
at time t0 with its value | (t) at time t
| (t) = u (t, t0 ) | (t0 ) .
(10.2)
As we have seen in chapter 4, when the Hamiltonian is time independent

u (t, t0 ) is given by

i (t t0 )
u (t, t0 ) = exp
H .
(10.3)

In this chapter we consider the more general case where H is allowed to vary
in time. We first derive a formal expression for the time evolution operator
u (t, t0 ) applicable for general H. Then we present the perturbation theory
expansion of the time evolution operator, and discuss approximation schemes
to evaluate u (t, t0 ).
10.1 Time Evolution

Dividing the time interval (t0 , t) into N sections of equal duration allows
expressing the time evolution operator as
u (t, t0 ) =
N
>
u (tn , tn1 ) ,
(10.4)
n=1
where
tn = t0 + n ,
(10.5)
Chapter 10. Time-Dependent Perturbation Theory

and where
=
t t0
.
N
(10.6)
Furthermore, according to the Schrdinger equation (4.7), the following holds

u (tn1 + , tn1 ) = 1

i
H (tn ) + O 2 .
(10.7)

In the limit where N higher than first order terms in , i.e. O 2
terms, are not expected to contribute, thus the time evolution operator can
be expressed as

N
>
i
1 H (tn ) .
u (t, t0 ) = lim
(10.8)
N n=1
10.2 Perturbation Expansion

Consider the case where
H = H0 +H1 ,
(10.9)
where is real. The perturbation expansion expresses the time evolution

operator u (t, t0 ) of the full Hamiltonian H as

u (t, t0 ) = u0 (t, t0 ) + u1 (t, t0 ) + 2 u2 (t, t0 ) + O 3 ,
(10.10)
where u0 (t, t0 ) is the time evolution of the Hamiltonian H0 . Such an expansion can be very useful for cases where u0 (t, t0 ) can be exactly calculated and
where the parameter is small, i.e. || 1. For such cases only low order
terms in this expansion are needed for approximately evaluating u (t, t0 ).
By employing Eq. (10.8)

N
>
i
u (t, t0 ) = lim
1 (H0 (tn ) + H1 (tn )) ,
(10.11)
N n=1

one easily obtains the terms u0 , u1 and u2

N
>
i
u0 (t, t0 ) = lim
1 H0 (tn ) ,
N n=1
N i
<
u0 (t, tn ) H1 (tn ) u0 (tn , t0 )
N n=1
i /t
=
dt u0 (t, t ) H1 (t ) u0 (t , t0 ) ,
t0
(10.12)
u1 (t, t0 ) = lim
Eyal Buks
(10.13)
338
10.2. Perturbation Expansion

and
u2 (t, t0 ) = lim
N1
<
N
<
N n=1 m=n+1
2
u0 (t, tn ) H1 (tn ) u0 (tn , tm ) H1 (tm ) u0 (tm , t0 )

=
1 /t /t
dt
dt
2 t0
t0
u0 (t, t ) H1 (t ) u0 (t , t ) H1 (t ) u0 (t , t0 ) .
The expansion can be expressed as

u (t, t0 ) = u0 (t, t0 ) O (t) + O 3 ,
(10.14)
(10.15)
where the operator O (t) is given by

O (t) = 1
/t
i /t
2 /t
dt H1I (t ) 2 dt dt H1I (t ) H1I (t ) ,
t0
t0
t0
(10.16)
and where H1I (t), which is defined by
H1I (t) u0 (t, t0 ) H1 (t) u0 (t, t0 ) ,
(10.17)
is the so called interaction representation of H1 with respect to u0 .

Exercise 10.2.1. Calculate the expectation value squared |O (t)|2 to lowest nonvanishing order in .
Solution 10.2.1. Since H1 (t) is Hermitian one finds that
|O (t)|2
!
"
i /t
2 /t /t
= 1
dt H1I (t ) 2 dt dt H1I (t ) H1I (t )
t0
t0
t0
!
"
i /t
2 /t /t
1+
dt H1I (t ) 2 dt dt H1I (t ) H1I (t )
t0
t0
t0
!
"2
2 /t
= 1+ 2
dt H1I (t )

t0
2 /t /t
2 dt dt (H1I (t ) H1I (t ) + H1I (t ) H1I (t ))
t0
t0
3
+O ,
Eyal Buks
(10.18)
339

or
2
|O (t)| = 1 + 2

2
/t
t0
"2
dt H1I (t )
2 /t /t
dt dt H1I (t ) H1I (t ) ,
2 t0
t0
(10.19)
thus
2
|O (t)| = 1
2 /t /t
dt dt
2 t0
t0
[H1I (t ) H1I (t ) H1I (t ) H1I (t )] ,
(10.20)
or
2
|O (t)| = 1
where
2 /t /t
dt dt H1I (t ) H1I (t ) ,
2 t0
t0
H1I (t) = H1I (t) H1I (t) .
(10.21)
(10.22)
10.3 Transition Probability

Consider the case where the unperturbed Hamiltonian H0 is time independent. The eigenvectors of H0 are denoted as |an , and the corresponding
eigenenergies are denoted as En
H0 |an = En |an ,
(10.23)
where
an |an = nn .
(10.24)
In this basis u0 (t, t0 ) is given by

i (t t0 )
iEn (t t0 )
u0 (t, t0 ) = exp
H0 =
exp
|an an | .

n
(10.25)
Assuming that initially at time t0 the system is in state |an , what is the
probability to find it later at time t > t0 in the state |am ? The answer to
this question is the transition probability pnm , which is given by
Eyal Buks
340
10.3. Transition Probability

pnm = |am | u (t, t0 ) |an |2 .
(10.26)

e
iEm (tt0 )
am | u (t, t0 ) |an = am | O (t) |an

i /t
dt am | H1I (t ) |an
= nm
t0
2 /t /t
dt dt am | H1I (t ) H1I (t ) |an
2 t0
t0
3
+O ,
(10.27)
thus
i /t
=
nm
dt am | H1I (t ) |an
t0
pnm
3
2 /t /t
2 dt dt am | H1I (t ) H1I (t ) |an + O

.
t0
t0
(10.28)
In what follows, we calculate the transition probability pnm to lowest nonvanishing order in for the case where n = m, for which the dominant
contribution comes from the term of order in Eq. (10.28). For simplicity
the initial time t0 , at which the perturbation is turned on, is taken to be zero,
i.e. t0 = 0. We consider below the following cases:
10.3.1 The Stationary Case
In this case H1 is assumed to be time independent (after being turned on at
t0 = 0). To lowest nonvanishing order in Eq. (10.28) yields
2
= 2
pnm
where
mn =
/ i t
2
dt e mn
|am | H1 |an |2 ,
(10.29)
Em En
.
(10.30)
Using the identity

/t
it
dt e
i t
2
= 2e
sin
t
2
(10.31)
one finds that

Eyal Buks
341

pnm =
t
4 sin2 mn
2
2
|am | H1 |an | .
2
2

mn
(10.32)
Note that in the limit t one finds with the help of Eq. (10.31) that

/ it
2
4 sin2 t
2
lim
=
lim
dt
t
t
0
2
/t
/t
= lim dt dt ei(t t )
t 0
/t
= 2 () dt
0
= 2t () .
(10.33)
In this limit pnm is proportional to the time t, i.e. pnm can be expressed as
pnm = wnm t, where wnm is the transition rate, which is given by
wnm =
2
( mn ) |am | H1 |an |2 .
2
(10.34)
The delta function (mn ) ensures that energy is conserved in the limit
of long time, and transitions between states having different energies are
excluded. However, such transitions have finite probability to occur for any
finite time interval t. On the other hand, as can be see from Eq. (10.32)
(see also the figure below, which plots the function f (x) = sin2 x/x2 ), the
probability is significant only when mn t 1, or alternatively when
Et ,
(10.35)
where E = mn .
0.8
0.6
0.4
0.2
-10
-8
-6
-4
-2
4 x 6
10
The function f (x) = sin x/x2 .
Eyal Buks
342

10.3.2 The Near-Resonance Case
In this case H1 is assumed to be given by
H1 (t ) = Keit + K eit ,
(10.36)
where K is an operator that is assumed to be time independent (after being

turned on at t0 = 0), and where the angular frequency is a positive constant.
The transition probability is given by [see Eq. (10.28)]

(mn )t
(mn )t
ei
2
sin
am | K |an
2
4
pnm = 2
mn

2
( mn +)t
2
ei
sin (mn2+)t am | K |an
,
+
mn +
(10.37)
We refer to the case where = mn as absorption resonance, and to the

case where = mn as stimulated emission resonance. Near any of these
resonances mn the dominant contribution to pnm comes from only
one out of the two terms in Eq. (10.37), thus
)t
2 (
42 sin mn2 2 |am | K |an |2 mn

(mn )
.
pnm
(10.38)
(
+)t
4 sin2 mn2 2
a | K |a
2
(mn +)
mn
In the long time limit, i.e. in the limit t , the probability pnm is
found using Eq. (10.33) to be proportional to the time t, i.e. pnm = wnm t,
where the transition rate wnm is given by
?
2
2
mn
2 ( mn ) |am | K |an |
2
wnm 2
(10.39)
mn
.
2 ( mn + ) am | K |an
In many cases of interest the final state |am lie in a band of dense states. Let
wn be the total transition rate from the initial state |an . Assume that the
matrix element am | K |an does not vary significantly as a function of the
energy Em . For this case the total rate wn can be expressed in terms of the
density of states g (Em ) (i.e. number of states per unit energy) in the vicinity
of the final state |am [see Eq. (10.39)] as
wn =
2
g (Em ) |am | K |an |2 ,
(10.40)
where Em = En + . This result is known as the Fermis golden rule.
Eyal Buks
343

10.3.3 H1 is Separable
For this case it is assumed that H1 can be expressed as
1 ,
H1 (t ) = f (t ) H
(10.41)
1 is time independent
where f (t ) is a real function of time and where H
Hermitian operator. To lowest nonvanishing order in Eq. (10.28) yields
pnm
/t imn t

1 |an
2 .
= 2
dt e
f (t )
am | H
0
(10.42)
10.4 Problems
1. Find the exact time evolution operator u (t, 0) of the Hamiltonian H,
which is given by
H = H0 + Hp ,
(10.43)
where
H0 = a a ,

Hp = i (t) e2i(t) a2 e2i(t) a2 ,
(10.44)
(10.45)
a and a are the annihilation and creation operators (as defined in chapter
5), is positive, is real and (t) is an arbitrary real function of time t.
2. Consider a particle having mass m moving under the influence of a one
dimensional potential given by
V (x) =
m20 x2
,
2
(10.46)
where the angular resonance frequency 0 is a constant. A perturbation

given by
H1 (t ) = 2x cos (t ) ,
(10.47)
where the real constant is assumed to be small, is turned on at time

t = 0. Given that the system was initially at time t = 0 in the ground
state |0 of the unperturbed Hamiltonian, calculate the transition probability pn0 (t) to the number state |n to lowest nonvanishing order in the
perturbation expansion.
Eyal Buks
344
10.4. Problems
3. Repeat the previous exercise with the perturbation
H1 (t ) = xf (t ) ,
where the force f (t ) is given by
2
exp t 2
f (t ) =
,
(10.48)
(10.49)
and where both and are real. Given that the system was initially at
time t in the ground state |0 of the unperturbed Hamiltonian,
find the transition probability pn0 to the number state |n in the limit
t . Compare your approximated result with the exact result given
by Eq. (5.339).
4. Consider a spin 1/2 particle. The Hamiltonian is given by
H = Sx ,
(10.50)
where is a Larmor frequency and where Sx is the x component of the

angular momentum operator. Given that the spin is initially at time t = 0
in the eigenstate |+;
z of the operator Sz (having eigenvalue +/2), what
is the probability p++ (t) to find the spin at the same state |+;
z at a
later time t. Compare the exact result with the approximated value that
is obtained from Eq. (10.21).
5. Consider a particle having mass m confined in a potential well given by
+
0
if 0 x a
V (x) =
.
(10.51)
if x < 0 or x > a
The particle is initially at time t in the ground state of the well.
A small perturbation
H1 (t) =
( t )2
xe
,
a
(10.52)
where 1 and where is a positive constant having the dimensions

of time, is applied. Calculate the probability to find the particle in the
first excited state in the limit t .
6. Consider the transition between the energy eigenstates |an and |am of
the unperturbed Hamiltonian H0 , which is assumed to be time indepen
dent, due to harmonic perturbation given by H1 (t ) = Keit + K eit

[see Eq. (10.36)], where K is an operator that is assumed to be time independent (after being turned on at t0 = 0). Calculate to second order in
perturbation theory the transition rate wnm in the long time limit for the
case where the first order contribution vanishes, i.e. for the case where
am | K |an = 0.
Eyal Buks
345
10.5 Solutions
1. Expressing the ket vector state as
| = eiH0 t/ |I ,
(10.53)
and substituting into the Schrdinger equation, which is given by

i
d |
= (H0 + Hp ) | ,
dt
(10.54)
d |I
= HI |I .
dt
(10.55)
yield
i
where HI , which is given by

HI = eiH0 t/ Hp eiH0 t/ ,
(10.56)
is the interaction picture representation of Hp . With the help of the vector

identity (2.173), which is given by
eL AeL = A + [L, A] +
1
1
[L, [L, A]] + [L, [L, [L, A]]] + , (10.57)
2!
3!
and the relations

it ,
H0 , a2 = 2ita2 ,
(10.58)
and
it ,
H0 , a2 = 2ita2 ,
(10.59)
one finds that
eiH0 t/ a2 eiH0 t/ = a2 e2it ,
(10.60)
iH0 t/ 2 iH0 t/
(10.61)
a e
2 2it
=a e
thus

HI = i (t) e2i a2 e2i a2 .
Since [HI (t) , HI (t )] = 0 the solution of Eq. (10.55) is given by

i t
|I (t) = exp
dt HI (t ) |I (0)
0
= S (, ) |I (0) ,
Eyal Buks
(10.62)
(10.63)
346
10.5. Solutions
where
,
S (, ) = exp e2i a2 e2i a2 ,
and where
t
=
dt (t ) ,
(10.64)
(10.65)
and thus the time evolution operator is thus given by

u (t, 0) = eiH0 t/ S (, ) .
(10.66)
2. To lowest nonvanishing order in the perturbation expansion one finds

using Eq. (10.38) together with Eqs. (5.11), (5.28) and (5.29) that
pn0 (t) =
( )t
22 sin2 0 2
n,1 .
m ( 0 )2
(10.67)
3. To lowest nonvanishing order in perturbation expansion Eq. (10.42) yields

pn0 = n,1 ,
where
(10.68)
1
i0 t

dt
e
f
(t
)
2m 0
2
1 2 2
=
e 2 0 .
2m 0
(10.69)
The exact result is found from Eq. (5.339)
e n
.
n!
To first order in both results agree.
4. While the exact result is [see Eq. (6.305)]
pn =
p++ (t) = cos2
t
,
2
(10.70)
(10.71)
Eq. (10.21) yields

1 /t /t
dt dt +;
z| (Sx +;
z| Sx |+;
z)2 |+;
z
2 0
0
2 +;
z| Sx2 |+;
z /t /t
= 1
dt dt
2
0
0
2
t
= 1
.
2
(10.72)
p++ (t) = 1
Eyal Buks
347

5. The normalized wavefunctions n (x ) of the wells energy eigenstates are
given by
0
2
nx
n (x ) =
sin
,
(10.73)
a
a
2 2 n2
,
2ma2
En =
(10.74)
where n = 1, 2, . The transition probability is calculated to lowest

with the help of Eq. (10.42)
nonvanishing order in
2
t
/ i(E E )t / t 2
2
2
1
dt e
e
p2,1 =
a
0
2 /
2x
x
dx x sin
sin
,
a 0
a
a
(10.75)
thus [see Eq. (5.136)

2562
94 2 2
exp
.
813
8m2 a4
p2,1 =
(10.76)
6. For the present case to second order in Eq. (10.27) becomes [see Eq.
(10.25)]
e
iEm (tt0 )
am | u (t, t0 ) |an

t

/
/t
= 2
dt dt eiml t +iln t am | H1 (t) |al al | H1 (t) |an ,
l 0
0
2
(10.77)
where
mn =
Em En
,
(10.78)
or
Eyal Buks
348
10.5. Solutions
iEm (tt0 )

e
am | u (t, t0 ) |an
2
Kml Kln /t
/t i(ml )t +i(ln )t
=
dt
dt e
2
0
0
2 K Kln /t
/t i( +)t +i( )t
lm
ln
dt dt e ml
2
0
0
l
2 Kml K /t
/t i(ml )t +i(ln +)t
nl
dt
dt e
2
0
0
l
2 K K /t
/t i(ml +)t +i(ln +)t
lm nl
dt dt e
,
2
0
0
(10.79)
where
Kln = al | K |an .
(10.80)
With the help of the identity

/t
dt
/t
dt ei1 t +i2 t =
/t
one finds that
dt
ei(1 +2 )t ei1 t
,
i2
(10.81)
iEm (tt0 )

e
am | u (t, t0 ) |an

2 Kml Kln /t
i(mn 2)t
i(ml )t
=
dt
e
e
i2 ( ln ) 0
l

2 K Kln /t
imn t
i(ml +)t
lm
dt
e
e
i2 ( ln ) 0
l

2 Kml K /t
imn t
i(ml )t
nl
dt
e
e
i2 ( ln + ) 0
l

2 K K /t
i(mn +2)t
i(ln +)t
lm nl
dt
e
e
,
i2 ( ln + ) 0
(10.82)
or
Eyal Buks
349

iEm (tt0 )

e
am | u (t, t0 ) |an
2
Kml Kln /t
dt ei(mn 2)t
=
2
i ( ln ) 0
l
2 K K /t
lm nl
dt ei(mn +2)t
2
i ( ln + ) 0
l

Kml K
2
Kln
nl
+
Kml
/t i(ml )t
ln +
ln
+
dt e
i2
0
l

Kml K
2 Klm Kln
nl
+

/t imn t
ln
ln +
dt e
2
i
0
2 K Kln /t
lm
+
dt ei(ml +)t
i2 ( ln ) 0
l
2 K K /t
lm nl
+
dt ei(ln +)t .
2
i ( ln + ) 0
l
(10.83)
By employing the identity [see Eq. (10.33)]
/ it
2
lim
e
dt
= 2t () ,
t
(10.84)
the transition rate wnm can be evaluated in the long time limit. To that
end it is assumed that = ml and = ln (i.e. it is assumed that the
harmonic perturbation is not at resonance with any first order transition
between the initial |an or final |am states and an intermediate state
|al ), and it is further assumed that mn > 0 and that 0. Under
/t
these assumptions only the terms proportional to 0 dt ei(mn 2)t in

Eq. (10.83) are taken into account, and consequently the transition rate
wnm becomes
2
2
2 Kml Kln
wnm = 4
(10.85)
(mn 2) .

(ln )
Eyal Buks
350
11. WKB Approximation
The theory of geometrical optics provides an approximated solution to

Maxwells equation that is valid for systems whose typical size scales are
much larger than the wavelength of electromagnetic waves. In 1926 using a
similar approach the physicists Wentzel, Kramers and Brillouin (WKB) independently found an approximated solution to the Schrdinger equation in
the coordinate representation for the case where the wavelength associated
with the wavefunction (to be defined below) can be considered as short. Below the WKB approximation is discussed for the time independent and one
dimensional case. This chapter is mainly based on Ref. [3].
11.1 WKB Wavefunction

Consider a point particle having mass m moving under the influence of a onedimensional potential V (x). The time independent Schrdinger equation for
the wavefunction (x) is given by [see Eq. (4.50)]
d2 (x) 2m
+ 2 (E V (x)) (x) = 0 ,
dx2
(11.1)
where E is the energy. In terms of the local momentum p (x), which is defined
by

p (x) = 2m (E V (x)) ,
(11.2)
the Schrdinger equation becomes
d2 (x) p 2
+
(x) = 0 .
dx2
(11.3)
Using the notations
(x) = eiW (x)/ ,

and the relation
!

2 "
d2 (x)
i d2 W
1 dW
=
(x) ,
dx2
dx2
dx
(11.4)
(11.5)
Chapter 11. WKB Approximation

one finds that the Schrdinger equation can be written as

2
dW
d2 W
i 2
+ p2 = 0 .
dx
dx
(11.6)
In the WKB approach the Planks constant is treated as a small parameter. Expanding W as a power series in
W = W0 + W1 + 2 W2 +
one finds that

2

dW0
d2 W0
dW0 dW1
+ p2 + O 2 = 0 .
+ i
2
2
dx
dx
dx dx
The terms of order zero in yield
2

dW0
+ p2 = 0 .
dx
(11.7)
(11.8)
(11.9)
thus
W0 (x) =
x
dx p (x ) ,
(11.10)
x0
where x0 is a constant.
What is the range of validity of the zero order approximation? As can be
seen by comparing Eq. (11.6) with Eq. (11.9), the approximation W W0
is valid when the first term in Eq. (11.6) is negligibly small in absolute value
in comparison with the second one, namely when
d W
dW
2
.

(11.11)
dx
dx
It is useful to express this condition in terms of the local wavelength (x),

which is given by
(x) =
2
.
p (x)
(11.12)
By employing the lowest order approximation dW/dx = p the condition

(11.11) becomes

2 .
(11.13)
dx
This means that the approximation is valid provided that the change in
wavelength over a distance of one wavelength is small.
Eyal Buks
352
11.1. WKB Wavefunction

The terms of 1st order in of Eq. (11.8) yield an equation for W1
dW1
i
=
dx
2
d2 W0
dx2
dW0
dx
i d
log
2 dx
dW0
dx
Using Eq. (11.9) one thus has

d
1
iW1 log
=0.
dx
p
(11.14)
(11.15)
Therefor, to 1st order in the wave function is given by

(x) = C+ + (x) + C (x) ,
(11.16)
where
1
i
(x) = exp
p
x
x0
dx p (x ) ,
(11.17)
and where both C+ and C are constants.

In general, the continuity equation (4.73), which is given by
d dJ
+
=0,
dt
dx
(11.18)
relates the probability distribution function = ||2 and the current density
J = (/m) Im ( d/dx) associated with a given one dimensional wavefunction (x). For a stationary (x) the probability distribution function
is time independent, and thus J is a constant. Consider a region where
E > V (x). In such a region, which is classically accessible, the momentum
p (x) is real and positive, and thus the probability distribution function (x)
of the WKB wavefunctions (x) is proportional to 1/p. This is exactly
what is expected from a classical analysis of the dynamics, where the time
spent near a point x is inversely proportional to the local classical velocity at
that point v (x) = p (x) /m. With the help of Eq. (4.200) one finds that the
current density J associated with the wavefunction (11.16) is given by

d+
d

J=
Im C+ + + C C+
+ C
m
dx
dx

d
d

2
2
+
=
|C+ | Im +
+ |C | Im
m
dx
dx

d
d

+ Im C+
C + + C+ C
+
.
dx
dx
(11.19)
As can be seen from Eq. (11.17), the last term vanishes since (x) = + (x).
Therefor, with the help of Eq. (6.361) one finds that
Eyal Buks
353

J=

1
|C+ |2 |C |2 .
m
(11.20)
Thus, the current density J associated with the state + (x) is positive,
whereas J < 0 for (x). Namely, + (x) describes a state propagating from
left to right, whereas (x) describes a state propagating in the opposite
direction.
11.2 Turning Point

Consider a point x = a for which E = V (a), namely p (a) = 0 [see Fig.
11.1 (a)]. Such a point is called a turning point since a classical particle that
reaches the point x = a momentarily stops and changes its direction. Near
a turning point the local wavelength diverges, and consequently, as can be
seen from Eq. (11.13), the WKB approximation breaks down. Consider the
case where E > V (x) for x > a and where E < V (x) for x < a. In the region
x > a the WKB wave function is expressed using Eq. (11.16), where, for
convenient, the constant x0 is chosen to be a. However, on the other side of
the turning point, namely for x < a, the momentum p (x) becomes imaginary
since E < V (x). Thus, in this region, which is classically forbidden, the wave
function given by Eq. (11.16) contains one exponentially decaying term in
the limit x and another exponentially diverging term in the same
limit. To ensure that the wavefunction remains normalizable, the coefficient
of the exponentially diverging term is required to vanish, and thus we seek a
solution having the form
x

C
1
exp dx |p|
x<a
a
x|p|

(x) =
.
(11.21)
x
/
/
C
C
+p exp i dx p + p exp i dx p x > a

a
Note that the pre-factor 1/ p in the classically forbidden region x < a is

substituted in Eq. (11.21) by 1/ |p|. The ratio between these two factors
in the region x < a is a constant, which is assumed to be absorbed by the
constant C. For given value of C, what are the values of C+ and C ? It
should be kept in mine that Eq. (11.21) becomes invalid close to the turning
point x = a where the WKB approximation breaks down. Thus, this question
cannot be simply answered by requiring that (x ) and its first derivative
are continuous at x = a [e.g., see Eq. (4.127)].
As we have seen above, the WKB approximation breaks down near the
turning point x = a. However the two regions x < a and x > a can be tailored
together by the technique of analytical continuation. In the vicinity of the
turning point, namely for x a, the potential V (x) can be approximated by
V (x) V (a) (x a) ,
Eyal Buks
(11.22)
354
11.2. Turning Point
Im( x )
V (x ) E
(a)
(b)
x
a
Im( x )
Im( x )
(c)
+
a
Re( x )
(d)
Re( x )
Re( x )
Fig. 11.1. (a) The turning point at x = a. (b) The integration trajectory . The
singly connected region + (c) and (d).
where
dV
=
,
dx
x=a
and thus for x a

p (x) 2m (x a) .
(11.23)
(11.24)
Formally, the coordinate x can be considered as complex. Consider a circle

in the complex plane centered at x = a having a radius . The radius > 0
is assumed to be sufficiently large to ensure the validity of the WKB approximation outside it. On the other hand, it is also assumed to be sufficiently
small to allow the employment of the approximation (11.24), namely, for any
point on that circle
x = a + ei ,
(11.25)
where is real, it is assumed that

p (x) 2mei/2 .
Eyal Buks
(11.26)
355

We consider below analytical continuation of the wavefunction given by Eq.
(11.21) for the case x > a into a region in the complex plane. Such a region
must excludes the vicinity of the turning point x = a where the WKB approximation breaks down and in addition it is required to be singly connected to
allow analytical continuation. Two such regions are considered below, the first
one, which is labeled as + (see Fig. 11.1 (c)), excludes the circle |x a|
and also excludes the negative imaginary line x = a ib, whereas the second one, which is labeled as (see Fig. 11.1 (d)), also excludes the circle
|x a| and in addition excludes the positive imaginary line x = a + ib,
where in both cases the parameter b is assumed to be real and positive.
To perform the tailoring it is convenient to define the term

i
I () =
dx p ,
where the integration trajectory [see Fig. 11.1 (b)] contains two sections,
the first along the real axis from x = a to x = a + and the second along the
arc x = a + ei from = 0 to = . With the help of the approximation

(11.26) one finds that

i 2m
I () =
d + i3/2 d ei3 /2

0
0
3
2 i 1
2i
e
i 2m 2 3/2
=
i3/2

3
3
2i 2m3/2 32 i
=
e
3
2 2m3/2 i((1 12 )+ 32 )
=
,
e
3
(11.27)
thus
2 2m3/2
I () =
,
(11.28)
3 3/2
2 2m
I () =
.
(11.29)
3
The last result allows expressing the analytical continuation of the wavefunction given by Eq. (11.21) for the case x > a and evaluate its value at the
point x = a . For the case where the singly connected region + ( ) is
employed, this is done using integration along the trajectory ( ), and
the result is labeled as + (a ) [ (a )]
Eyal Buks
356
11.2. Turning Point
+ (a ) =
(a ) =
C+ exp
2 2m3/2
3

3/2
+ C exp 2 2m
3
(2m)1/4 ei/4

3/2
2 2m3/2
+
C
C+ exp 2 2m
exp
3
3
(2m)1/4 ei/4
(11.30)
(11.31)
Note that the denominators of Eqs. (11.30) and (11.31) are evaluated by
analytical continuation of the factor p [see Eq. (11.26)] along the trajectories
and respectively. On the other hand, according to Eq. (11.21) in the
region x < a one finds by integration along the real axis that
!
"
3
C
2 2m 2
(a ) =
exp
.
(11.32)
3
(2m)1/4
Comparing Eqs. (11.30) and (11.31) with Eq. (11.32) shows that for each of
the two choices + and the analytical continuation yields one exponential
term having the same form as the one in Eq. (11.32), and another one, which
diverges in the limit x . Excluding the diverging terms one finds that
continuity of the non diverging term requires that
C=
C+
i/4
e
C
,
ei/4
and thus the tailored wavefunction is given by

x

C
1
|p| exp dx |p| x < a

xa

(x) =
.
/
2C
1
p cos dx p 4 x > a
(11.33)
(11.34)
The fact that analytical continuation of the wavefunction in the region x < a
along the trajectory ( ) yields only the right to left (left to right) propagating term in the region x > a, and the other term is getting lost along the
way, can be attributed to the limited accuracy of the WKB approximation.
As can be seen from Eq. (11.27), along the integration trajectory near the
point = /3 one term becomes exponentially larger than the other, and
consequently, within the accuracy of this approximation the small term gets
lost.
It is important to keep in mind that the above result (11.34) is obtained
by assuming a particular form for the solution in the region x < a, namely
by assuming that in the classically forbidden region the coefficient of the
exponentially diverging term vanishes. This tailoring role will be employed in
the next section that deals with bound states in a classically accessible region
between two turning points [see Fig. 11.2(a)]. On the other hand, a modified
tailoring role will be needed when dealing with quantum tunneling. For this
case, which will be discussed below, we seek a wave function having the form
Eyal Buks
357
V (x ) E
V (x ) E
(a)
(b)
x
a
x
a
Fig. 11.2. The region a x b bounded by the two turning points at x = a

and x = b is classically accessible in panel (a), whereas it is classically forbidden in
panel (b) .
x

/
/x
C+
C
1
1
|p| exp dx |p| + |p| exp dx |p| x < a

a
x
a
(x) =
. (11.35)
/
C
i
i
exp
dx
p
+
x
>
a
p

4
a
Thus, in this case only the term describing propagation from left to right
is kept in the region x > a, and the coefficient of the other term in that
region that describes propagation in the opposite direction is assumed to
vanish. Using the same tailoring technique as in the previous case one find
that C+ = 0 and C = C, and thus

/x
C
1
exp
dx
|p|
x<a
|p|

x a

(x) =
.
(11.36)
/
C
i
i
exp
dx
p
+
x
>
a
p

4
a
11.3 Bohr-Sommerfeld Quantization Rule

Consider a classical accessible region a x b bounded by two turning
points at x = a and x = b, namely, consider the case where E > V (x) for
a x b and where E < V (x) for x < a and for x > b [see Fig. 11.2(a)].
We seek a normalizable solution, thus the wave function in the classically
forbidden regions x < a and for x > b is assumed to vanish in the limit
x . Employing the tailoring role (11.34) with respect to the turning
point at x = a yields the following wave function for the region a x b
x

2Ca
1
a (x) = cos
dx p ,
(11.37)
p

4
a
Eyal Buks
358
11.3. Bohr-Sommerfeld Quantization Rule

where Ca is a constant. Similarly, employing the tailoring role (11.34) with
respect to the turning point at x = b yields
b

2Cb
1
(11.38)
b (x) = cos
dx p .
p

4
x
The requirement a (x) = b (x) can be satisfied for any x in the region
a x b only if
1
b
dx p =
+ n .
2
(11.39)
where n is integer. Alternatively, this result, which is known as BohrSommerfeld quantization rule, can be expressed as
@
1
1
(11.40)
dx p = n + ,
2
2
where
@
dx p = 2
b
dx p .
(11.41)
To normalize the wavefunction a (x) = b (x) we assume that (a) only

the accessible region a x b contributes, since outside this region the
wavefunction exponentially decays; and (b) in the limit of large n the cosine term rapidly oscillates and therefore the average of its squared value is
approximately 1/2. Applying these assumptions to a (x), which is given by
Eq. (11.37), implies that
1
b
a
dx |a (x)| 2 |Ca |
b
a
dx
.
p
(11.42)
Note that the time period T of classical oscillations between the turning
points x = a and x = b is given by
T =2
b
a
dx
.
v
(11.43)
where v (x) = p (x) /m is the local classical velocity. Thus, by choosing the
pre-factor to be real, one finds that the normalized wavefunction is given by
x
0
m
1
(x) = 2
cos
dx p .
(11.44)
pT

4
a
Eyal Buks
359

The Bohr-Sommerfeld quantization rule (11.40) can be used to relate the
classical time period T with the energy spacing E = En+1 En between
consecutive quantum eigenenergies. As can be seen from the validity condition
of the WKB approximation (11.13), the integer n is required to be large to
ensure the validity of Eq. (11.40). In this limit E E, and thus by taking
the derivative of Eq. (11.40) with respect to energy one finds that

1
@
E
E dx
= 2 .
(11.45)
p
In classical mechanics E/p is the velocity of the particle v, therefor

1
@
E
dx
=T ,
(11.46)
p
thus
E =
2
.
T
(11.47)
11.4 Tunneling
In this case we consider a classical forbidden region a x b bounded by
two turning points at x = a and x = b, namely, it is assumed that E < V (x)
for a x b and E > V (x) for x < a and for x > b [see Fig. 11.2(b)].
In classical mechanics a particle cannot penetrate into the potential barrier
in the region a x b, however such a process is possible in quantum
mechanics. Consider a solution having the form
x

/
/x
1
i
r
i
exp
dx p + p exp dx p
x<a
p
xa

ax

/
/
C+
C
exp 1 dx |p| +
exp 1 dx |p| a x b ,
(x) = |p|
|p|
x
b
t
i
i
exp
dx p + 4
x>b
p

b
where we have introduced the transmission and reflection coefficients t and

r respectively. Such a solution describes an incident wave of unit amplitude
propagating in the region x < a from left to right, a reflected wave having
amplitude r in the same region, and a transmitted wave having amplitude t
in the opposite side of the barrier x > b.
Employing the tailoring role (11.36) yields C+ = 0 and C = t. Moreover,
employing the tailoring role (11.34) and using the identity
x
x
1
1
dx |p| = 1/2 exp
dx |p| ,
(11.48)
exp

b
Eyal Buks
360
11.6. Solutions
where
2
= exp
b
a
dx |p| ,
yield |t| 1/2 = 1, thus the transmission probability is given by
b
2
2
|t| = = exp
dx |p| .
(11.49)
(11.50)
It is important to keep in mind that this approximation is valid only

when 1. One way of seeing this is by noticing that the second tailoring
step, as can be seen from Eq. (11.34), also leads to the conclusion that |r| =
1. This apparently contradicts Eq. (11.50), which predicts a nonvanishing
value for |t|, whereas current conservations, on the other hand, requires that
|t|2 +|r|2 = 1 [see Eq. (4.202)]. This apparent contradiction can be attributed
to limited accuracy of the WKB approximation, however, Eq. (11.50) can be
considered to be a good approximation only provided that 1.
11.5 Problems
1. Calculate the transmission probability of a particle having mass m
and energy E through the potential barrier V (x), which vanishes in the
region x < 0 and which is given by V (x) = U ax in the region x 0,
where a > 0 and where U > E.
2. Calculate the transmission probability of a particle having mass m and
energy E through the potential barrier V (x) = m 2 x2 /2, where > 0.
Consider the general case without assuming 1.
3. Consider a particle having mass m moving in a one dimensional double
well potential (see Fig. 11.3), which is assumed to be symmetric, i.e.
V (x) = V (x). In the limit where the barrier separating the two wells
can be considered as impenetrable, each well is characterized by a set of
eigenstates having eigenenergies {En }. To lowest nonvanishing order in
the penetrability of the barrier calculate the eigenenergies of the system.
4. Employ the WKB approximation to derive the eigenenergies of the hydrogen atom.
11.6 Solutions
1. The classical turning points are x = 0 and x = (U E) /. Thus with
the help of Eq. (11.50) one finds that
Eyal Buks
361
V (x )
En
Fig. 11.3. Double well potential.
2 2m
= exp

!
(U E)/
dx
"
3
4 2m
= exp
(U E) 2 .
3
U E
(11.51)
2. The factor p/ can be expressed as
.

2 2
2m E + m2 x
p (x)
=
=

:
=
x
x20
1+
2Ex20
,
E0 x2
(11.52)

where x0 = /m and where E0 = . For sufficiently large |x|, namely
for x2 Ex20 /E0 , one has
p (x)
x
E
2+
,

x0 E0 x
(11.53)
where x is assumed to be positive. The corresponding WKB wavefunctions (11.17) in the same limit of large large |x| are given (up to multiplication by a constant) by
Eyal Buks
362
11.6. Solutions
x
1
i
dx p (x )
(x) =
exp

x0 p/
!

/x
/x
i
exp x2 dx x exp i EE0
0
x0
dx
x
1/2
1/4
2Ex20
x
1
+
2
E0 x2
x0
i EE 12

0
x
ix2
exp 2 .
x0
2x0
"
(11.54)
Consider a solution having the asymptotic form
(x ) =
x
x0
i EE
12

i EE 12
2
0
ix2
ix
exp 2x
+ r xx0
exp 2x
x
2
2
0
,
i EE0 12
2
0
ix
t xx0
exp 2x
x
2
0
(11.55)
where t and r are transmission and reflection coefficients respectively,

which can be related one to another by the technique of analytical continuation. Consider x as a complex variable
x
= ei ,
x0
(11.56)
where > 0 and is real. The transmitted term in the limit x

along the upper semicircle x/x0 = ei , where 0 is given by
2 2i
i i EE 12
i e
0
t e
exp
,
(11.57)
2
thus for = this term becomes identical to the reflected term at x/x0 =
, which is given by
2
i
i EE 12
0
r ()
exp
,
(11.58)
2
provided that
or

i E 1
t ei E0 2 exp
i2 e2i
2
= r ()i E0 2 exp
ite E0 = r .
Eyal Buks
i2
2
(11.59)
(11.60)
363

Moreover, current conservation requires that |t|2 + |r|2 = 1, thus
2
E
|t| +
ite E0
= 1 ,
(11.61)
2
and therefor the transmission probability = |t| is given by
1
1 + e
2E
E0
(11.62)
As we have seen above, the analytical continuation of the transmitted

term in the region x leads to the reflected term in the region
x . What about the incident term in the region x (the first
term)? Note that this term (the incident one) becomes exponentially
small compared with the reflected term in a section near =
3/4 along

the upper semicircle [due to the exponential factors exp ix2 /2x20 ].
Consequently, within the accuracy of the WKB approximation it does
not contribute to the analytically continued value.
3. Consider one of the single-well eigenenergies En . The associated eigenstate corresponding to the left well is labeled as |n, L and the one corresponding to the right well as |n, R. The effect of finite penetrability of
the barrier can be evaluated using time independent perturbation theory
for the degenerate case [see Eq. (9.38]. For the unperturbed case, where
the barrier separating the two wells can be considered as impenetrable,
the level En is doubly degenerate, and the corresponding eigen space
is spanned by the vectors {|n, L , |n, R}. The projection of the Hamiltonian of the system H = p2 /2m + V on this eigen space is represented
by the 2 2 matrix Hn , which is given by

n, L| H |n, L n, L| H |n, R
Hn =
.
(11.63)
n, R| H |n, L n, R| H |n, R
Employing the approximations
H |n, L En |n, L ,
H |n, R En |n, R ,
one finds that

1
Hn = En
,
1
(11.64)
(11.65)
(11.66)
where
= n, L |n, R ,
(11.67)
or in the coordinate representation
Eyal Buks
364
11.6. Solutions
=
dx n,L (x) n,R (x) ,
(11.68)
where n,L (x) and n,R (x) are the wavefunctions of the states |n, L
and |n, R respectively, i.e.
n,L (x) = x |n, L ,
n,R (x) = x |n, R .
(11.69)
(11.70)
The main contribution to the overlap integral (11.68) comes from the
classically forbidden region |x| a, where x = a are turning points
(i.e., En = V (a) = V (a)). With the help of Eq. (11.36) one finds that
!
"

/x
/a
2
1
1
|C|
exp
dx
|p|
exp
dx
|p|
a

a
x
dx
(11.71)
|p|
a
a
a

1
dx
2
= |C| exp
dx |p|
,
(11.72)

|p|
a
(11.73)
where C is the normalization factor

of the WKB wavefunction, which is
approximately given by C = 2 m/T (T is the time period of classical
oscillations of a particle having energy En in a well) in the limit of large
n [see Eq. (11.44)], thus
4
/a
dx
|p/m|
1
exp
a
dx |p| .
(11.74)
Finally, By diagonalizing the matrix Hn one finds that the two eigenenergies are En (1 ).
4. The radial equation for the case of hydrogen is given by [see Eq. (7.61)]

2 2
d
+
V
(r)
(11.75)
eff
2 dr2
where me is the reduced mass (me is the electrons mass), and where
Veff (r) =
e2 l (l + 1) 2
+
.
r
2r2
(11.76)
The eigenenergies Ekl are calculated using the Bohr-Sommerfeld quantization rule (11.40)
Eyal Buks
365

1
r2
dr
r1

1
2 (Ekl Veff (r)) = k + .
2
(11.77)
where k is required to be an integer. The points r1,2 are classical turning

points that satisfy
Ekl = Veff (r1,2 ) .
Using the notation
r1,2
,
1,2 =
a0
Ekl
=
,
EI
where
a0 =
2
e2
(11.78)
(11.79)
(11.80)
(11.81)
is the Bohrs radius and where

EI =
e4
22
(11.82)
is the ionization energy, Eq. (11.78) becomes

=
2 l (l + 1)
(11.83)
thus
1,2 =

1
1 1 l (l + 1) .
Similarly Eq. (11.77) becomes

2 :
2 l (l + 1)
1
d
k
+
2
2
(11.84)
(11.85)
or

1
I = k +
,
2
(11.86)
where the integral I, which is given by

I=
2
Eyal Buks

( 1 ) (2 )
d
,
(11.87)
366
11.6. Solutions
can be calculated using the residue theorem
!
"
:
1 + 2
41 2
I =
1
.
2
(1 + 2 )2
(11.88)
Thus the quantization condition (11.86) becomes

=
1
Ekl
=
2 .
EI
l (l + 1) + k + 12
(11.89)
Comparing with the exact result (7.84) shows that the WKB result is a
good approximation provided that the quantum numbers are large.
Eyal Buks
367
12. Path Integration
In this chapter, which is mainly based on Ref. [4], the technique of Feynmans
path integration is briefly reviewed.
12.1 Charged Particle in Electromagnetic Field

Consider a point particle having mass m and charge q moving under the
influence of electric field E and magnetic field B, which are related to the
scalar potential and to the vector potential A by
E =
1 A
,
c t
(12.1)
and
B=A.
(12.2)
The classical Lagrangian of the system is given by Eq. (1.43)

1
q
L = mr2 q + A r ,
2
c
(12.3)
and the classical Hamiltonian is given by Eq. (1.62)

2
p qc A
H=
+ q .
2m
(12.4)
The solution of the Euler Lagrange equations (1.8) yields the classical equation of motion of the system, which is given by Eq. (1.60)

1
m
r = q E + r B .
(12.5)
c
In what follows, we consider for simplicity the case where both and A are
time independent. For this case H becomes time independent, and thus the
quantum dynamics is governed by the time evolution operator, which is given
by Eq. (4.9)
Chapter 12. Path Integration

iHt
u (t) = exp
.
(12.6)
The propagator K (rb , t; ra ) is defined by

K (rb , t; ra ) = rb | u (t) |ra ,
(12.7)
where |r denotes a common eigenvector of the position operators x, y, and z

with vector of eigenvalues r = (x , y , z ). As can be seen from the definition,
the absolute value squared of the propagator K (rb , t; ra ) is the probability
distribution function to find the particle at point rb at time t given that it
was initially localized at point ra at time t = 0.
Dividing the time interval (0, t) into N sections of equal duration allows
expressing the time evolution operator as
N
t
u (t) = u
.
N
(12.8)
The identity operator in the position representation [see Eq. (3.65)] is given
by

1r = d3 r |r r | .
(12.9)
Inserting 1r between any two factors in Eq. (12.8) and using the notation
ra = r0 ,
rb = rN ,
t
=
,
N
one finds that
(12.10)
(12.11)
(12.12)
K (rb , t; ra ) = rN | u () u () u () u () |r0

= d3 r1 d3 r2 d3 rN1
*)
*)
rN | u ()
rN1 rN1
u ()
rN2 rN2
u () |r1 r1 | u () |r0 ,
(12.13)
thus
K
(rb , t; ra )
N1
#
n=1
d3 rn
N1
#
m=0

K rm+1 , ; rm .
(12.14)
In what follows the limit N will be taken,

and therefor
it is sufficient

to calculate the infinitesimal propagator K rm+1 , ; rm to first order only
in .
Eyal Buks
370
12.1. Charged Particle in Electromagnetic Field

With the help of the relation (12.120), which is given by

e(A+B) = eA eB + O 2 ,
(12.15)
one has

iH
u () = exp

!

2 "

i p qc A
iq
= exp
exp
+ O 2 .
2m
Equation (12.122), which is given by

0

2

ir
iV2
1
=
d
r
exp
i
V
r
,
exp
2m
2
m
(2i)3/2
(12.16)
(12.17)
allows expressing the first term in Eq. (12.16) as

!

2 "
0
2

i p qc A
ir
1

q
3
exp
d
r
exp
=
i
p
A
r
.
2m
2
m
c
(2i)3/2
(12.18)
Moreover, with the help of Eq. (12.121), which is given by

e(A+B) = eB/2 eA eB/2 + O 3 ,
(12.19)
one finds that

0

q
exp i
p A r
m
c

0

0

0

q Ar
q A r
= exp i
exp i
p r exp i
+ O 3/2 .
m c 2
m
m c 2
(12.20)
Combining these results yields

)
K rm+1 , ; rm = rm+1
u () |rm
!0
"
,

2

ir
q A (rm ) + A rm+1 r
1
3
=
d r exp
exp i
2
m c
2
(2i)3/2
0

iq (rm ) )
exp
rm+1
exp i
p r |rm

m

+O 3/2 .
(12.21)
Eyal Buks
371

In the next step the identity operator in the momentum representation [see
Eq. (3.71)], which is given by

1p = d3 p |p p | ,
(12.22)
is inserted to the left of the ket vector |rm . With the help of Eq. (3.75),
which is given by

1
ip r
r |p =
exp
,
(12.23)
3/2

(2)
one finds that
0

p r |rm
rm+1
exp i
m
!

"
0

ip rm+1 rm

1
3
d p exp
=
exp i
p r .

m
(2)3
(12.24)
Thus, by using Eq. (3.84), which is given by

ip r
1
3
d
p
exp
= (r ) ,
3

(2)
(12.25)
one finds that

!
"
0
0

rm+1
exp i
p r |rm = rm+1 rm
r , (12.26)
m
m
and thus

m 3/2
K rm+1 , ; rm =
exp
2i

i
Lm + O 3/2 ,
(12.27)
where
m
Lm =
r
m+1 rm
2

A (rm ) + A rm+1 rm+1 rm
. (12.28)
c
2
q
q (rm )+
Comparing Eq. (12.28) with the classical Lagrangian of the system, which
is given by Eq. (1.43)
1
q
L = mr2 q + A r ,
2
c
(12.29)
shows that Lm is nothing but the Lagrangian at point rm

Eyal Buks
372
12.2. Classical Limit

Lm = L (rm ) .
(12.30)
As we have discussed above, the terms of order 3/2 in Eq. (12.27) are not
expected to contribute to K (rb , t; ra ) in the limit of N . By ignoring
these terms Eq. (12.14) becomes
!
"
N/2 N1

N1
#
i t
Nm
d rn exp
K (rb , t; ra ) = lim
L (rm ) .
N 2it
N m=0
n=1
(12.31)
Recall that the action in classical physics [see Eq. (1.4)] associated with a
given path is given by

S = dt L .
(12.32)
Thus, by defining the integral operator

N/2 N1
rb
#
Nm
D [r (t)] = lim
d3 rn ,
N 2it
ra
n=1
the propagator K (rb , t; ra ) can be written as

rb
i
K (rb , t; ra ) =
Sr (t ) ,
D [r (t )] exp

ra
(12.33)
(12.34)
where
Sr (t) =
t
0
dt L [r (t)] .
(12.35)
Equation (12.34), which is known as Feynmans path integral, expresses the

propagator K (rb , t; ra ) in terms of all possible paths r (t ) satisfying r (0) =

ra and
given by
r (t) = rb , where each path r (t ) contributes a phase factor
exp iSr (t ) / , where Sr (t ) is the classical action of the path r (t ).
A note regarding notation: In the above derivation of Eq. (12.34) eigenvalues and eigenvectors were denoted with prime (e.g., r , |r , r |, p ) to make
them distinguishable from the corresponding operators (e.g., r and p). This
distinction is no longer needed for the rest of this chapter, since no quantum operators are used to evaluate path integrals, and therefore, to make the
notation less cumbersome, we omit the prime notation.
12.2 Classical Limit

Recall that the Hamiltons principle of least action states that the path taken
by a classical system is the one for which the action S obtains a local minimum. This implies that for any infinitesimal change in the path the resultant
Eyal Buks
373

change in the action S vanishes (i.e., S = 0). As we have seen in chapter 1,
this principle leads to Lagranges equations of motion (1.8), which are given
by
L
d L
=
.
dt qn
qn
(12.36)
While in classical mechanics a definite path is associated with the systems

dynamics, in quantum mechanics all possible paths are included in Feynmans
path integral. However, as we show below, in the classical limit the dominant
contribution to the path integral comes only from paths near the classical
one. The classical limit is defined to be the limit where the Planks constant
approaches zero 0. In this limit the exponent exp (iS/) in the path
integral rapidly oscillates, and consequently contributions from neighboring
paths tend to cancel each other. However, near the classical path, such averaging out does not occur since according to the principle of least action
S = 0 for the classical path. Consequently, constructive interference between
neighboring paths is possible near the classical path, and as a result the main
contribution to the path integral in the classical limit comes from the paths
near the classical path.
1
0.8
0.6
0.4
0.2
-4
-2
0
-0.2
2 x
-0.4
-0.6
-0.8
-1
Graphical demonstration of the stationary phase approximation The plot

shows the function cos x2 for the case = 1. According to the
stationary phase approximation,
/
in the limit , the main contribution
to the integral dx cos x2 comes from the region near the point

x = 0, where d x2 /dx = 0.
12.3 Aharonov-Bohm Effect

Using Eq. (1.43) for the classical Lagrangian of a charged particle in stationary electromagnetic field one finds that the classical action (12.35) associated
with a path r (t) in the time interval (0, t) is given by
Eyal Buks
374
12.3. Aharonov-Bohm Effect
S=
t
dt

1 2
q
mr q + A r .
2
c
(12.37)
Consider first the case where the vector potential vanishes, i.e. A = 0. For
this case, the system is said to be conservative, and therefor, as we have seen
in chapter 1 [see Eq. (1.29)], the energy of the system
1
E = mr2 + q
(12.38)
2
is a constant of the motion (see exercise 5 below). In terms of E the action
S (12.37), which is labeled as S0 for this case where A = 0, can be expressed
as

t
1 2
mr q
S0 = dt
2
0
t
0

dt E + mr2
= Et + m
rb
ra
dr r .
(12.39)
where ra = r (0) and rb = r (t). Employing Eq. (12.38) again allows rewriting
S0 as
rb
(12.40)
S0 = Et + dr p (r) ,
ra
where p (r) is the local classical momentum

p (r) = 2m (E q (r)) .
(12.41)
The phase factor in the path integral corresponding to S0 is given by

r

b
iS0
iEt
i
exp
= exp
exp
dr p (r) .
(12.42)

ra
Note the similarity between the second factor in the above equation and
between the WKB wavefunction [see Eq. (11.17)]. In the general case, where
A can be nonzero, the phase factor in the path integral becomes [see Eq.
(12.37)]

rb
iS
iS0
iq
exp
= exp
exp
dr A .
(12.43)

c
ra
Eyal Buks
375
collector
rb , t
impenetrable
long coil
B0
ra ,0
source
Fig. 12.1. Two-slit interference experiment with a very long impenetrable cylinder
placed near the gap between the slits.
12.3.1 Two-slit Interference

Consider a two-slit interference experiment where electrons having charge
q = e are injected from a point source at ra (see Fig. 12.1). A collector at
point rb measures the probability density to detect an electron at that point.
A very long impenetrable cylinder is placed near the gap between the slits in
order to produce a magnetic field inside the cylinder in the direction normal
to the plane of the figure. The field outside the cylinder, however, can be
made arbitrarily small, and in what follows we assume that it vanishes.
The probability density Pb to detect the electron at time t by the collector
located at point rb is given by
Pb = |K (rb , t; ra )|2 ,
(12.44)
where the propagator (12.34) is given for this case by

rb

iS0,r(t)
ie
D [r (t)] exp
exp
dr A , (12.45)
K (rb , t; ra ) =

c
ra
r(t)
where the trajectories r (t) satisfy r (0) = ra and r (t) = rb .

How Pb is modified when the magnetic field is turned on, and consequently
the last factor in Eq. (12.45) starts to play a role? To answer this question
it is convenient to divide the sum over all paths into two groups, one for all
Eyal Buks
376
12.3. Aharonov-Bohm Effect

paths going through the left slit, and another for all paths going through the
right one. Here we disregard paths crossing a slit more than one time, as their
contribution is expected to be small. In general, the difference 12 between
the vector potential phase factor in Eq. (12.45) associated with two different
paths r1 (t) and r2 (t) is given by

e
12 =
dr A
dr A
c
e
=
c
r1 (t)
r2 (t)
dr A ,
(12.46)
where the closed path integral is evaluated along the path r1 (t) in the forward
direction from ra to rb , and then along the path r2 (t) in the backward
direction from rb back to ra . This integral can be calculated using Stokes
theorem [see Eq. (12.2)]
@

e
e
12 =
,
(12.47)
dr A =
ds B = 2
c
c
0
where is the magnetic flux threaded through the area enclosed by the closed
path, and where
0 =
hc
e
(12.48)
is the so called flux quantum. While 12 vanishes for pairs of paths going
through the same slit, it has the same value 12 = 2/0 (12 = 2/0 )
for all the pairs where r1 (t) goes through the left (right) path and where r2 (t)
goes through the right (left) one. Thus, we come to the somewhat surprising
conclusion that the probability density Pb is expected to be dependent on
the magnetic field. The expected dependence is periodic in the magnetic
flux with flux quantum 0 period. Such dependence cannot be classically
understood, since in this example the electrons can never enter the region in
which the magnetic field B is finite, and thus the Lorentz force vanishes in
the entire region accessible for the electrons outside the impenetrable coil.
12.3.2 Gauge Invariance
Consider the following gauge transformation
= A + ,
AA
=,
(12.49)
(12.50)
where = (r) is an arbitrary smooth and continuous function of r, which

is assumed to be time independent. As can be seen from Eqs. (12.1) and
Eyal Buks
377

(12.2), this transformation leaves E and B unchanged, since () = 0.
In chapter 1 we have seen that such a gauge transformation [see Eqs. (1.44)
and 1.45)] modifies the Lagrangian [see Eq. (1.43)]
q
L L = L + r ,
(12.51)
c
and also the action [see Eq. (12.37)]
S S = S +
= S+
q
c
rb
ra
= S+
t
0
q
dt r
c
dr
q
[ (rb ) (ra )] ,
c
(12.52)
however, the classical motion is unaffected.

In quantum mechanics, the propagator is expressed as a path integral
[see Eq. (12.34)],
where each path r (t) contributes a phase factor given by

exp iSr(t) / . As can be seen from Eq. (12.52), this phase factor is generally
not singly determined, since it depends on the chosen gauge. This result,
however, should not be considered as paradoxical, since only phase difference
between different paths has any physical meaning. Indeed, as we have seen
above [see Eq. (12.47)], phase difference 12 , which determines the relative
phase between two different paths, is evaluated along a closed path, which is
singly determined, and therefore gauge invariant.
Exercise 12.3.1. Given that the wavefunction (r , t ) solves the Schrdinger
equation with vector A and scalar potentials, show that the wavefunction
(r , t ), which is given by

iq (r )

(r , t ) = exp
(r , t ) ,
(12.53)
c
= A + and scalar
solves the Schrdinger equation with vector A
=
potentials.
Solution 12.3.1. Using Eq. (3.76) one finds that

iq
iq
p exp
exp
c
c

iq
iq
= exp
p, exp
+p
c
c
q
= p+
.
c
(12.54)
Eyal Buks
378
12.4. One Dimensional Path Integrals

This result implies that

iq
q
q
iq
q
p A exp
= p A ,
exp
c
c
c
c
c
and therefore the following holds

iq
iq
exp
H exp
=H,
c
c
(12.55)
(12.56)

2
p qc A
H=
+ q
2m
(12.57)
is the Hamiltonian corresponding to the vector potential A, whereas

2

p qc A qc
+ q ,
(12.58)
H=
2m
Thus, one finds
is the Hamiltonian corresponding to the vector potential A.
that the state vector

(
iq
|
(12.59)
= exp
c
provided that the state vector |

solves the Schrdinger equation with H,
solves the Schrdinger equation with H, and therefore

(r , t ) = exp iq (r ) (r , t ) .
(12.60)
c
12.4 One Dimensional Path Integrals

Consider a point particle having mass m moving in one dimension along the
x axis under the influence of the potential V (x). The path integral (12.34)
for this case becomes
!

N/2 N1
"
N1
#
Nm
i t
xm+1 xm
K (xb , t; xa ) = lim
dxn exp
L xm ,
.
t
N 2it
N m=0
N
n=1
(12.61)
where the Lagrangian is

L (x, x)
=
Eyal Buks
1
mx 2 V (x) .
2
(12.62)
379

The solution of the Euler Lagrange equation, which is given by Eq. (1.8)
d L
L
=
,
dt x
x
(12.63)
yields the classical equation of motion of the system

m
x=
dV
.
dx
(12.64)
12.4.1 One Dimensional Free Particle

For this case V (x) = 0.
0

m
im
K (xb , t; xa ) =
exp
(xb xa )2 .
2it
2t
(12.65)
Solution 12.4.1. The path integral (12.61) for this case becomes

N/2 N1
N1
#
imN
imN
2
dxn exp
(xm+1 xm ) ,
N
2t
2t m=0
n=1
(12.66)
or
N/2 N1
#
n=1
dxn exp
N1

m=0
(xm+1 xm )
(12.67)
where
=
imN
.
2t
The first integral

(12.68)
/
3
4 0
3
4
2
2
dx1 exp (x2 x1 ) (x1 x0 ) =
exp (x2 x0 )2 ,
2
2
(12.69)
The second integral

dx1 can be calculated using the identity
dx2 can be calculated using the identity
3
4 0 2
3
4
2
2
exp (x3 x0 )2 .
dx2 exp (x3 x2 ) (x2 x0 ) =
2
3
3
Eyal Buks
380

(12.70)
/
Similarly, the nth integral dxn yields

0
n
2
exp
(xn+1 x0 ) .
(n + 1)
n+1
(12.71)
Therefor, the propagator is given by

N/2
4
3
2
(N 1)
2
exp (xb xa ) ,
2 3
N
N
(12.72)
or
K (xb , t; xa ) =

m
im
2
exp
(xb xa ) .
2it
2t
(12.73)
As can be seen from the classical equation of motion (12.64), a free particle
moves at a constant velocity. Thus, the classical path satisfying x (0) = xa
and x (t) = xb is given by
xc (t ) = xa +
(xb xa ) t
.
t
The corresponding classical action Sc is

m (xb xa )2
.
Sc =
dt L (x, x)
=
2t
xc
(12.74)
(12.75)
(t )
Note that the following holds

d2 Sc
m
= .
dxa dxb
t
(12.76)
Thus the propagator can be expressed in terms of the classical action Sc as

:

i
i d2 Sc
K (xb , t; xa ) =
exp
Sc .
(12.77)
2 dxa dxb

As we will see below, a similar expression for the propagator is obtained also
for other cases.
12.4.2 Expansion Around the Classical Path
Motivated by the previous example of a free particle, we attempt below to
relate the propagator for the more general case, where V (x) is allowed to be
Eyal Buks
381

x dependent, with the classical path xc (t ) and the corresponding classical
action Sc . Consider a general path x (t ) satisfying the boundary conditions
x (0) = xa and x (t) = xb . It is convenient to express the path as
x (t ) = xc (t ) + (t ) ,
(12.78)
where the deviation (t ) from the classical path xc (t ) vanishes at the end
points (0) = (t) = 0. The action associated with the path x (t ) can be
expressed as

S=
dt L (x, x)
,
(12.79)
x(t )
where the Lagrangian is given by Eq. (12.62).

Expanding S is orders of yields
S = Sc + S1 + S2 + ,
where
(12.80)
dt L (x, x)
,
(12.81)
!
"

L
S1 = dt
+
,
(12.82)
x
x=xc
x
x=xc
!
"

2
2
2
L
L
L
S2 = dt
2 + 2
+
. (12.83)
x2
x=xc
x x
x=xc
x 2
x=xc
Sc =
In the general case, higher orders in such an expansion may play an important
role, however, as will be discussed below, in the classical limit the dominant
contribution to the path integral comes from the lowest order terms.
Claim. S1 = 0.
Proof. Integrating by parts the term proportional to in the expression for

S1 yields
t
!
!
"
L
d
L
S1 =

+ dt
.
(12.84)
x
x=xc
x
x=xc
dt x
x=xc
0
The first term in Eq. (12.84) vanishes due to the boundary conditions (0) =
(t) = 0, whereas the second one vanishes because xc (t ) satisfies the Euler
Lagrange equation (12.63), thus S1 = 0. The fact that S1 vanishes is a direct
consequence of the principle of least action of classical mechanics that was
discussed in chapter 1.
Eyal Buks
382

Employing the coordinate transformation (12.78) and the expansion of S
around the classical path allows rewriting the path integral (12.61) as
K (xb , t; xa ) = Pc (xb , t; xa ) K (t) ,
(12.85)
where

iSc
,

3
i
S2 + O
K (t) = D [ (t )] exp
,

Pc (xb , t; xa ) = exp
(12.86)
(12.87)
and where
D [ (t )] = lim
Nm
2it
N/2 N1
#
d n .
(12.88)
n=1
The term K (t) is evaluated by integrating over all paths (t ) satisfying the
boundary conditions (0) = (t) = 0.
/
dx Pc (xb , t2 ; x ) Pc (x , t1 ; xa )
K (t1 + t2 )
=
.
Pc (xb , t1 + t2 ; xa )
K (t1 ) K (t2 )
(12.89)
Solution 12.4.2. As can be seen from the definition of the propagator

(12.7), the following holds

dx K (xb , t2 ; x ) K (x , t1 ; xa ) = dx xb | u (t2 ) |x x | u (t1 ) |xa
= xb | u (t1 + t2 ) |xa
= K (xb , t1 + t2 ; xa ) .
(12.90)
Requiring that this property is satisfied by the propagator K (xb , t; xa ) that
is given by Eq. (12.85) leads to
/
dx Pc (xb , t2 ; x ) Pc (x , t1 ; xa )
K (t1 + t2 )
=
.
(12.91)
Pc (xb , t1 + t2 ; xa )
K (t1 ) K (t2 )
12.4.3 One Dimensional Harmonic Oscillator
For this case the Lagrangian is taken to be given by
L (x, x)
=
Eyal Buks
mx 2 m 1 xx
m 2 x2
+
,
2
2
2
(12.92)
383

where m, and 1 are assumed to be real constants. As we will see below,
the term (m1 /2) xx doesnt affect the dynamics, however, it is taken into
account in order to allows studying the more general case where the Lagrangian contains all possible types of quadratic (in x and x)
terms (though,
for simplicity, all coefficients in the Lagrangian are assumed to be time independent). Consider a general path x (t ) satisfying the boundary conditions
x (0) = xa and x (t) = xb . Using the notation
x (t ) = xc (t ) + (t ) ,
the Lagrangian becomes

2

m
2
2
x c + + 1 (xc + ) x c + (xc + ) , (12.93)

L (x, x)
=
2
thus the action associated with the path x (t ) can be expressed as

S=
dt L (x, x)
= Sc + S1 + S2 ,
(12.94)
x(t )
where
m
Sc =
2
t
S1 = m
t
4
3

1
dt x c +
xc + x c 2 xc ,
2
t
dt
m
S2 =
2

dt x 2c + 1 xc x c 2 x2c ,
2

+ 1 2 2 .
(12.95)
(12.96)
(12.97)
As we have seen above, the principle of least action implies that S1 = 0. Note
that in this case the expansion to second order in is exact and all higher
order terms vanish. Thus, the exact solution of this problem will also provide
an approximate solution for systems whose Lagrangian can be approximated
by a quadratic one.
Exercise 12.4.3. Find the classical action Sc of a classical path satisfying
x (0) = xa and x (t) = xb .
Solution 12.4.3. The Euler Lagrange equation (12.63)
d L
L
=
,
dt x
x
(12.98)
for this case yields

Eyal Buks
384

x
= 2 x ,
(12.99)
thus, indeed the term (m 1 /2) xx doesnt affect the dynamics. Requiring also
boundary conditions x (0) = xa and x (t) = xb leads to
xc (t ) =
xb sin (t ) xa sin ( (t t))

.
sin (t)
(12.100)
To evaluate the corresponding action we calculate the following integrals

t
0

dt x 2c 2 x2c =
2
sin2 (t)
t
0
3
2
dt (xb cos (t ) xa cos ( (t t)))
4
2
(xb sin (t ) xa sin (t t))

2

2xa xb
2
= xa + xb cot (t)
sin (t)

t
2
= (xa xb ) cot (t) 2xa xb tan
,
2
(12.101)
and
t
dt xc x c
=
=
/t
0
dt (xb sin (t ) xa sin ( (t t))) (xb cos (t ) xa cos ( (t t)))

sin2 (t)
x2b x2a
,
2
(12.102)
thus, the action is given by

Sc =
dt L (x, x)
xc (t )

m 1 x2b x2a
m
t
2
=
(xa xb ) cot (t) 2xa xb tan
+
.
2
2
4
(12.103)
Eyal Buks
385

To evaluate the propagator according to Eq. (12.85) the factor K (t) has
to be determined. This can be done by employing relation (12.89) for the
case where xa = xb = 0
/
dx Pc (0, t2 ; x ) Pc (x , t1 ; 0)
K (t1 + t2 )
=
.
(12.104)
Pc (0, t1 + t2 ; 0)
K (t1 ) K (t2 )
0
m
K (t) =
.
2i sin (t)
(12.105)
Solution 12.4.4. By using Eqs. (12.103) and (12.104) one finds that

3 m
4
K (t1 + t2 )
dx exp i
(cot (t2 ) + cot (t1 )) x2 =
,
(12.106)
2
K (t1 ) K (t2 )
thus, using the general integral identity
0

i
dx exp ix2 =
,
where is real, one finds that

:
K (t1 + t2 )
2i
=
.
m (cot (t2 ) + cot (t1 ))
K (t1 ) K (t2 )
(12.107)
(12.108)
Alternatively, using the identity

1
sin (t1 ) sin (t2 )
=
,
cot (t2 ) + cot (t1 )
sin ( (t1 + t2 ))
this can be rewritten as
:
2i sin (t1 ) sin (t2 )
K (t1 + t2 )
=
.
m sin ( (t1 + t2 ))
K (t1 ) K (t2 )
0
m
K (t) = ef (t)
,
2i sin (t)
(12.109)
(12.110)
(12.111)
where f (t) is an arbitrary function of time. Substituting this into Eq. (12.110)
yields
f (t1 ) + f (t2 ) = f (t1 + t2 ) ,
(12.112)
thus f (t) = At, where A is a constant. Combining all these results the propagator (12.85) for the present case becomes
Eyal Buks
386
12.5. Semiclassical Limit

0
m
K (xb , t; xa ) = eAt
2i sin (t)
!

"

m 1 x2b x2a
i m
t
2
exp
(xa xb ) cot (t) 2xa xb tan
+
.
2
2
4
(12.113)
In addition we require that in the limit , 1 0 the above result will
approach the result given by Eq. (12.65) for the propagator of a free particle.
This requirement yields A = 0. Note that
(xa xb )2 cot (t) 2xa xb tan
t
2

2xa xb
= x2a + x2b cot (t)
.
sin (t)
(12.114)
For the case 1 = 0 the propagator K (xb , t; xa ) becomes

0
m
K (xb , t; xa ) =
2i sin (t)

, 2

im
2
exp
xa + xb cos (t) 2xa xb
.
2 sin (t)
(12.115)
d2 Sc
m
=
,
dxa dxb
sin (t)
(12.116)
thus, similar to the case of a free particle [see Eq. (12.77)], also for the present
case of a harmonic oscillator, the propagator can be expressed in terms of
the classical action Sc as
:

i d2 Sc
i
K (xb , t; xa ) =
exp
Sc .
(12.117)
2 dxa dxb
12.5 Semiclassical Limit

In the semiclassical limit the Planks constant is considered to be small.
In this limit the dominant contribution to the path integral comes only from
paths near the classical one, which has the least action. This implies that
in
the expansion of S around the classical path (12.80) terms of order O 3
can be approximately neglected. Thus, as can be seen from Eq. (12.87), in
this limit [see also Eq. (12.85)] the propagator K (xb , t; xa ) is evaluated by
path integration over the quadratic terms S2 only of the action [see Eq.
Eyal Buks
387

(12.83)]. In the previous section we have exactly calculated the propagator
associated with the quadratic Lagrangian of a harmonic oscillator. The result
was expressed in Eq. (12.117) in terms of the classical action Sc . As can be
seen from Eq. (12.77), the same expression is applicable also for the case
of a free particle. It can be shown that the same form is also applicable
for expressing the propagator K (xb , t; xa ) in the semiclassical limit for the
general case
:

i
i d2 Sc
exp
Sc .
(12.118)
K (xb , t; xa ) =
2 dxa dxb

The proof of the above result, which requires generalization of the derivation
that led to Eq. (12.117) for the case of a general quadratic Lagrangian, will
not be given here. Another important result, which also is given here without
a proof, generalizes Eq. (12.118) for the case of motion in n spacial dimensions
:

i d2 Sc
i
K (rb , t; ra ) = det
exp
(12.119)
Sc .
2 dra drb
12.6 Problems
1. Show that

e(A+B) = eA eB + O 2 ,
(12.120)
where A and B are operators.

2. Show that

e(A+B) = eB/2 eA eB/2 + O 3 ,
(12.121)
where A and B are operators.

3. Show that
0
2

ir
iV2
1
exp
d
r
exp
, (12.122)
=
i
V
r
3/2
2m
2
m
(2i)
where V is a vector operator.
4. Show that the energy (12.38) is indeed a constant of the motion.
5. Consider a quantum system having time independent Hamiltonian H and
a discrete energy spectrum. Express its partition function Z in terms of
the systemss propagator K (xb , t; xa ).
6. Consider a one dimensional harmonic oscillator having mass m and resonance angular frequency in thermal equilibrium at temperature T .
Calculate the matrix elements x | |x of the density operator in the
basis of eigenvectors |x of the position operator x.
Eyal Buks
388
12.7. Solutions
7. Consider a free particle in one dimension having mass m. Calculate the
position wavefunction (x , t) at time t given that the position wavefunction (x , 0) at time t = 0 is given by
!
2 "
1
1
x
(x , 0) =
exp
.
(12.123)
1/2
1/4
2
x
0
x
0
where x0 is a constant.
8. A particle having mass m is in the ground state of the potential well
V0 (x) = (1/2) m2 x2 for times t < 0 . At time t = 0 the potential
suddenly changes and becomes V1 (x) = mgx.
a) Calculate the propagator K (xb , t; xa ) from point xa to point xb in
the semiclassical limit for the case where the potential is V1 (x) (i.e.
for the Hamiltonian after the change at t = 0).
b) Use
the result
previous section to calculate the variance
'
( ) of the
*
(x)2 (t) = x2 (t) x (t)2 of the position operator x at time t.
12.7 Solutions
1. Consider the operator
C () = eA e(A+B) eB .
(12.124)
Clearly, C (0) = 1. Moreover, with the help of Eq. (2.170) one finds that
dC
= eA Ae(A+B) eB + eA e(A+B) (A + B) eB eA e(A+B) eB B ,
d
(12.125)
thus
dC
= A + (A + B) B = 0 ,
d
=0
(12.126)
namely

C () = 1 + O 2 ,
and therefor

e(A+B) = eA eB + O 2 .
(12.127)
2. Consider the operator
C () = eB/2 e(A+B) eB/2 eA .
Eyal Buks
(12.128)
389

As in the previous exercise, it is straightforward (though, somewhat tedious) to show that
C (0) = 1 ,
dC
=0,
d
=0
d2 C
=0,
d2
(12.129)
(12.130)
(12.131)
=0
thus

C () = 1 + O 3 ,
(12.132)
and therefor

e(A+B) = eB/2 eA eB/2 + O 3 .
3. The proof is trivial using the identity

0

2
2 /4
ex +x dx =
e
.
(12.133)
(12.134)
4. By taking the time derivative of E one has

dE
= mr
r + qr = r (m
r qE) .
dt
(12.135)
However, according to the equation of motion (1.60) the term in the

brackets vanishes, and therefor dE/dt = 0.
5. Assume that the energy eigenstates of the Hamiltonian H are labeled by
|an and the corresponding eigenenergies by En , i.e.
H |an = En |an ,
(12.136)
where
an |an = nn .
With the help of the closure relation

1=
|an an | ,
(12.137)
(12.138)
one finds that the propagator K (xb , t; xa ) can be expressed as

iHt
K (xb , t; xa ) = xb | exp
|xa

iEn t
=
xb |an exp
an |xa .

n
Eyal Buks
(12.139)
390
12.7. Solutions
Taking xb = xa and integrating over xa yields

iEn t
dxa K (xa , t; xa ) =
exp
.
(12.140)
Thus, the partition function Z, which is given by Eq. (8.41)

Z=
eEn ,
(12.141)
where = 1/kB T , can be expressed as

Z=
dx K (x , i; x ) .
(12.142)
6. Using Eq. (8.42) one finds that

x | |x =
x | eH |x
,
Z
(12.143)

where the partition function Z = Tr eH can be expressed in terms
of the propagator K (x , t; x ) [see Eq. (12.142)]

Z=
dx K (x , i; x ) .
(12.144)
Furthermore, as can be seen from the definition of the propagator [see

Eq. (12.7)]
K (x , t; x ) = x | e
iHt
|x ,
(12.145)
the following holds

x | eH |x = K (x , i; x ) .
(12.146)
Thus, with the help of Eq. (12.115) one finds for the case of a harmonic
oscillator that (recall that sin (ix) = i sinh x and cos (ix) = cosh x)

Z=
dx K (x , i; x )

0
m
m [cosh () 1] x2
dx exp
=
2 sinh ()
sinh ()
:
1
=
2 [cosh () 1]
=
1
,
2 sinh
2
(12.147)
Eyal Buks
391

and therefore one finds, in agreement with Eq. (8.352), that
x | |x =
K (x , i; x )
Z:
2m
2
sinh ()

, 2

m
x + x2 cosh () 2x x
exp
2 sinh ()
= sinh
where
x0 =
2
2
+x
x
tanh( 2 ) x 2x
coth( 2 ) x 2x
0

x0 coth
2

.
m
(12.148)
(12.149)
7. Denoting the state ket vector of the system by | (t) and the time evolution operator by u (t) one has
(x , t) = x | (t)
= x | u (t) | (0)

=
dx x | u (t) |x x | (0)

=
dx K (x , t; x ) (x , 0) ,
(12.150)
where the propagator K (x , t; x ) is given by Eq. (12.73)

:

1
i (x x )2
K (x , t; x ) =
exp
,
2itx20
2t
x20
(12.151)
and where
=

,
mx20
(12.152)
thus
(x , t) =

1
dx
1/2
2itx20
1/4 x0

2
2
1
i
x
i x x
i
x
exp
1
+
.
2
t
x0
t x20
2t x0
1
(12.153)
Eyal Buks
392
12.7. Solutions
1

2
1 14 4ca+b
a
e
,
a
(12.154)
one finds that

(x , t) =
1
1/2
1/4 x0
!
2 "
1
1
x
exp
. (12.155)
1 + it
2 (1 + it) x0
8. The Lagrangian for times t > 0 is given by

1
L (x, x)
= mx 2 mgx .
2
(12.156)
The Euler Lagrange equation yields the classical equation of motion of

the system
x
= g .
(12.157)
The general solution reads

x = x0 + v0 t
gt2
,
2
(12.158)
where the constants x0 and v0 are the initial values of the position and
velocity at time t = 0. Given that x = xa at time t = 0 and x = xb at
time t one finds that x0 = xa and
v0 =
xb xa gt
+
,
t
2
(12.159)
thus the classical trajectory xc (t ) is given by

xb xa g
g
xc (t ) = xa +
+ t t t2 .
t
2
2
(12.160)
Using the notation

xt =
gt2
,
2
(12.161)
the trajectory xc (t ) is expressed as

xc (t ) = xa + (xb xa xt )
t
t2
+ xt 2 ,
t
t
(12.162)
and the corresponding velocity x c (t ) is expressed as

Eyal Buks
393

xb xa xt 2xt t
+ 2 .
t
t
x c (t ) =
(12.163)
The Lagrangian along the classical trajectory is given by

L (xc , x c ) =
=
1
mx 2c mgxc
2

m xb xta xt +
2xt t
t2
2

t2
t
mg xa + (xb xa xt ) + xt 2 ,
t
t
(12.164)
and the corresponding action Sc is given by

Sc =
dt L (x, x)
xc (t )
=m
t
0
=m
dt
xb xta xt +
2
2xt t
t2
2
(xb xa )2 + 2xt (xb + xa )

2t

t
t2
g xa + (xb xa xt ) + xt 2
t
t
x2t
3
(12.165)
a) In general, the propagator in the semiclassical limit is given by Eq.
(12.118)
:

i d2 Sc
i
exp
K (xb , t; xa ) =
Sc ,
(12.166)
2 dxa dxb

where for the present case Sc is given by Eq. (12.165) and
d2 Sc
m
= ,
dxa dxb
t
thus
(12.167)
!
"
x2
2
m
im (xb xa ) + 2xt (xb + xa ) 3t
K (xb , t; xa ) =
exp
2it

2t
!
"
0
x2
2
1
1
i (xb xa ) + 2xt (xb + xa ) 3t
=
exp
x0 2it
t
2x20
!
"
0
2
1
1
i 83 x2t + x2a + 2 (2xt xa ) (xb xt ) + (xb xt )
=
exp
,
x0 2it
t
2x20
0
(12.168)
where
Eyal Buks
394
12.7. Solutions
x0 =

.
m
(12.169)
b) Initially at time t = 0 the wavefunction (x ) is given by [see Eq.

(5.120)]
!
2 "
1
1 x
exp
,
(12.170)
(x , t = 0) =
1/2
2 x0
1/4 x
0
where
x0 =

.
m
(12.171)
The wave function

at time t is evaluated using the propagator
(x , t) =
dx K (x , t; x ) (x , 0)
0
1
1
1
=
x0 2it 1/4 x1/2
0
2
2
2
8 2

i 3 xt +(x ) +2(2xt x )(x xt )+(x xt )
12 xx
t
2x2
0
0
dx e
1
x0
1 e
2it
8 2
i 3 xt +4xt (x xt )+(x xt )
t
2x2
0
1/2
1/4 x0
i
(1 t
)(
2
dx e
x
x0
ix (x xt )
tx2
0
.
(12.172)

1

exp ax2 + bx dx =
1 4a
b2
e ,
a
(12.173)
one finds that

(x , t) =
12
1
1+it
x xt
x0
2
8 ix2 +4ix (x x )
t
t
t
3
2tx2
0
1/2
1/4 x0
1 + it
(12.174)
The probability distribution function f (x ) = | (x , t)|2 to find the

particle near point x at time t is thus given by
1
1+(t)
2
x xt
x0
2
e
.
f (x ) =
,
x0 1 + (t)2
Eyal Buks
(12.175)
395

therefore f (x ) has a Gaussian distribution with a mean value xt and
variance given by
'

( x2
2
2
.
(x) (t) = 0 1 + (t)
2
(12.176)
Note that this result is in agreement with Eq. (5.270).
Eyal Buks
396
13. Adiabatic Approximation
The adiabatic approximation can be employed for treating systems having

slowly varying Hamiltonian. This chapter is mainly based on Ref. [5].
13.1 Momentary Diagonalization

The Schrdinger equation (4.1) is given by
i
d |
= H | .
dt
(13.1)
For any given value of the time t the Hamiltonian H (t) is assumed to have
a discrete spectrum
H (t) |n (t) = En (t) |n (t) ,
(13.2)
where n = 1, 2, , the momentary eigenenergies En (t) are real, and the set
of momentary eigenvectors is assumed to be orthonormal
n (t) |m (t) = nm .
(13.3)
The general solution can be expanded using the momentary eigenvectors as

a momentary basis

| (t) =
an (t) ei n (t) |n (t) .
(13.4)
n
The phase factors n (t) in the expansion (13.4) are chosen to be given by
n (t) = n (t) + n (t) ,
where the phase factors

1 t
n (t) =
dt En (t )

are the so-called dynamical phases, and the other phase factors
(13.5)
(13.6)
Chapter 13. Adiabatic Approximation

n (t) = i
dt n (t ) |n (t )
(13.7)
are the so-called geometrical phases. As we will see below, choosing the phase
factor n (t) to be given by Eq. (13.5) ensures that the coefficients an (t)
become constants in the adiabatic limit.
Exercise 13.1.1. Show that the term n (t ) |n (t ) is pure imaginary.
Solution 13.1.1. Note that by taking the derivative with respect to t (denoted by upper-dot) of the normalization condition (13.3) one finds that
n |m + n |m
=0,
(13.8)
thus
n |m
= m |n
.
(13.9)
The last result for the case n = m implies that n (t ) |n (t ) is pure imaginary, and consequently n (t) are pure real.
Substituting Eq. (13.4) into Eq. (13.1) leads to

i
a n (t) ein (t) |n (t)
n

n
+i
=

n
an (t) n (t) ein (t) |n (t)

n
an (t) ei n (t) |n (t)
an (t) ein (t) En (t) |n (t) .

(13.10)
i m (t)
Taking the inner product with m (t)| e
yields

Em (t)
a m (t)+i m (t) am (t)+
an (t) ein (t) ei m (t) m (t) |n (t) =
am (t) .
i
n
(13.11)
Since, by definition, the following holds

i m (t) =
Em (t)
m (t) |m
(t) ,
i
Eq. (13.11) can be rewritten as

a m =
ei(n (t) m (t)) m (t) |n (t) an .
(13.12)
(13.13)
n
=m
Eyal Buks
398
13.3. Adiabatic Limit

Exercise 13.1.2. Show that for n = m the following holds
m (t) |n (t) =
m (t)| H |n (t)
.
En (t) Em (t)
(13.14)
Solution 13.1.2. Taking the time derivative of Eq. (13.2)

,
H |n + H |n
= E n |n + En |n
(13.15)
and the inner product with m (t)|, where m = n, yields the desired identity.
13.2 Gauge Transformation

The momentary orthonormal basis {|n (t )}n , which is made of eigenvectors of H (t), is clearly not singly determined. Consider the following gauge
transformation [see for comparison Eq. (12.49)]
|n (t ) |
n (t ) = ei(t ) |n (t ) ,
(13.16)
where (t ) is arbitrary real function of time. The geometrical phase n (t),

which is given by Eq. (13.7)
n (t) = i
t0
dt n (t ) |n (t ) ,
(13.17)
is transferred to
n (t) n (t) = n (t) + (t) (t0 ) .
(13.18)
Thus, in general the geometrical phase is not singly determined. However, it

becomes singly determined, and thus gauge invariant, if the path is closed,
namely if H (t) = H (t0 ), since for such a case (t) = (t0 ).
13.3 Adiabatic Limit

In the adiabatic limit the terms m (t) |n (t) are considered to be negligibly
small. As can be seen from Eq. (13.14), this limit corresponds to the case
where the rate of change in time of the Hamiltonian approaches zero. In
this limit the coefficients an (t) become constants [see Eq. (13.13)], and the
solution (13.4) thus becomes

| (t) =
an ein (t) |n (t) .
(13.19)
n
Eyal Buks
399
13.4 The Case of Two Dimensional Hilbert Space

In this case the Hilbert space is two dimensional and the Hamiltonian can be
represented by a 2 2 matrix, which is conveniently expressed as a combination of Pauli matrices
H=h
0I + h ,
(13.20)
where I is the 2 2 identity matrix, h0 is a real scalar, h = (h1 , h2 , h3 ) is a

three-dimensional real vector, and the components of the Pauli matrix vector
are given by

01
0 i
1 0
1 =
, 2 =
, 3 =
.
(13.21)
10
i 0
0 1
where H = h h, and where h

is a unit vector,
Using the notation h = H h,
given in spherical coordinates by
= (cos sin , sin sin , cos ) ,
h
(13.22)
one finds that

H=h
0I + H
cos
sin exp (i)
sin exp (i)
cos
(13.23)
The orthonormal eigenvectors are chosen to be given by [see Eqs. (6.213) and
(6.214)]

cos 2 exp i
2

|+ =
,
(13.24)
sin 2 exp i
2

sin 2 exp i
2

| =
,
(13.25)
cos 2 exp i
2
and the following holds +|+ = | = 1, +| = 0, and

H | = (h0 H) | .
(13.26)
Note that the eigenstates | are independent of both h0 and H.

The geometrical phase (13.7) can be evaluated by integration along the
path h (t)
n (t) = i
dt n (t ) |n (t ) = i
h(t)
h(0)
dh n (h)| h |n (h) . (13.27)

| h | =
Eyal Buks
ctg .
2H
(13.28)
400
13.4. The Case of Two Dimensional Hilbert Space

Solution 13.4.1. Using the expression for a gradient in spherical coordinates (the radial coordinate r in the present case is H)
f =
f
1 f
1 f
r+
+
,
r
r
r sin
(13.29)
one finds that

h |+ =

sin exp i
cos 2 exp i
2
2
2

+
,
cos 2 exp i
sin 2 exp i
2H
2H sin
2
2
(13.30)
thus

i

+| h |+ =
sin 2 exp i
cos 2 exp i
2
2
2H sin
i
=
ctg .
2H

cos 2 exp i
2

sin 2 exp i
2
(13.31)
The term | h | is calculated in a similar way.

For the case of a close path, Stocks theorem can be used to express the
integral in terms of a surface integral over the surface bounded by the close
curve h (t)
@

n = i dh n| h |n = i da ( n| h |n) .
(13.32)
S

ih
| h | =
.
2 |h|3
(13.33)
Solution 13.4.2. Using the general expression for the curl operator in spherical coordinates (again, note that the radial coordinate r in the present case
is H)

1
(sin A ) A
A =
r
r sin

1
1 Ar
(rA )
+
r sin
r

1 (rA ) Ar
+
,
r
r
(13.34)
one finds that

| h | =
Eyal Buks
ih
cos
ih
=
.
2H 2 sin
2 |h|3
(13.35)
401

With the help of the last result one thus finds that

1
h
1
=
da
= ,
3
2
2
|h|
(13.36)
where is the solid angle subtended by the close path h (t) as seen from the
origin. Due to the geometrical nature of the last result, these phase factors
were given the name geometrical phases.
13.5 Transition Probability

The set of equations of motion (13.13) can be rewritten in a matrix form as
d
|a) = H |a) ,
dt
where

a1
a2
|a) =
..
.
(13.37)
(13.38)
is a column vector of the coefficients an C, and where the matrix elements

of H are given by
Hmn = Hnm
= iei(n (t) m (t)) m (t) |n (t)
for the case n = m and Hnn = 0 otherwise.

The inner product between the vectors

b1
a1

|a) = a2 , |b) = b2 ,
..
..
.
.
(13.39)
(13.40)
is defined by
(a |b) = (b |a) =
am bm .
(13.41)
The set of vectors {|n)} (n = 1, 2, ), having coefficients am = nm , forms

an orthonormal basis for the vector space
(n1 |n2 ) = n1 n2 .
(13.42)
Consider the case where the system is initially at time t0 in the state |n).
What is the probability pnn (t) to find it later at time t > t0 at the same state
|n)? The adiabatic approximation is valid only when pnn 1. Considering
the matrix H as a perturbation, the probability pnn can be approximated
using time dependent perturbation theory.
Eyal Buks
402

Exercise 13.5.1. Show that to lowest nonvanishing order the following
holds
/t
pnn (t) = 1
(13.43)
dt Hnm (t )
.
t0
Solution 13.5.1. By employing Eqs. (10.21) and (10.27) one finds that (recall that Hnn = 0)
pnn (t) = 1
/t
dt
t0
/t
t0
dt (n| H (t ) H (t ) |n) .
(13.44)
<
Inserting the identity operator 1 = m |m) (m| between H (t ) and H (t )
and recalling that Hmn = Hnm

lead to

pnn (t) = 1
pmn (t) ,
(13.45)
m
where
/t
pmn (t) =
dt Hnm (t )
.
t0
(13.46)
As can be seen from Eq. (13.39), the matrix elements Hnm (t ) are proportional to the oscillatory dynamical phase factors
Hmn

i t
exp (i ( n (t) m (t))) = exp

dt (En (t ) Em (t )) .

(13.47)
In the adiabatic limit these terms rapidly oscillate and consequently the probabilities pmn (t) are exponentially small. From the same reason, the dominant
contribution to the integral is expected to come from regions where the energy gap En (t )Em (t ) is relatively small. Moreover, it is also expected that
the main contribution to the total survival probability pnn will come from
those states whose energy Em (t ) is close to En (t ). Having this is mind, we
study below the transition probability for the case of a two level system. As
we will see below, the main contribution indeed comes from the region near
the point where the energy gap obtains a minimum.
13.5.1 The Case of Two Dimensional Hilbert Space
We calculate below p+ for the case H=h
, where h (t) is the straight line
h (t) = (0, 1, t) ,
Eyal Buks
(13.48)
403

where is a positive constant, is a real constant, and where the initial time
is taken to be t0 = and the final one is taken to be t = . In spherical
coordinates h (t) is given by
h (t) = H (0, sin , cos ) ,
(13.49)
where
H =
.
1 + (t)2 ,
cot = t ,
(13.50)
(13.51)
and where = /2. Thus, the energy gap 2H obtains a minimum at time
t = 0. As can be seen from Eqs. (13.24) and (13.25), for any curve lying on
a plane with a constant azimuthal angle , the following holds
+
= | ,
2
(13.52)
and therefor
and
*
= ,

+
2
(13.53)
*
=
=0.
+
+
(13.54)
For the present case one finds using Eq. (13.51) that
*
1
=

+
.
2 1 + (t)2
This together with Eqs. (13.39) and (13.46) leads to
/
i(+ (t ) (t ))

p+ =
dt e
(t ) + (t )

2
.
/ t

exp 2i 0 dt
1 + (t )
dt
1 + (t )2

/ t

exp 2i
1 + 2
0 d
dt
1 + (t )2
,

-2

2
2
exp i
1 /
1 + ln + 1 +
.
=
2
1+
Eyal Buks
(13.55)
(13.56)
404

Exercise 13.5.2. Show that if / 1 then

p+ exp
.
Solution 13.5.2. The variable transformation

= sinh z ,
(13.57)
and the identities

1 + 2 = cosh z ,

1
1 + 2 = sinh (2z) ,
2

ln + 1 + 2 = z ,
d = cosh z dz ,
(13.58)
(13.59)
(13.60)
(13.61)
yield
p+
1
4

exp i
sinh
(2z)
+
z
1 /
.
=
dz
cosh z
(13.62)
In the limit / the phase oscillates rapidly and consequently p+ 0.

The stationary phase points zn in the complex plane are found from the
condition

d 1
0=
sinh 2z + z = cosh 2z + 1 ,
(13.63)
dz 2
thus

1
zn = i n +
,
2
(13.64)
where n is integer. Note, however that the term 1/ cosh z has poles at the same
points. Using the Cauchys theorem the path of integration can be deformed
to pass close to the point z1 = i/2. Since the pole at z1 is a simple one,
the principle value of the integral exists. To avoid passing through the pole at
z1 a trajectory forming a half circle "above" the pole with radius is chosen
were 0. This section gives the dominant contribution which is iR, where
R is the residue at the pole. Thus the probability p+ is approximately given
by

p+ exp
.
(13.65)
The last result can be used to obtain a validity condition for the adiabatic approximation. In the adiabatic limit p+ 1, and thus the condition
/ 1 is required to ensure the validity of the approximation.
Eyal Buks
405
13.6 Slow and Fast Coordinates

Consider a system whose Hamiltonian is given by
H = H0 + H1 .
(13.66)
The Hamiltonian H0 is assume to depend on a set of coordinates x =

(x1 , x2,... ) and on their canonically conjugate variables p = (p1 , p2,... ), i.e.
H0 = H0 (
x, p). In what follows the coordinates x = (x1 , x2,... ) will be considered as slow, and thus H0 will be considered as the Hamiltonian of the
slow subsystem. The other part of the system is a fast subsystem, which is
assumed to have a much faster dynamics and its energy spectrum is assumed
to be discrete. The Hamiltonian of the fast subsystem H1 is assumed to parametrically depend on the slow degrees of freedom x, i.e. H1 = H1 (
x). This
dependence gives rise to the coupling between the slow and fast subsystems.
An adiabatic approximation can be employed in order to simplify the
equations of motion of the combined system. In what follows, for simplicity,
this method will be demonstrated for the case where the slow subsystem is
assumed to be composed of a set of decoupled harmonic oscillators. For this
case the Hamiltonian H0 is taken to be given by

p2
ml 2l x2l
l
H0 =
+
,
(13.67)
2ml
2
l
where ml and l are the mass and angular frequency of mode l, respectively.
The Hamiltonian of the fast subsystem H1 (
x), which depends parametrically on x, has a set of eigenvectors and corresponding eigenvalues for any
given value of x
H1 |n (
x) = n (
x) |n (
x) ,
(13.68)
where n = 1, 2, . The set of local eigenvectors {|n (

x)} is assumed to
form an orthonormal basis of the Hilbert space of the fast subsystem, and
thus the following is assumed to hold
m (
x) |n (
x) = mn ,
and [see Eq. (2.23)]

|k (
x) k (
x)| = 1F ,
(13.69)
(13.70)
where 1F is the identity operator on the Hilbert space of the fast subsystem.
The state of the entire system (t) at time t is expanded at any point x
using the local basis {|n (

x)}

(t) =
n (
x, t) |n (
x) .
(13.71)
n
Eyal Buks
406
13.6. Slow and Fast Coordinates

In the above expression a mixed notation is being employed. On one hand,
the ket notation is used to denote the state of the fast subsystem (the terms
|n (
x)). On the other hand, a wavefunction in the position representation
x, t)) is employed to denote the state of the slow subsystem.
(the terms n (
Substituting the expansion (13.71) into the Schrdinger equation (4.1)
i
d
= H ,
dt
(13.72)
leads to

[H0 + n (
x)] n (
x, t) |n (
x) = i
n (
x, t) |n (
x) ,
n
(13.73)
where overdot represents time derivative. Projecting m (

x)| leads to

m (
x)| H0 n (
x, t) |n (
x) + m (
x) m (
x, t) = i m (
x, t) .
(13.74)
n
By calculating the term [see Eq. (13.69)]

m (
x)| p2l n (
x, t) |n (
x) = n (
x, t) m (
x)| p2l |n (
x)
+ 2 (pl n (
x, t)) m (
x)| pl |n (
x) + mn p2l n (
x, t) ,
(13.75)
introducing the notation

Am,n;l m (
x)| pl |n (
x) ,
(13.76)
and using [see Eq. (13.70)]

m (
x)| p2l |n (
x) =
m (
x)| pl |k (
x) k (
x)| pl |n (
x)
= pl Am,n;l +
Am,k;l Ak,n;l ,
(13.77)
one obtains
m (
x)| p2l n (
x, t) |n (
x)
!
= n (
x, t) pl Am,n;l +
2Am,n;l pl n (
x, t) +
Am,k;l Ak,n;l
k
mn p2l n (
x, t) .
"
(13.78)
With the help of Eqs. (13.67) and (13.74) one finds that
Eyal Buks
407

!
"
1

(
x, t) pl Am,n;l +
Am,k;l Ak,n;l
2ml n n
l
k
1 ,
2Am,n;l pl n (
x, t) + mn p2l n (
x, t)
2ml n
l
ml 2 x2
l
x, t) + m (
x) m (
x, t) = i m (
x, t) .
m (
(13.79)

Defining the matrices Al
= Am,n;l , ()m,n = m mn , and the vector
m,n

= , the above can be written in a matrix form as
n
n
?
D
2 m 2 x2
1
l
l
l
pl Al +
+ = i .
2ml
2
(13.80)
To calculate the off-diagonal matrix elements of Al we apply pl to Eq.

(13.68) and project m (
x)|, where m = n
m (
x)| pl H1 |n (
x) = m (
x)| pl n (
x) |n (
x) .
(13.81)
Using Eq. (13.69), the definition (13.76) and pl = i x

[see Eq. (3.29)] one
l
finds that [compare with Eq. (13.14)]
Am,n;l = i
1
x)
m (
x)| H
xl |n (
.
n m
(13.82)
In the adiabatic approximation the off diagonal elements of Al [see Eq.

(13.82)] are considered as negligible small. In this case the set of equations
of motion (13.80) becomes decoupled
?

D

2
2 2
(pl Am,m;l (
m
x
))
x
l
l
l
i m =
+
+ m (
x) m .
(13.83)
2ml
2
l
As can be seen from the above result (13.83), the adiabatic approximation
greatly simplifies the systems equations of motion. The effect of the fast
subsystem on the dynamics of the slow one is taken into account by adding a
vector potential Am,m;l (
x) and a scalar potential m (
x) to the Schrdinger
equation of the slow subsystem [compare with Eq. (4.194)]. However, both
potential terms depend on the state m that is being occupied by the fast
subsystem.
Exercise 13.6.1. Show that if m (
x)| /xl |m (
x) is pure real then
Amm;l (
x) = 0 .
Eyal Buks
(13.84)
408
13.7. Problems
Solution 13.6.1. Note that in general the diagonal elements Am,m;l are real
since pl is Hermitian [see Eq. (13.76)]. On the other hand, if m (
x)| /xl |m (
x)
is pure real then Amm;l (
x) is pure imaginary, thus for this case Amm;l (
x) = 0.
13.7 Problems
1. Consider the following gauge transformation

*
cos 2 exp i
cos 2
2
+ =
|+ =
,
sin 2 exp (i)
sin 2 exp i
2

*
sin 2 exp i
sin 2
2
| =
.
cos 2 exp (i)
cos 2 exp i
2
(13.85)
(13.86)
Find an expression for the transformed geometrical phase (t).

2. Consider a particle having mass m confined by a time dependent potential
well given by
+
0
if 0 x a
V (x) =
.
(13.87)
if x < 0 or x > a
where the width of the well a oscillates in time according to

a (t) = a0 1 sin2 ( p t) ,
(13.88)
where a0 , and p are positive, and where < 1.

a) Under what condition the adiabatic approximation is expected to be
valid.
b) Calculate the geometrical phases n [see Eq. (13.7)] for a cyclic evolution from time t = 0 to time t = / p .
3. Consider a particle of mass m moving in one dimension along the x axis,
whose time-dependent Hamiltonian H (t ) is given by
H (t ) =
p2
m 20 x2
+
+ xf (t ) ,
2m
2
(13.89)
where p is the variable canonically conjugate to x, the force f (t ) is given

by
2
exp t 2
,
(13.90)
f (t ) =
and 0 , and are real constants. When is sufficiently large the problem can be treated using the adiabatic approximation. Expand the state
of the system | (t) in a basis of momentary eigenvectors of the Hamiltonian H (t ) and derive the equations of motion for the coefficients in
Eyal Buks
409

that basis. The system is initially prepared, at time t , in the
ground state of the momentary Hamiltonian limt H (t). Using the
equations of motion for the coefficients in the momentary basis calculate to lowest nonvanishing order in the adiabatic approximation the
transition probability pn0 to any of the momentary eigenvectors of the
Hamiltonian limt H (t) in the limit t .
13.8 Solutions

*
i
+
= exp
|+ ,
2

*
i
= exp
| ,
2
(13.91)
(13.92)
thus the transformed geometrical phase (t) [see Eq. (13.18)] becomes
n (t) n (t) = n (t)
(t) (t0 )
+
.
2
2
(13.93)
2. The momentary eigenenergies of the system are given by

En (t) =
2 2 n2
2 ,

2ma20 1 sin2 ( p t)
(13.94)
where n = 1, 2, . The corresponding eigenvectors are denoted by

|n (t).
a) In general, the main contribution to interstate transitions come from
time periods when the energy gap between neighboring eigenenergies is relatively small. At any given time the smallest energy gap
between momentary eigenenergies is the one between the two lowest
states H21 (t) = E2 (t) E1 (t). Furthermore, the main contribution
to the transition probability is expected to come from the regions near
minima points of the energy gap H21 (t). Near the minima point at
time t = 0 the energy gap H21 (t) is given by
32 2
H21 (t) =

2
2ma20 1 sin2 ( p t)

32 2
=
1 + 2 (p t)2 + O t3
2
2ma0
.

32 2
=
1 + 4 ( p t)2 + O t3 .
2ma20
(13.95)
Eyal Buks
410
13.8. Solutions
The estimated transition probability for the two dimensional case
is given by Eq. (13.65). In view of the fact that all other energy
gaps between momentary eigenenergies is at least 5/3 larger than
H21 , it is expected that this estimate is roughly valid for the present
case. The requirement that the transition probability given by Eq.
(13.65) is small is taken to be the validity condition for the adiabatic
approximation. Comparing the above expression for H21 (t) near t =
0 with Eq. (13.50) yields the following validity condition
2p 1/2
33
.
2ma20
(13.96)
b) In general, the term n (t ) |n (t ) is pure imaginary [see Eq. (13.9)].

On the other hand, the fact that the wavefunctions of one dimensional
bound states can be chosen to be real (see exercise 7 of chapter 4),
implies that n (t ) |n (t ) is pure real. Thus n (t ) |n (t ) = 0 and
therefore all geometrical phases vanish.
3. The Hamiltonian H0 limt H (t) is given by
H0 =
p2
m 20 x2
+
.
2m
2
(13.97)
The eigenvectors |n and eigenenergies En,0 = 0 (n + 1/2) of H0 satisfy

the following relation
H0 |n = En,0 |n ,
(13.98)
where n = 0, 1, , 2 .The momentary Hamiltonian H (t ) can be rewritten as

2
p2
m 20
f (t )
f 2 (t )
,
(13.99)
H (t ) =
+
x+
2
2m
2
m 0
2m20
thus H (t ) describes a harmonic oscillator of angular frequency 0 having a parabolic potential centered at f (t ) /m 20 , and consequently the
momentary eigenvectors |n (t) of the Hamiltonian H (t ) can be chosen
to be coherent states given by [see Eqs. (5.36) and (5.46)]
|n (t ) = D ( (t )) |0 ,
(13.100)

where D () = exp a a is the displacement operator, (t ) is
given by
(t ) =
f (t )
21/2 m20 x0
(13.101)
and
Eyal Buks
411

x0 =

.
m0
(13.102)
The following holds

H (t ) |n (t ) = En (t ) |n (t ) ,
(13.103)
where n = 1, 2, , the momentary eigenenergies En (t ) are given by

En (t ) = En,0
f 2 (t )
,
2m20
(13.104)
and n (t) |m (t) = nm . The adiabatic expansion is given by [see Eq.

(13.4)]

| (t ) =
an (t ) ein (t ) |n (t ) ,
(13.105)
n
where n (t ) = n (t ) + n (t ),
1
n (t ) =
dt En (t ) ,
(13.106)
and
n (t ) = i
dt n (t ) |n (t ) .
(13.107)
To lowest nonvanishing order in the adiabatic approximation the transition probability p0n is given by [see Eqs. (13.14) and (13.46)]
n
(t)|
H
|0
(t)
pn0 =
dt ei(0 (t )n (t ))
(13.108)
En (t) E0 (t)
where overdot denotes derivative with respect to time. In general the

term n (t ) |n (t ) can be expressed as [see Eq. (5.37)]

n (t ) |n (t ) = 0| D ( (t )) a
a D ( (t )) |0
=

= 2i ||2 Im ,
(13.109)
thus for the current case, for which (t ) is real, n (t ) |n (t ) = 0, and

consequently the geometrical phase n (t) vanishes. Furthermore, for the
current case H = xf, and thus [see Eqs. (5.11), (5.28) and (5.29)]
n (t)| H |0 (t) = 21/2 x0 f n,1 .
Eyal Buks
(13.110)
412
13.8. Solutions
With the help of the above results one finds that
/
x20

i0 t
pn0 = 2 2
dt e
f
n,1 ,
2 0
or
pn0 = n,1 ,
(13.111)
where
=
2
1 2 2
e 2 0 .
2m 0
(13.112)
Note that the above result is identical to (10.68). Note also that the exact
result of this problem is given by [see Eq. (5.339)]
pn0 =
Eyal Buks
e n
.
n!
(13.113)
413
14. The Quantized Electromagnetic Field
This chapter discusses the quantization of electromagnetic (EM) field for the
relatively simple case of a free space cavity.
14.1 Classical Electromagnetic Cavity

Consider an empty volume surrounded by conductive walls having infinite
conductivity. The Maxwells equations (in Gaussian units) are given by
1 E
,
c t
1 B
E =
,
c t
E = 0 ,
B = 0 ,
B =
(14.1)
(14.2)
(14.3)
(14.4)
where c = 2.99 108 m s1 is the speed of light in vacuum. In the Coulomb

gauge the vector potential A is chosen such that
A =0 ,
(14.5)
and the scalar potential vanishes in the absence of sources (charge and
current). In this gauge both electric and magnetic fields E and B can be
expressed in terms of A only as [see Eqs. (1.41) and (1.42)]
1 A
,
c t
(14.6)
B=A.
(14.7)
E=
and

2 A =
1 2A
.
c2 t2
(14.8)
Chapter 14. The Quantized Electromagnetic Field

Solution 14.1.1. The gauge condition (14.5) and Eqs. (14.6) and (14.7)
guarantee that Maxwells equations (14.2), (14.3), and (14.4) are satisfied
E=
1 ( A)
1 B
=
,
c
t
c t
1 ( A)
=0,
c
t
B = ( A) = 0 ,
E=
(14.9)
(14.10)
(14.11)
where in the last equation the general vector identity ( A) = 0 has

been employed. Substituting Eqs. (14.6) and (14.7) into the only remaining
nontrivial equation, namely into Eq. (14.1), leads to
( A) =
1 2A
.
c2 t2
(14.12)
Using the vector identity

( A) = ( A) 2 A ,
(14.13)
and the gauge condition (14.5) one finds that

2 A =
1 2A
.
c2 t2
(14.14)
Exercise 14.1.2. Consider a solution having the form

A = q (t) u (r) ,
(14.15)
where q (t) is independent on position r and u (r) is independent on time t.

Show that q (t) and u (r) must satisfy
2 u+2 u = 0 ,
(14.16)
d2 q
+ 2 q = 0 ,
dt2
(14.17)
and
where is a constant and where

= c .
(14.18)
Solution 14.1.2. The gauge condition (14.5) leads to

u =0 .
(14.19)
From Eq. (14.8) one finds that

Eyal Buks
416
14.1. Classical Electromagnetic Cavity

q2 u =
1 d2 q
u
.
c2 dt2
(14.20)
Multiplying by an arbitrary unit vector n

leads to
2
u n
1 d2 q
= 2
.
un
c q dt2
(14.21)
The left hand side of Eq. (14.21) is a function of r only while the right hand
side is a function of t only. Therefore, both should equal a constant, which is
denoted as 2 , thus
2 u+2 u = 0 ,
(14.22)
d2 q
+ 2 q = 0 ,
dt2
(14.23)
and
where
= c .
(14.24)
Exercise 14.1.3. Show that the general solution can be expanded as

A=
qn (t) un (r) .
(14.25)
n
where the set {un } forms a complete orthonormal basis spanning the vector
space of all solutions of Eq. (14.16) satisfying the proper boundary conditions
on the conductive walls having infinite conductivity.
Solution 14.1.3. Equation (14.16) should be solved with the boundary conditions of a perfectly conductive surface. Namely, on the surface S enclosing
the cavity we have B
s = 0 and E
s = 0, where
s is a unit vector normal
to the surface. To satisfy the boundary condition for E we require that u be
normal to the surface, namely, u =
s (u
s) on S. This condition guarantees
also that the boundary condition for B is satisfied. To see this we calculate
the integral of the normal component of B over some arbitrary portion S of
S. Using Eq. (14.7) and Stokes theorem one finds that

(B
s) dS = q
[( u)
s] dS
S
@S
=q
u dl ,
C
(14.26)
where the close curve C encloses the surface S . Thus, since u is normal to
the surface, one finds that the integral along the close curve C vanishes, and
therefore
Eyal Buks
417

(B
s) dS = 0 .
(14.27)
Since S is arbitrary we conclude that B

s =0 on S. Each solution of Eq.
(14.16) that satisfies the boundary conditions is called an eigen mode. As can
be seen from Eq. (14.23), the dynamics of a mode amplitude q is the same
as the dynamics of a harmonic oscillator having angular frequency =
c. Inner product between different solutions of Eq. (14.16) that satisfy the
boundary conditions can be defined as

u1 , u2
(u1 u2 ) dV ,
(14.28)
V
where the integral is taken over the volume of the cavity. Using Eq. (14.16)
one finds that

2

2 21
(u1 u2 ) dV =
u1 2 u2 u2 2 u1 dV .
(14.29)
V
Using Greens theorem one finds that

2

(u1 u2 ) dV =
(u1 [(
s ) u2 ] u2 [(
s ) u1 ]) dS .
2 21
V
(14.30)
Using Eq. (14.19), the boundary condition u =

s (u
s) on S, and writing =
s (
s )
s (
s ), we find that the right hand side of (14.30) vanishes.
Thus, solutions with different 2 are orthogonal to each other. Let {un } be a
complete orthonormal basis spanning the vector space of all solutions of Eq.
(14.16) satisfying the boundary conditions. For any two vectors in this basis
the orthonormality condition is

un , um =
(un um ) dV = n,m .
(14.31)
V
Using such a basis we can expand the general solution as

A=
qn (t) un (r) .
(14.32)
Exercise 14.1.4. Show that the total electric energy in the cavity is given
by
1 2
UE =
q ,
(14.33)
8c2 n n
and the total magnetic energy is given by
1 2 2
UB =
q .
8 n n n
Eyal Buks
(14.34)
418
14.1. Classical Electromagnetic Cavity

Solution 14.1.4. Using Eqs. (14.6),(14.7) and (14.25) one finds that the
fields are given by
1
E=
qn un ,
(14.35)
c n
and
B=

n
qn un .
(14.36)
The total energy of the field is given by UE +UB , where UE (UB ) is the energy
associated with the electric (magnetic) field, namely,

1
UE =
E2 dV ,
(14.37)
8 V
and
1
UB =
8
B2 dV .
(14.38)

UE =
1 2
q ,
8c2 n n
and using Eq. (14.36) one finds that

1
UB =
qn qm
( un ) ( um ) dV .
8 n,m
V
(14.39)
(14.40)
The last integral can be calculated by using the vector identity

(F1 F2 ) = ( F1 ) F2 F1 ( F2 ) ,
(14.41)
applied to un ( um ), thus
( un )( um ) = (un ( um ))+un [ ( um )] . (14.42)
Using the divergence theorem and the fact that un and ( um ) are orthogonal to each other on S one finds that the volume integral of the first
term vanishes. To calculate the integral of the second term it is convenient
to use the identity
( um ) = ( um ) 2 um .
(14.43)
This together with Eqs. (14.19), (14.16), and (14.31) lead to

UB =
Eyal Buks
1 2 2
q .
8 n n n
(14.44)
419

The Lagrangian of the system is given by [see Eq. (1.16)]

1 qn2
2n qn2
L = UE UB =
,
4c2 n
2
2
(14.45)
where the symbol overdot is used for derivative with respect to time, and
where n = cn . The Euler-Lagrange equations, given by

d L
L
=0,
(14.46)
dt qn
qn
lead to Eq. (14.23).
The variable canonically conjugate to qn is [see Eq. (1.20)]
pn =
L
1
=
qn .
qn
4c2
(14.47)
The classical Hamiltonian HF of the field is given by [see Eq. (1.22)]

4c2 p2
1 2n qn2
n
HF =
pn qn L =
+
.
(14.48)
2
4c2 2
n
n
The Hamilton-Jacobi equations of motion, which are given by
qn =
HF
= 4c2 pn ,
pn
p n =
HF
2
= n2 qn ,
qn
4c
(14.49)
(14.50)
lead also to Eq. (14.23). Note that, as expected, the following holds
HF = UE +UB ,
(14.51)
namely the Hamiltonian expresses the total energy of the system.
14.2 Quantum Electromagnetic Cavity

The coordinates qn and their canonically conjugate variables pn are regarded
as Hermitian operators satisfying the following commutation relations [see
Eqs. (3.5) and (4.41)]
[qn , pm ] = i n,m ,
(14.52)
[qn , qm ] = [pn , pm ] = 0 .
(14.53)
and
Eyal Buks
420
14.2. Quantum Electromagnetic Cavity

In general, the Heisenberg equation of motion (4.37) of an operator A(H) is
given by
dA(H)
1 3 (H) (H) 4 A(H)
=
A , HF +
.
dt
i
t
(14.54)
qn = 4c2 pn ,
(14.55)
Thus, with the help of Eq. (14.48) one finds that
and
p n =
2n
qn .
4c2
(14.56)
It is useful to introduce the annihilation and creation operators

0

n
4ic2 pn
in t
an = e
qn +
,
(14.57)
8c2
n
0

n
4ic2 pn
q
an = ein t
.
(14.58)
n
8c2
n
The phase factor ein t in the definition of an is added in order to make it

time independent. The inverse transformation is given by
:

2c2 in t
qn =
e
an + ein t an ,
(14.59)
n
0

n in t
pn = i
e
an + ein t an .
(14.60)
2
8c
The commutation relations for the these operators are derived directly from
Eqs. (14.52) and (14.53)
,
an , am = n,m ,
(14.61)
, [an , am ] = an , am = 0 .
(14.62)
The Hamiltonian (14.48) can be expressed using Eqs. (14.59) and (14.60) as

1
HF =
n an an +
.
(14.63)
2
n
The eigenstates are the photon-number states |s1, s2 , ..., sn , ..., which satisfy
[see chapter 5]

1
HF |s1, s2 , , sn , =
n sn +
|s1, s2 , , sn , . (14.64)
2
n
The following holds [see Eqs. (5.28) and (5.29)]
Eyal Buks
421
an |s1, s2 , , sn , = sn |s1, s2 , , sn 1, ,
an |s1, s2 , , sn , = sn + 1 |s1, s2 , , sn + 1, .
(14.65)
(14.66)
The non-negative integer sn is the number of photons occupying mode n.

The vector potential A (14.25) becomes
:
2c2

A (r, t) =
ein t an + ein t an un (r) .
(14.67)
n
n
14.3 Periodic Boundary Conditions

Consider the case where the EM field is confined to a finite volume V , which
for simplicity is taken to have a cube shape with edge L = V 1/3 . The
eigen modes and eigen frequencies of the EM field are found in exercise 1
of this chapter for the case where the walls of the cavity are assumed to
have infinite conductance [see Eq. (14.89)]. It is however more convenient to
assume instead periodic boundary conditions, since the spatial dependence
of the resulting eigen modes [denoted by un (r)], can be expressed in terms of
exponential functions, rather than trigonometric functions [see Eqs. (14.82),
(14.83) and (14.84)]. For this case Eq. (14.63) becomes

1
HF =
,
(14.68)
k ak, ak, +
2
k,
and Eq. (14.67) becomes

:

2c2
A (r, t) =
k, ei(krk t) ak, +
k, ei(krk t) ak, , (14.69)
k V
k,
where the eigen frequencies are given by k = c |k|. In the limit of large
volume the discrete sum over wave vectors k can be replaced by an integral

V
dk
dk
dkz ,
(14.70)
x
y
(2)3
k
and the commutation relations between field operators become
3
4
[ak, , ak , ] = ak, , ak , = 0 ,
3
4

ak, , ak , = , k k .
(14.71)
(14.72)
The sum over contains two terms corresponding to two polarization

vectors
k, , which are normalized to unity and mutually orthogonal, i,e.
k,
k, = , . Furthermore, the Coulomb gauge condition requires that
Eyal Buks
422
14.4. Problems
k, k =
k, k = 0, i.e. the polarization vectors are required to be orthogonal
to the wave vector k. Linear polarization can be represented by two mutually
orthogonal real vectors
k,1 and
k,2 , which satisfy
k,1
k,2 = k/ |k|. For
the case of circular polarization the polarization vectors can be chosen to be
given by
1
k,+ = (
k,1 + i
k,2 ) ,
2
1
k, = (
k,1 i
k,2 ) .
2
For this case of circular polarization the following holds
k,
k, = , ,
k
k,
k, = i , .
|k|
(14.73)
(14.74)
(14.75)
(14.76)
where {+, }.
14.4 Problems
1. Find the eigen modes and eigen frequencies of a cavity having a pizza
box shape with volume V = L2 d.
2. Consider two perfectly conducting metallic plates placed in parallel to
each other. The gap between the plates is d and the temperature is assumed to be zero. Calculate the force per unit area acting between the
plates (the Casimir force).
3. Find the average energy per unit volume of the electromagnetic field in
thermal equilibrium at temperature T .
4. Consider an electromagnetic cavity having a set of normal modes. The
waveform of mode n is denoted by un (r), the angular frequency by n ,
the annihilation operator by an , and the creation operator by an . The
electric filed operator at point r and time t can be expressed as [see Eqs.
(14.6) and (14.67)]
E (r, t) = E() (r, t) + E(+) (r, t) ,
(14.77)
where
E() =
E(+) =

2n iein t un (r) an ,
(14.78)

2 n iein t un (r) an .
(14.79)
The correlation function G(l) (r1 , , rl ; rl+1 , , r2l ) of degree l is defined by

Eyal Buks
423

'
(
G(l) (r1 , , rl ; rl+1 , , r2l ) E () (r1 ) E () (rl ) E (+) (rl+1 ) E (+) (r2l ) .
(14.80)
The normalized coherence function g(l) (r1 , , rl ; rl+1 , , r2l ) of degree l is defined by
g (l) (r1 , , rl ; rl+1 , , r2l )
G(l) (r1 , , rl ; rl+1 , , r2l )

. (14.81)
2l
>
G(1) (rs ; rs )
s=1
Consider the case where all modes in the cavity are in their ground state
except of a single mode, which is in a number state with m photons.
Calculate g(l) (r1 , , rl ; rl+1 , , r2l ) for such a state.
14.5 Solutions
1. We seek solutions of Eq. (14.16) satisfying the boundary condition that
the tangential component of u vanishes on the walls. Consider a solution
having the form
0
8
(14.82)
ux (r) =
ax cos (kx x) sin (ky y) sin (kz z) ,
V
0
8
uy (r) =
(14.83)
ay sin (kx x) cos (ky y) sin (kz z) ,
V
0
8
uz (r) =
az sin (kx x) sin (ky y) cos (kz z) .
(14.84)
V
While the boundary condition on the walls x = 0, y = 0, and z = 0 is
guaranteed to be satisfied, the boundary condition on the walls x = L,
y = L, and z = d yields
nx
kx =
,
(14.85)
L
ny
ky =
,
(14.86)
L
nz
kz =
,
(14.87)
d
where nx , ny and nz are integers. This solution clearly satisfies Eq.
(14.16), where the eigen value is given by
.
= kx2 + ky2 + kz2 .
(14.88)
Alternatively,
. using the notation n = (nx , ny , nz ) one has = (/L) n,
where n = n2x + n2y + n2z . Using Eq. (14.24) one finds that the angular
frequency of a mode characterized by the vector of integers n is given by
Eyal Buks
424
14.5. Solutions
n = c
0
n 2
x
n 2
y
n 2
z
(14.89)
In addition to Eq. (14.16) and the boundary condition, each solution has
to satisfy also the transversality condition u = 0 (14.19), which in
the present case reads
ka= 0 ,
(14.90)
where k = (kx , ky , kz ) and a = (ax , ay , az ). Thus, for each set of integers

{nx , ny , nz } there are two orthogonal modes (polarizations), unless nx =
0 or ny = 0 or nz = 0. In the latter case, only a single solution exists.
The inner product between two solutions u1 and u1 having vectors of
integers n1 = (nx1 , ny1 , nz1 ) and n2 = (nx2 , ny2 , nz2 ), and vectors of
amplitudes a1 = (ax1 , ay1 , az1 ) and a2 = (ax2 , ay2 , az2 ), respectively, can
be calculated using Eq. (14.28)

u1 , u2 =
(u1 u2 ) dV
V

=
(ux1 ux2 + uy1 uy2 + uz1 uz2 ) dV .
V
(14.91)
The following holds

ux1 ux2 dV
V
8
= ax1 ax2
V
L
n
n
x1
x2
cos
x cos
x dx
L
L
0
L
n
n
y1
y2
sin
y sin
y dy
L
L
0
d
n
n
z1
z2
sin
z sin
z dz ,
d
d
0
8
L2 d
= ax1 ax2
nx1 ,nx2 ny1 ,ny2 nz1 ,nz2 .
V
8
(14.92)
Similar results are obtained for the contribution of the y and z components. Thus
u1 , u2 = (a1 a2 ) nx1 ,nx2 ny1 ,ny2 nz1 ,nz2 ,
(14.93)
and therefore the vectors of amplitudes a are required to be normalized,

i.e. to satisfy a a = 1, in order to ensure that the solutions u are
normalized.
Eyal Buks
425

2. Employing the results of the previous exercise, the eigen frequencies n
are taken to be given by Eq. (14.89), where L is assumed to be much
larger than d. As can be seen from Eq. (14.64), each mode contributes
energy of n /2 to the total energy of the ground state of the system,
which is denoted by E (d). Let E () be the ground state energy in the
limit where d and let U (d) = E (d)E () be the potential energy
of the system. Formally, both E (d) and E () are infinite, however, as
we will show below, the divergence can be regulated when evaluating the
difference U (d). The assumption that L is large allows substituting the
discrete sums over nx and ny by integrals when evaluating E (d) and
E (). Moreover the discrete sum over nz is substituted by an integral
in the expression for E (). The prime on the summation symbol over
nz in the expression for E (d) below implies that a factor of 1/2 should
be inserted if nz = 0, when only one polarization exists (see previous
exercise). Using these approximations and notation one has
U (d) = E (d) E ()
0
2

n 2
L
z
= c
dkx
dky kx2 + ky2 +
d
0
nz 0
2

.
L
d
c
dkx
dky
dkz kx2 + ky2 + kz2 .
0
0
0
(14.94)
.
In polar coordinates u =
kx2 + ky2 and = tan1 (ky /kx ) one has
dkx dky = udud, thus
0
2

n 2
L
z
U (d) = c
du u u2 +
2 n 0
d
z
2

L
d
c
du u
dkz u2 + kz2 .
2 0
0
(14.95)
Changing the integration variables
2
ud
x=
,
kz d
Nz =
,
leads to
Eyal Buks
(14.96)
(14.97)
426
14.5. Solutions
2 cL2
U (d) =
4d3
2 cL2
=
4d3
!

!
nz
F (nz )
"
dNz F (Nz )
"

1
F (0) +
F (nz )
dNz F (Nz ) ,
2
0
n =1
z
where the function F () is given by

.

2
F () =
dx x + =
dy y .
(14.98)
(14.99)
Formally, the function F () diverges. However, the following physical

argument can be employed in order to regulate this divergency. The assumption that the walls of the cavity perfectly conduct is applicable at
low frequencies. However, any metal becomes effectively transparent in
the limit of high frequencies. Thus, the contribution to the ground state
energy of high frequency modes is expected to be effectively d independent, and consequently U (d) is expected to become finite. Based on this
argument the divergency in F () is removed by introducing a cutoff function f (y) into the integrand in Eq. (14.99)

dy yf (y) .
(14.100)
F () =
2
While near y = 0 (low frequencies) the cutoff function is assumed to be

given by f (y) = 1, in the limit of large y (high frequencies) the function
f (y) is assumed to approach zero sufficiency fast to ensure that F () is
finite for any . Moreover, it is assumed that F () 0. In this case the
Euler-Maclaurin summation formula, which is given by
<
1
F (n)
F (0) +
2
n=1
dN F (N) =
1
1
F (0) +
F (0) + ,
12
720
(14.101)
can be employed to evaluate U (d). The following holds

F () = 2 2 f 2 ,
(14.102)
thus for small [where the cutoff function is assumed to be given by

f (y) = 1] F () = 4 and F () = 4, and therefore
U (d) =
2 cL2
.
720d3
(14.103)
The force per unit area (pressure) P (d) is found by taking the derivative
with respect to d and by dividing by the area L2
Eyal Buks
427

P (d) =
2 c
.
240d4
(14.104)
The minus sign indicates that the force is attractive.

3. The average energy U in thermal equilibrium is given by Eq. (8.448),
which is given by
log Zc
,
(14.105)

where Zc = Tr eH is the canonical partition function, H is the
Hamiltonian [see Eq. (14.64)], = 1/kB T and kB is the Boltzmanns
constant. The partition function is found by summing over all photonnumber states |s1, s2 , ...
U =
Zc =
s1, s2 ,...=0
s1, s2 , ...| eH |s1, s2 , ...

e
s1, s2 ,...=0
#
n
#
n
<
n (sn + 12 )
sn =0
n (sn + 12 )
1
n
2 sinh
2
"
"
,
(14.106)
where n labels the cavity modes. Using the last result one finds that
U =
=
=
log Zc
log
n
n
n
1
2 sinh 2n
coth
n
.
2
(14.107)
It is easy to see that the above sum diverges since the number of modes
in the cavity is infinite. To obtain a finite result we evaluate below the
difference Ud = U (T ) U (T = 0) between the energy at temperature
T and the energy at zero temperature, which is given by (recall that
coth (x) 1 in the limit x )
Eyal Buks
428
14.5. Solutions

n
n
coth
1
2
2
n
n
=
.
en 1
n
Ud =
(14.108)
The angular frequencies n of the modes are given by Eq. (14.89). For
simplicity a cubical cavity having volume V = L3 is considered. For this
case Ud is given by (the factor of 2 is due to polarization degeneracy)
Ud = 2kB T

nx =0 ny =0 nz
where n =
given by
=
n
.
n 1
e
=0
(14.109)
.
n2x + n2y + n2z , and where the dimensionless parameter is
c
.
L
(14.110)
In the limit where 1 (macroscopic limit) the sum can be approximated by the integral

4
n
Ud = 2kB T
dn n2 n
8
e 1
0
kB T
3
x3 dx
,
ex 1
0

4
15
(14.111)
thus the energy per unit volume is given by

Ud
2 (kB T )4
=
.
V
153 c3
(14.112)
4. When only a single mode in the cavity is excited the normalized coherence function g(l) (r1 , , rl ; rl+1 , , r2l ) becomes [see Eqs. (14.78) and
(14.79) and the definition of g (l) ]
' l (
a al
(l)
,
(14.113)
g (r1 , , rl ; rl+1 , , r2l ) =
l
a a
where a and a are the annihilation and creation operators
of the excited
cavity mode. With the help of the relation a |m = n |m 1 [see Eq.
Eyal Buks
429

(5.28)] one finds that for the given single mode m photon state g(l) is
given by
+
m!
l m l
g(l) = (ml)!m
.
(14.114)
0
m<l
Eyal Buks
430
15. Light Matter Interaction
In this chapter the transitions between atomic states that result from interaction with an electromagnetic (EM) field are discussed.
15.1 Hamiltonian
Consider an atom in an EM field. The classical Hamiltonian HF of the EM
field is given by Eq. (14.48). For the case of hydrogen, and in the absence
of EM field, the Hamiltonian of the atom is given by Eq. (7.2). In general,
the classical Hamiltonian of a point particle having charge e and mass me
in an EM field having scalar potential and vector potential A is given by
Eq. (1.62). In the Coulomb gauge the vector potential A is chosen such that
A = 0, and the scalar potential vanishes provided that no sources
(charge and current) are present. The EM field is assumed to be sufficiently
small to allows employing the following approximation

e 2
e
p A p2 2 A p ,
(15.1)
c
c
where p is the momentum vector. Recall that in the Coulomb gauge the vector
operators p and A satisfy the relation p A = A p, as can be seen from
Eqs. (6.167) and (6.324). These results and approximation allow expressing
the Hamiltonian of the system as
H = H0 + HF + Hp ,
(15.2)
where H0 is the Hamiltonian of the atom in the absence of EM field, and

where Hp , which is given by
Hp =
e
Ap ,
me c
(15.3)
is the coupling Hamiltonian between the atom and the EM field.

In the quantum case The Hamiltonian HF of the EM field is given by Eq.
(14.68)

1
HF =
k ak, ak, +
,
(15.4)
2
k,
Chapter 15. Light Matter Interaction

and the vector potential A is given by Eq. (14.69)
:

2c2
k, ei(krk t) ak, +
A (r, t) =
k, ei(krk t) ak, . (15.5)
k V
k,
15.2 Transition Rates

While the Hamiltonian Hp is considered as a perturbation, the unperturbed
Hamiltonian is taken to be H0 + HF . The eigenvectors of H0 + HF are labeled as |{sk, } , . While the integers sk, represent the number of photons
occupying each of the modes of the EM field, the index labels the atomic
energy eigenstate. The following holds
H0 |{sk, } , = E |{sk, } , ,
where E is the energy of the atomic state, and

1
HF |{sk, } , =
k sk, +
|{sk, } , .
2
(15.6)
k,
15.2.1 Spontaneous Emission

Consider the case where the system is initially in a state |i = |{sk, = 0} , i ,
for which all photon occupation numbers are zero, and the atomic state is labeled by the index i . The final state is taken to be |f = ak, |{sk, = 0} , f ,
for which one photon is created in mode k, , and the atomic state is labeled
by the index f . To lowest nonvanishing order in perturbation theory the
transition rate wi,f is given by Eq. (10.34)
2
( k i,f ) |f| Hp |i|2 ,
(15.7)
2

where i,f = Ei Ef /. With the help of Eqs. (14.69) and (15.3) wi,f
can be rewritten as

2 2 2
2
e
4 c
wi,f =
( k i,f )
f|
k, peikr ak, |i
.
(15.8)
me c
k V
wi,f =

[H0 , r] =
1
(i) p ,
me
(15.9)
thus
Eyal Buks
432
15.2. Transition Rates
2
42 e2 k
( k i,f )
f|
k, reikr ak, |i
V
42 e2 k
( k i,f ) |Mi,f |2 ,
=
V
wi,f =
(15.10)
where the atomic matrix element Mi,f is given by
k, reikr |i .
Mi,f = f |
(15.11)
15.2.2 Stimulated Emission and Absorption

The process of spontaneous emission of a photon in mode k, can be labeled
as (i,sk, ) (f,sk, + 1), where sk, = 0. In the case of stimulated emission,
on the other hand, the initial photon occupation is assumed to be nonzero,
(e)
i.e. sk, 1. Let w(i,sk, )(f,sk, +1), be the rate of emission of photons in
mode k, , given that the initial photon occupation number is sk, . With
the help of Eq. (14.66) the expression for the case of spontaneous emission
(15.10) can be easily generalized for arbitrary initial photon occupation sk,
(e)
w(i,sk, )(f,sk, +1), =
42 e2 k (sk, + 1)
(k i,f ) |Mi,f |2 .
V
(15.12)
Note that for the case of emission it is assumed that the energy of the
atomic
state

i is larger than the energy of the atomic state f, i.e. i,f =
Ei Ef / > 0.
(a)
Absorption is the reverse process. Let w(i,sk, )(f,sk, 1), be the rate of
absorption of photons in mode k, , given that the initial photon occupation
number is sk, . With the help of Eq. (14.65) one finds using a derivation
similar to the one that was used above to obtain Eq. (15.12) that
(a)
w(i,sk, )(f,sk, 1), =
4 2 e2 k sk,
( k + i,f ) |Mi,f |2 .
V
(15.13)
Note that in this case it is assumed that i,f < 0.

The emission (15.12) and absorption (15.13) rates provide the contribu(e)
(a)
tion of a single mode of the EM field. Let d(i,s)(f,s+1), /d (d(i,s)(f,s1), /d)
be the total emitted (absorbed) rate in the infinitesimal solid angle d having
polarization . For both cases s denotes the photon occupation number of the
initial state. To calculate these rates the contributions from all modes in the
EM field should be added. In the limit of large volume the discrete sum over
wave vectors k can be replaced by an integral according to Eq. (14.70). Note
that both the emitted and absorbed radiations are expected to be isotropic.
Thus, by using the relation k = ck, where k = |k|, one finds that
Eyal Buks
433

(e)
d(i,s)(f,s+1),
d

V
(e)
dk k2 w(i,sk, )(f,sk, +1),
(2)3 0

e2 (s + 1)
2
=
|Mi,f |
dx x3 (x i,f )
2c3
0
fs (s + 1) 3i,f
2
=
|Mi,f | .
2c2
=
(15.14)
2
where fs = e /c 1/137 is the fine-structure constant. In a similar way,

one finds for the case of absorption that
(a)
d(i,s)(f,s+1),
d
fs s3i,f
|Mi,f |2 .
2c2
(15.15)
15.2.3 Selection Rules

While the size of an atom aatom is on the order of the Bohrs radius a0 =
0.531010 m (7.64), the energy difference Ei Ef is expected to be on the
order of the ionization
energy of hydrogen atom EI = 13.6 eV (7.66). Using

the relation k = Ei Ef / = ck one finds that aatom k 103 . Thus,
to a good approximation the term eikr in the expression for the matrix
element Mi,f can be replaced by unity
Mi,f f |
k, r |i .
(15.16)
This approximation is called the dipole approximation.

The atomic energy eigenstates | can be chosen to be also eigenvectors
of the angular momentum operators Lz and L2 . It is convenient to employ
the notation |k, l, m, to label these states, where k, l and m are orbital
quantum numbers and where labels the spin state. As can be seen from
Eqs. (7.42), (7.43) and (7.44) the following holds
H0 |k, l, m, = Ekl |k, l, m, ,
L2 |k, l, m, = l (l + 1) 2 |k, l, m, ,
Lz |k, l, m, = m |k, l, m, .
(15.17)
(15.18)
(15.19)
Since it is assumed that no magnetic field is externally applied, the eigenenergies Ekl are taken to be independent on the quantum numbers m and .
Radiation transitions between a pair of states |ki , li , mi , i and |kf , lf , mf , f
can occur only when the corresponding matrix element (15.16) is nonzero.
This requirement yields some conditions known as selection rules. The first
one refers to the spin quantum number . Note that Mi,f is a matrix element
of an orbital operator (15.16), and consequently it vanishes unless f = i , or
alternatively, unless = f i = 0. It is important to keep in mind that
this selection rule is valid only when spin-orbit interaction can be neglected.
Eyal Buks
434
15.2. Transition Rates

Exercise 15.2.1. Show that the selection rule for the magnetic quantum
number m is given by
m = mf mi {1, 0, 1} .
(15.20)
Solution 15.2.1. Using the relations Lz = xpy ypx and [xi , pj ] = i ij

it is easy to show that [Lz , z] = 0 and [Lz , x iy] = (x iy). The first
relation together with Eq. (15.19) imply that
0 = kf , lf , mf , f | [Lz , z] |ki , li , mi , i
= (mf mi ) kf , lf , mf , f | z |ki , li , mi , i ,
(15.21)
whereas the second relation together with Eq. (15.19) imply that
kf , lf , mf , f | [Lz , x iy] |ki , li , mi , i
= (mf mi ) kf , lf , mf , f | x iy |ki , li , mi , i
= kf , lf , mf , f | (x iy) |ki , li , mi , i ,
(15.22)
thus
(mf mi 1) kf , lf , mf , f | x iy |ki , li , mi , i = 0 .
(15.23)
Therefore Mi,f = 0 [see Eq. (15.16)] unless m {1, 0, 1}. The transition
m = 0 is associated with linear polarization in the z direction, whereas
the transitions m = 1 are associated with clockwise and counterclockwise
circular polarizations respectively.
Exercise 15.2.2. Show that the selection rule for the quantum number l is
given by
l = lf li {1, 1} .
Solution 15.2.2. Using Eq. (15.40), which is given by

, 2 , 2 -
L , L , r = 22 rL2 +L2 r ,
(15.24)
(15.25)
together with Eq. (15.18) yield

,
,
-kf , lf , mf , f | L2 , L2 , r |ki , li , mi , i
= 24 (lf (lf + 1) + li (li + 1)) kf , lf , mf , f | r |ki , li , mi , i
= 4 (lf (lf + 1) li (li + 1))2 kf , lf , mf , f | r |ki , li , mi , i ,
(15.26)
Eyal Buks
435

(lf (lf + 1) li (li + 1))2 2 (lf (lf + 1) + li (li + 1))
3
4
2
= (li + lf ) (li + lf + 2) (li lf ) 1 ,
(15.27)
one finds that

3
4
(li + lf ) (li + lf + 2) (li lf )2 1 kf , lf , mf , f | r |ki , li , mi , i = 0 . (15.28)
Since both li and lf are non negative integers, and consequently li +lf +2 > 0,
one finds that kf , lf , mf , f | r |ki , li , mi , i can be nonzero only when li =
lf = 0 or |l| = 1. However, for the first possibility, for which li = mi = lf =
mf = 0, the wavefunctions of both states |ki , li , mi , i and |kf , lf , mf , f is a
function of the radial coordinate r only [see Eq. (6.130)], and consequently
kf , lf , mf , f | r |ki , li , mi , i = 0. Therefore the selection rule is given by l
{1, 1}.
15.3 Semiclassical Case

Consider the case where one mode of the EM field, which has angular frequency and polarization vector
, is externally driven to a coherent state
|, where || 1. In the semiclassical approximation the annihilation operator of the driven mode a is substituted by the complex constant (and
the operator a by ). Furthermore, all other modes are disregarded. When
is close to a specific transition frequency a = (E+ E ) / between two
atomic states, which are labeled by |+ and |, the atom can be approximately considered to be a two level system. In the dipole approximation the
matrix element +| Hp | is given by [see Eqs. (15.3), (15.9) and (15.16)]
+| Hp | =
where
= 2ie a

it
e
eit ,
2
+| r | .
V
(15.29)
(15.30)
It is convenient to express the complex frequency as = 1 ei , where

both 1 and are real. The frequency 1 is called the Rabi frequency. Due
to selection rules the diagonal matrix elements of Hp vanish.
The Schrdinger equation is given by
i
d
| = H | ,
dt
(15.31)
where the matrix representation in the basis {|+ , |} of the Hamiltonian
H is given by [see Eq. (15.29)]
Eyal Buks
436
15.3. Semiclassical Case

H=

2
i(t+)

ei(t+)
i(t+)a i(t+) 1 e
.
1 e
e
a
It is convenient to express the general solution as

it
it
| (t) = b+ (t) exp
|+ + b (t) exp
| .
2
2
(15.32)
(15.33)
Substituting into the Schrdinger equation yields [see Eq. (6.244)]

i
d
dt
b+
b
1
2

1 ei ei(2t+)
b+

,
b
ei ei(2t+)
(15.34)
where
= a .
(15.35)
In the rotating wave approximation the rapidly oscillating terms ei(2t+)

are disregarded, since their influence in the long time limit is typically negligible. This approximation is equivalent to the assumption that the second
term in Eq. (15.29) can be disregarded. Furthermore, the phase factor can
be eliminated by resetting the time zero point accordingly. Thus, the Hamiltonian can be taken to be given by

a 1 eit
H=
,
(15.36)
2 1 eit a
and the equation of motion in the rotating frame can be taken to be given
by

d b+
1 1
b+
i
=
.
(15.37)
b
dt b
2 1
The time evolution is found using Eq. (6.138) [see also Eq. (6.248)]

b+ (t)
b (t)
sin

cos i
i 21 sin 2
2
2
b+ (0)
1 +()
1 +()
=
,
sin
b (0)
i 21 sin 2
cos + i
2
2
1 +()
1 +()
(15.38)
where
=
.
21 + ()2 t
Eyal Buks
(15.39)
437
15.4 Problems
1. Show that
, 2 , 2 -

L , L , r = 22 rL2 +L2 r .
(15.40)
2. Consider an atom having a set of orthonormal energy eigenstates {|n }.

The oscillator strength fnm associated with the transition between state
|n to state |m is defined by
fnm =
2me n,m
2
|f | r | i | .
3
Show that

fn,n = 1 .
(15.41)
(15.42)
3. Calculate the lifetime of all states of hydrogen atom having principle

quantum number n = 2.
4. Consider a hydrogen atom that is initially at time t in its ground
state. An electric field in the z direction given by
E (t) = E0
z
2
,
2 + t2
(15.43)
where is a constant having the dimension of time, is externally applied.

Calculate the probability p2p to find the atom in the sub-shell 2p at time
t .
5. Consider a particle having mass m and charge q moving in a one dimensional harmonic oscillator having angular resonance frequency . Calculate using the dipole approximation the rate of spontaneous emission
from the number state |n to the ground state |0.
6. A hydrogen atom is initially in its ground state. An electric field given
by E0 cos (t), where both E0 and are constants, is externally applied.
Assume that > EI , where EI is the ionization energy of the atom.
Calculate the rate of ionization.
15.5 Solutions
1. Using the relations [Lx , z] = iy, [Ly , z] = ix and [Lz , z] = 0 one finds
that
, 2 - , 2 - , 2 L , z = Lx , z + Ly , z
= i (Lx y yLx + Ly x + xLy )
= iV
z,
(15.44)
Eyal Buks
438
15.5. Solutions
,
where V = r L L r. Thus the following holds L2 , r = iV. With
the help of the identities
[Lx , Vz ] = Lx [Lx , y] [Lx , y] Lx + [Lx , Ly ] x + x [Lx , Ly ] = iVy ,
[Ly , Vz ] = [Ly , Lx ] y y [Ly , Lx ] + Ly [Ly , x] + [Ly , x] Ly = iVx ,
[Lz , Vz ] = [Lz , Lx y] [Lz , yLx ] + [Lz , Ly x] + [Lz , xLy ] = 0 ,
one finds that
, 2 , 2 -,
L , L , z = i L2 , Vz
= 2 (Lx Vy + Vy Lx Ly Vx Vx Ly )
= 2 (L V V L)
z,
(15.45)
thus
, 2 , 2 -L , L , r = 2 (L V V L)
= 2 (L (r L) L (L r) (r L) L+ (L r) L)

= 22 rL2 +L2 r .
(15.46)
2. Trivial by the Thomas-Reiche-Kuhn sum rule (4.69).

(se)
3. The rate of spontaneous emission if, in general is given by Eq. (15.14)
(se)
if, = 4
fs 3i,f
|Mi,f |2 .
2c2
(15.47)
The transition frequency i,f between the levels n = 2 to n = 1 is given

by [see Eq. (7.66)]

me e4
1
1
i,f =
2+ 2 ,
(15.48)
23
2
1
thus in terms of the Bohrs radius a0 [see Eq. (7.64)] one has
(se)
if, =
33 5fs me c2
28
Mi,f
2
a0
.
The matrix element of z = r cos and of u = 21/2 (x iy) =

21/2 r sin ei are given by [see Eq. (7.92)]
n , l , m | z |n, l, m

1
2

3
= dr r Rn l Rnl d (cos ) d cos Ylm
Ylm ,
Eyal Buks
(15.49)
439

and
n , l , m | u |n, l, m
0
1
2

1
3
dr r Rn l Rnl d (cos ) d sin ei Ylm
Ylm ,
=
2
0
(15.50)
where n, l and m are the quantum members of hydrogens energy eigenvectors. The final state is taken to be the ground state (n, l, m) = (1, 0, 0).
In the dipole approximation the transition (2, 0, 0) (1, 0, 0) is forbidden
due to the selection rule l {1, 1}. Using the identities
3/2
1
R10 (r) = 2
er/a0 ,
(15.51)
a0

3/2
r
1
R20 (r) = (2 r/a0 )
e 2a0 ,
(15.52)
2a0

3/2
r
1
r
e 2a0 ,
(15.53)
R21 (r) =
2a0
3a0
0
1
Y00 (, ) =
,
(15.54)
4
0
1
3
Y11 (, ) =
sin ei ,
(15.55)
2 2
0
1 3
Y10 (, ) =
cos ,
(15.56)
2
0
1
3
(15.57)
Y11 (, ) =
sin ei ,
2 2
where a0 is Bohrs radius [see Eq. (7.64)], one finds for the radial part
that
dr r R10 R21
27 6
=
a0 ,
35
(15.58)
and for the angular part that
Eyal Buks
440
15.5. Solutions
1
d (cos )
1
1
2
0
d (cos )
2
0
d (cos )
2
0

1
d cos Y00 Y10 = ,
3

d cos Y00 Y11 = 0 ,

d sin ei Y00 Y10 = 0 ,
(15.59)
(15.60)
(15.61)
and
1
d (cos )
2
0

d sin ei Y00 Y11 = 0 ,
0
0 1
2
0 1
1
1
i
d (cos ) d sin e
Y0 Y1 =
,
2
3
1
(15.63)
0 1
0
2
0 1
1
1
i
,
d (cos ) d sin e Y0 Y1 =
2
3
1
1
(15.62)
(15.64)
d (cos )
2
0

d sin ei Y00 Y11 = 0 .
(15.65)
Thus, by combining all these results one finds that the inverse lifetime of
the states (2, 1, 1), (2, 1, 0) and (2, 1, 1) is given by
(se) =
33 5fs me c2
28
7
2
35 2
,
(15.66)
whereas the lifetime of the state (2, 0, 0) is infinite (in the dipole approximation).
4. The probability p2pm to find the atom in the state |n = 2, l = 1, m is
calculated using Eq. (10.42) together with Eq. (7.84)
p2pm
e2 E02 2
=
2
where
EI =
3E
2
i 4 I t
dt e
|2, 1, m| z |1, 0, 0| . (15.67)

2
2
+t
e4
22
(15.68)
is the ionization energy. The following holds

Eyal Buks
441
dt ei
3EI
4 t
/
1
dxeix
=

2
+t
1 + x 2
/
dxeix
,
=
(x i) (x + i)
(15.69)
where
=
3EI
,
4
(15.70)
thus with the help of the residue theorem one finds that
dt ei
3EI
4 t
= e .
+ t2
(15.71)
The matrix element 2, 1, m| z |1, 0, 0 is calculated with the help of Eq.
(15.49)
2, 1, m| z |1, 0, 0 =

0
dr r R21 R10
2 2a0
=
m,0 ,
35
7
1
d (cos )
2
d cos (Y1m ) Y00
(15.72)
where
a0 =
2
e2
(15.73)
is the Bohrs radius, thus

2
3EI
215 eE0 a0
p2pm = 10
e 2 m,0 .
3
(15.74)
5. The oscillator is assumed to move along the z direction. The rate of

(se)
spontaneous emission |n|0, with polarization into solid angle d
is given in the dipole approximation by [see Eqs. (15.14) and (15.16)]
(se)
d|n|0, =

2
q 2 (n)3
|0| z |n|2
k,
z d ,
3
2c
(15.75)

2
q 2 (n)3 n
n,1
k,
z d .
3
2c 2m
(15.76)
where
k, is the polarization unit vector. With the help of Eqs. (5.11),
(5.28) and (5.29) one finds that
(se)
d|n|0, =
Eyal Buks
442
15.5. Solutions
Integrating over d in spherical coordinates and with the help of the
relation

1
4
,
(15.77)
d cos2 =
d
d (cos ) cos2 =
3
1
and summing over the two orthogonal polarization yields the total rate
of spontaneous emission
(se)
|n|0 =
2q 2 2
n,1 .
3mc3
(15.78)
6. The wave function of the final state |k has the form r |k = V 1/2 eik r ,
where V is the systemss volume. The perturbation that is induced by
the applied electric field can be expressed as H1 (t) = Keit + K eit ,
where
K=
eE0 r u
,
2
(15.79)
r = r (sin cos , sin sin , cos ) is the position vector operator and u
=
(sin 0 cos 0 , sin 0 sin 0 , cos 0 ) is a unit vector in the direction of the
applied electric field. The matrix element Mk = k | K |n = 1, l = 0, m = 0
corresponding to the transition from the ground state |n = 1, l = 0, m = 0
[see Eq. (7.92)] to the final state |k is given by (the z axis is taken to
be in the direction of k )
Mk
3/2
1/2 eE0 a0
=
2
V 1/2
dr r
1
d (cos )
2
d eik r cos er/a0 r

u,
(15.80)
where
ru
= r sin sin 0 cos ( 0 ) + r cos cos 0 ,
(15.81)
thus
Mk =
1/2
3/2
eE0 a0 V 1/2 cos 0
3/2
= 1/2 eE0 a0
Eyal Buks

0
r/a0 3
dr e
1
d (cos ) eik r cos cos

i eik r (k r+i)+eik r (k ri)
16a40 k a0
V 1/2 cos 0
3 .
i (k a0 )2 + 1
(k r)2
(15.82)
443

The rate of ionization w is obtained by summing over k [see Eq. (10.39)]
w=
2
(Ek ) |Mk |2 ,

(15.83)
where
Ek =
2 k2
+ EI
2me
(15.84)
is the change in the energy of the electron and where EI = me e4 /22 is

the ionization energy of the atom [see Eq. (7.66)]. Replacing the sum by
an integral according to (14.70) yields
256e2 E02 a30
w=
3

0
dk
2 k2
+ EI
2me
thus
w=
where
(k a0 )4

6 , (15.85)
(k a0 )2 + 1
256e2 me E02 a40

(k0 a0 )3

6 ,
33
(k0 a0 )2 + 1

2me ( EI )
k0 =
.
(15.86)
(15.87)
Note that for a given amplitude E0 the rate w obtains its maximum
value, which is given by [see Eq. (7.64)]
27 3 E02 a30
wmax =
,
(15.88)
16

when the angular frequency is chosen such that k0 a0 = 31/2 .
Eyal Buks
444
16. Identical Particles
This chapter reviews the identical particles postulate of quantum mechanics

and second quantization formalism. It is mainly based on the first chapter of
Ref. [6].
16.1 Basis for the Hilbert Space

Consider a system containing some integer number N of identical particles.
For the single particle case, where N = 1, the state of the system | can be
expanded using an orthonormal basis {|ai }i that spans the single particle
Hilbert space. Based on the single particle basis {|ai }i we wish to construct
a basis for the Hilbert space of the system for the general case, where N can
be any integer. This can be done in two different ways, depending on whether
the identical particles are considered to be distinguishable or indistinguishable
(see example in Fig. 16.1).
Suppose that the particles can be labelled by numbers as billiard balls.
In this approach the particles are considered as distinguishable. For this case
a basis for the Hilbert space of the many-particle system can be constructed
from all vectors having the form |1 : i1 , 2 : i2 , , N : iN . The ket vector
|1 : i1 , 2 : i2 , , N : iN represents a state having N particles, where the
particle that is labelled by the number m (m = 1, 2, , N ) is in the single
particle state |aim . Each ket vector |1 : i1 , 2 : i2 , , N : iN can be characterized by a vector of occupation numbers n
= (n1 , n2 , ), where ni is
the number of particles occupying the single particle state |ai . Let gn be
the number of different ket-vectors having the form |1 : i1 , 2 : i2 , , N : iN
that are characterized by the same vector of occupation numbers n
. It is easy
to show that
N!
gn = >
,
ni !
(16.1)
<
where N = i ni is the number of particles.
Alternatively, the particles can be considered as indistinguishable. In this
approach all states having the same vector of occupation numbers n
represent the same physical state, and thus should be counted only once. In
Chapter 16. Identical Particles

other words, when the particles are considered as indistinguishable the subspace corresponding to any given vector of occupation numbers n
is rather
than being gn - fold degenerate (as in the approach where the particles are
considered to be distinguishable) is taken to be nondegenerate. The identical
particle postulate of quantum mechanics states that identical particles should
be considered as indistinguishable. Consequently, a basis for the Hilbert space
of the many-particle system can be constructed from the set of ket vectors
{|
n}n . The ket vector |
n represents a state that is characterized by a vector
of occupation numbers n
= (n1 , n2 , ), where the integer ni is the number
of particles that are in the single particle state |ai . Such a basis is considered
to be both orthonormal, i.e.

n1 |
n2 = n 1 ,n2 ,
1 = n
2 and n 1 ,n2 = 0 otherwise, and complete
where n 1 ,n2 = 1 if n

|
n
n | = 1 .
(16.2)
(16.3)
It it convenient to introduce the creation operators ai . With analogy with

the case of a harmonic oscillator [see Eq. (5.32)] and the case of EM field [see
Eqs. (14.65) and (14.66)] the state |
n is expressed as

n
n
1
2
1
|
n =
a
a2
|0 ,
(16.4)
n1 !n2 ! 1
where |0 represents the state where all occupation numbers are zero. Equation (16.4) suggests that the creation operators ai maps a given state to a
state having additional particle in the single particle quantum state |ai . The
operator ai is the Hermitian conjugate of the annihilation operator ai . The
number operator Ni is defined by
Ni = ai ai .
(16.5)
In addition to the above discussed principle of indistinguishability, the

identical particle postulate of quantum mechanics also states that all particles
in nature are divided into two type: Bosons and Fermions. Moreover, while
for the case of Bosons, the creation and annihilation operators satisfy the
following commutation relations
3
4
[ai , aj ] = ai , aj = 0 ,
(16.6)
3
4
ai , aj = ij ,
(16.7)
for the case of Fermions the following holds
3
4
[ai , aj ]+ = ai , aj = 0 ,
+
3
4
ai , aj
= ij ,
+
Eyal Buks
(16.8)
(16.9)
446
16.1. Basis for the Hilbert Space
(a)
2
1
state 1 state 2 state 3 state 4

n1=0 n2=2 n3=1 n4=0
1
3

n1=0 n2=2 n3=1 n4=0
3
2

n1=0 n2=2 n3=1 n4=0
(b)

n1=0 n2=2 n3=1 n4=0
<
Fig. 16.1. In this example the number of particles is N = i ni = 3, where the
occupation numbers are given by n
= (n1 , n2 , n3 , n4 , ) = (0, 2, 1, 0, ). When
the particles are considered as distinguishable
> [see panel (a)] the corresponding
subspace is gn degenerate, where gn = N!/ ni ! = 3. On the other hand, when
i
the particles are considered as indistinguishable [see panel (b)], the corresponding
subspace is nondegenerate.
3
4
where ai , aj
denotes anti-commutation, i.e.
+
[A, B]+ = AB + BA
(16.10)
for general operators A and B.

Exercise 16.1.1. Show that for both Bosons and Fermions
[Ni , Nj ] = 0 .
Eyal Buks
(16.11)
447

Solution 16.1.1. For Bosons this result is trivial [see Eqs. (16.6) and (16.7)].
It is also trivial for Fermions when i = j. Finally, for Fermions when i = j
one has
Ni Nj = ai ai aj aj = ai aj ai aj = ai aj aj ai = aj ai aj ai = aj aj ai ai = Nj Ni .
(16.12)
16.2 Bosons
Based on Eqs. (16.2), (16.4), (16.6) and (16.7) a variety of results can be
obtained:
Exercise 16.2.1. Show that for Bosons
3 n 4
n1
ai , ai
= n ai
.
(16.13)
Solution 16.2.1. Trivial by Eq. (2.174), which states that for any operators
A and B
[A, B n ] = nB n1 [A, B] ,
(16.14)
and by Eq. (16.7).

ai |0 = 0 .
(16.15)
Solution 16.2.2. The norm of the vector ai |0 can be expressed with the
help of Eqs. (16.4) and (16.7)
3
4
0| ai ai |0 = 0| ai , ai + ai ai |0
= 0 |0 + 0, 0, , ni = 1, 0, |0, 0, , ni = 1, 0, ,

(16.16)
thus with the help of the normalization condition (16.2) one finds that
0| ai ai |0 = 0 and therefore ai |0 = 0.
Ni |
n = ni |
n .
(16.17)
Solution 16.2.3. With the help of Eqs. (16.4), (16.13) and (16.15) one finds
that
Eyal Buks
448
16.3. Fermions
Ni |
n = ai ai |
n
ni
n1 n2
1
a2
=
a1
ai ai ai
|0
n1 !n2 !
n1 n2
3 ni 4 ni
1
=
a2
+ ai
a1
ai ai , ai
ai |0
n1 !n2 !

n1
n2
ni 1
1
=
a2
ai ni ai
|0
a1
n1 !n2 !
= ni |
n .
(16.18)
ai |n1, n2 , , ni , = ni |n1, n2 , , ni 1, ,
ai |n1, n2 , , ni , = ni + 1 |n1, n2 , , ni + 1, .
(16.19)
(16.20)
Solution 16.2.4. Equation (16.20) follows immediately from Eqs. (16.4) and
(16.6). Moreover, with the help of Eqs. (16.4), (16.13) and (16.15) one finds
that
n1 n2
ni
1
ai |n1, n2 , , ni , =
ai ai
|0
a1
a2
n1 !n2 !
n1 n2
3 ni 4 ni
1
=
a1
a2
ai , ai
+ ai
ai |0
n1 !n2 !

n1
n2
ni 1
ni
=
a1
a2
ai
|0
n1 !n2 !
= ni |n1, n2 , , ni 1, .
16.3 Fermions
2
The anti-commutation relations (16.8) for the case i = j yields ai = 0. As
can be seen from Eq. (16.4), this implies that the only possible occupation
numbers ni are 0 and 1. This result is known as the Paulis exclusion principle,
according to which no more than one Fermion can occupy a given single
particle state. For Fermions Eq. (16.4) can be written as (recall that 0! =
1! = 1)
n1 n2
|
n = a1
a2
|0 ,
(16.21)
where ni {0, 1}.
Exercise 16.3.1. Show that for Fermions

ai |0 = 0 .
Eyal Buks
(16.22)
449

Solution 16.3.1. The norm of the vector ai |0 can be expressed with the
help of Eqs. (16.21) and (16.9)
3
4
0| ai ai |0 = 0| ai , ai ai ai |0
+
= 0 |0 0, 0, , ni = 1, 0, |0, 0, , ni = 1, 0, ,

(16.23)
thus with the help of the normalization condition (16.2) one finds that
0| ai ai |0 = 0 and therefore ai |0 = 0.
n = ni |
n ,
Ni |
(16.24)
where Ni = ai ai .
Solution 16.3.2. Using Eqs. (16.8), (16.9) and (16.21) one finds that
n1 n2
Ni |
n = ai ai a1
a2
|0
<
ni
2
nj n1 n2
= (1) j<i
a1
a2
ai ai ai
|0
n1 n2
ni
= a1
ai ai ai
|0 .
a2
(16.25)
For the case ni = 0 this

n = 0, whereas for the
yields
ni [see
3 Eq.
4 (16.22)] Ni |
case ni = 1 one has ai ai

= ai , ai ai ai = 1ai ai , thus Ni |
n = |
n.
+
Both cases are in agreement with Eq. (16.24).

<
ai |n1, n2 , , ni , = ni (1)j<i
nj
|n1, n2 , , ni 1, ,
<
ai |n1, n2 , , ni , = (1 ni ) (1)j<i
nj
(16.26)
|n1, n2 , , ni + 1, (16.27)
.
Solution 16.3.3. According to Eq. (16.8) ai aj = aj ai . For i = j this

2
= 0. These relations together with Eq. (16.21) leads to Eq.
yields ai
(16.27) (note that 1 ni = 1 if ni = 0 and 1 ni =0 if ni = 1). Similarly,
Eq. (16.26) is obtained by using the identity ai ai
considering both possibilities ni = 0 and ni = 1.
Eyal Buks
ni
= 1 ai ai and by
450
16.4. Changing the Basis
16.4 Changing the Basis

In the previous section the creation ai and annihilation ai operators were
defined based on a given single particle orthonormal basis {|ai }i . Consider
an alternative single particle basis {|bj }j , which is made of eigenvectors of
the single particle observable BSP , i.e. the following holds BSP
= BSP and
BSP |bj = j |bj ,
(16.28)
where j is the single particle eigenvalue corresponding to the eigenvector

|bj . Moreover, this basis is assumed to be both orthonormal, i.e.
bj |bj = j,j ,
(16.29)
and complete, i.e.

|bj bj | = 1 .
(16.30)
Exploiting the completeness of the original single particle orthonormal basis

{|ai }i , i.e. the fact that

|ai ai | = 1 ,
(16.31)
i
allows expressing the eigenvector |bj as

|bj =
ai |bj |ai .
(16.32)
The single particle state |ai can be expressed in the notation of many
particle states as ai |0, whereas the single particle state |bj can be expressed
as bj |0, where the operator bj , which is the creation operator of the single
particle state |bj , is given by [see Eq. (16.32)]

bj =
ai |bj ai .
(16.33)
i
The creation operator bj is the Hermitian conjugate of the annihilation operator

bj =
bj |ai ai .
(16.34)
i
An important example is the case where the single particle observable is

taken to be the position observable r. For this case Eq. (16.32) becomes

|r =
i (r ) |ai ,
(16.35)
i
Eyal Buks
451

where |r is a single particle position eigenvector, and where i (r ) = r |ai
is the wavefunction of the single particle state |ai .
Expressing the single particle state |ai in the notation of many particle
states as ai |0 allows expressing the single particle state |r in the notation
of many particle states as (r ) |0 [see Eq. (16.35)], where the operator
(r ), which is given by

(r ) =
i (r ) ai ,
(16.36)
i
is the Hermitian conjugate of the quantized field operator (r ), which is

given by

(r ) =
i (r ) ai .
(16.37)
i
Note that while i (r ) is a wave function, (r ) is an operator on the Hilbert

space of the many particle system.
,
Exercise 16.4.1. Calculate (r ) , (r ) , where [A, B] = AB BA
for general operators A and B, and where the minus sign is used for Bosons
and the plus sign for Fermions.
Solution 16.4.1. With he help of Eqs. (16.7) and (16.9) one finds that
3
4

,
(r ) , (r ) =
i (r ) i (r ) ai , ai
i,i
i (r ) i (r )

i
r |ai ai |r
= r |r ,
,
(r ) , (r ) = (r r ) .
(16.38)
(16.39)
Similarly, one finds that

,
[ (r ) , (r )] = 0 ,
(r ) , (r ) = 0 .

d3 r (r ) = N ,
Eyal Buks
(16.40)
(16.41)
(16.42)
452
16.5. Many Particle Observables

where
(r ) = (r ) (r ) ,
(16.43)
and where

Ni
N=
(16.44)
The operator (r ) is called the number density operator, and the operator
N is called the total number of particles operator.
Solution 16.4.2. Using the definition of (r ) one finds that

d3 r (r ) (r ) =
ai ai d3 r i (r ) i (r )
i,i
ai ai
=N .
(16.45)
16.5 Many Particle Observables

Observables of a system of identical particles must be defined and must be
represented by Hermitian operators in a way that is consistent with the principle of indistinguishability. Below we consider both, one-particle observables
and two-particle observables, and discuss their representation as operators
on the Hilbert space of the many-particle system.
16.5.1 One-Particle Observables
Consider a single particle observable such as the observable BSP , which was
introduced in the previous section [see Eqs. (16.28), (16.29), (16.30)]. It is
convenient to employ the single particle basis {|bj }j , which is made of singleparticle eigenvectors of BSP that satisfy BSP |bj = j |bj [see Eq. (16.28)],
in order to construct creation bj and annihilation bj operators. In the manyparticle case, the same physical variable that BSP represents for the single
particle case is represented by the operator B, which is given by

B=
j bj bj .
(16.46)
j
This can be seen by recalling that the operator bj bj represents the number
of particles in the single particle state |bj and that j is the corresponding
Eyal Buks
453

eigenvalue. With the help of Eqs. (16.28), (16.29), (16.30) and (16.33) (16.34)
the operator B can be expressed in terms of the operators ai and ai

B=
ai | BSP |ai ai ai .
(16.47)
i,i
16.5.2 Two-Particle Observables

Consider two-body interaction that is represented by an Hermitian operator
VTP on the Hilbert space of two-particle states. A basis for this Hilbert space
can be constructed using a given orthonormal basis for the single particle
Hilbert space {|bj }j . When the two particles are considered as distinguishable the basis of the Hilbert space of the two-particle states can be taken
to be {|j, j }j,j . The ket vector |j, j represents a state for which the first
particle is in single particle state |bj and the second one is in state |bj .
Assume the case where the single particle basis vectors |bj are chosen in
such a way that diagonalizes VTP , i.e.
VTP |j, j = vj,j |j, j ,
(16.48)
where the eigenvalue vj,j is given by

vj,j = j, j | VTP |j, j .
(16.49)
In the many-particle case, the same physical variable that VTP represents
for the two-particle case is represented by the operator V , which is given by
1
V =
vj,j bj bj bj bj .
(16.50)
2 j,j
To see that the above expression indeed represents the two particle interaction
consider the expectation value
n| V |
n with respect to the many body state
|
n = |n1, n2 , . The following holds [see Eqs. (16.6) , (16.7), (16.8) and
(16.9)]
bj bj bj bj = bj bj bj bj
3

4
= bj bj , bj bj bj bj
3

4
= bj bj , bj bj bj bj

= bj j,j bj bj bj
= Nj j,j + Nj Nj ,
(16.51)
where the upper sign is used for Bosons and the lower one for Fermions. Thus
V can be rewritten as
Eyal Buks
454
16.6. Hamiltonian
V =
1
vj,j Nj (Nj j,j ) .
2
(16.52)
j,j
Separating the terms for which j = j from the terms for which j = j yields
V =
vj,j Nj Nj +
j<j
1
vj,j Nj (Nj 1) ,
2 j
(16.53)
thus the matrix element

n| V |
n is given by

n| V |
n =
nj nj vj,j +
j<j
nj (nj 1)
vj,j .
2
j
(16.54)
While the factor nj nj represents the number of particle pairs occupying

single particle states j and j for the case j = j , the factor nj (nj 1) /2
represents the number of particle pairs occupying the same single particle
states j. Thus the above expression for V (16.50) properly accounts for the
two-particle interaction.
With the help of Eqs. (16.29), (16.30) and (16.33) (16.34) the operator V
can be expressed in terms of the operators ai and ai

1
V =
ai , ai |j, j j, j | VTP |j, j j , j |ai , ai ai ai ai ai
2 i ,i ,i ,i j,j

1
=
ai , ai |j, j j, j | VTP |j, j j, j |ai , ai ai ai ai ai
2 i ,i ,i ,i j,j

1
=
ai , ai |
|j, j j, j | VTP |ai , ai ai ai ai ai ,
2 i ,i ,i ,i
j,j
(16.55)
thus
V =

1
ai , ai | VTP |ai , ai ai ai ai ai .
2 i ,i ,i ,i
(16.56)
16.6 Hamiltonian
Consider the case where the single-particle Hamiltonian is given by
HSP = TSP + USP ,
(16.57)
where the operator TSP , which is given by

TSP =
Eyal Buks
p2SP
,
2m
(16.58)
455

where pSP is the single-particle momentum vector operator and where m is
the mass of a particle, is the single-particle kinetic energy operator, and
where the operator USP (r ) the the single-particle potential energy. The
many-particle kinetic energy operator is found using Eq. (16.47)
T =
1
ai | p2SP |ai ai ai .
2m i,i
(16.59)
The matrix element ai | p2SP |ai can be written using the wavefunctions
i (r ) = r |ai [recall Eq. (3.29), according to which r | p | = i
for a general state |]

2
d3 r (i (r )) (i (r )) .
(16.60)
ai | p2SP |ai =
2m
Thus, in terms of the quantized field operator (r ) [see Eqs. (16.36) and
(16.37)] the operator T can be expressed as

2
T =
d3 r (r ) (r ) .
(16.61)
2m
Integration by parts yields an alternative expression

2
T =
d3 r (r ) 2 (r ) .
2m
(16.62)
Similarly, the many-particle potential energy operator is found using Eq.

(16.47) [recall Eq. (3.23), according to which r | f (r) | = f (r ) (r ) for
a general state | and for a general function f (r)]

U =
ai | USP (r ) |ai ai ai
i,i
d3 r USP (r ) (r ) (r ) .
(16.63)
In addition, consider the case where the particles interact with each other
via a two-particle potential VTP (r1 , r2 ). The corresponding many-particle
interaction operator is found using Eq. (16.56). The two-particle matrix elements of VTP are given by
ai , ai | VTP |ai , ai

3
= d r
d3 r i (r ) i (r ) VTP (r , r ) i (r ) i (r ) ,
(16.64)
thus
Eyal Buks
456
16.6. Hamiltonian
V =
1
2
d3 r
d3 r VTP (r , r ) (r ) (r ) (r ) (r ) .
(16.65)
Combining all these results yields the total many-particle Hamiltonian

2
H=
d3 r (r ) (r )
2m

+ d3 r USP (r ) (r ) (r )

1
+
d3 r d3 r VTP (r , r ) (r ) (r ) (r ) (r ) .
2
(16.66)
Exercise 16.6.1. Show that the Heisenberg equation of motion for the field
operator (r ) is given by
d
i (r , t)
dt

2 2
=
+ USP (r ) (r , t)
2m

+ d3 r VTP (r , r ) (r , t) (r , t) (r , t) .
(16.67)
Note that in the absence of two-particle interaction the above equation for
the field operator (r , t) is identical to the single-particle Schrdinger equation for the single particle wavefunction (r ). Due to this similarity the
many-particle formalism of quantum mechanics is sometimes called second
quantization.
Solution 16.6.1. The Heisenberg equation of motion [see Eq. (4.37)] is given
by
i
d
= [H, ] .
dt
(16.68)
For general operators A, B and C the following holds

[AB, C] = A [B, C] [A, C] B
= A [B, C] [C, A] B .
(16.69)
Below we employ this relation for evaluating commutation relations. For

Fermions the upper sign (anti-commutation) is chosen, whereas for Bosons
the lower one is chosen (commutation). With the help of Eqs. (16.39), (16.40)
and (16.41) one finds (for both Bosons and for Fermions) that
Eyal Buks
457

,
2
d3 r (r ) 2 (r ) , (r )
2m

2
=
d3 r (r r ) 2 (r )
2m
2 2
=
(r ) ,
2m
[T, (r )] =
(16.70)
and
,
d3 r USP (r ) (r ) (r ) , (r )

= d3 r USP (r ) (r r ) (r )
[U, (r )] =
= USP (r ) (r ) .
(16.71)
(16.72)
Similarly
[V, (r )]

,
1
3
=
d r
d3 r VTP (r , r ) (r ) (r ) (r ) (r ) , (r )
2

,
1
=
d3 r d3 r VTP (r , r ) (r ) (r ) , (r ) (r ) (r )
2

1
3
d3 r VTP (r , r ) (r ) (r r ) (r ) (r )
=
d r
2

1
d3 r d3 r VTP (r , r ) (r r ) (r ) (r ) (r )
2

= d3 r VTP (r , r ) (r ) (r ) (r )
(16.73)
where in the last step it was assumed that VTP (r , r ) = VTP (r , r ). Combining these results lead to Eq. (16.67).
16.7 Momentum Representation

In the momentum representation the Hamiltonian is constructed using a
single-particle basis made of momentum eigenvectors |p . The wavefunc
tions of these single-particle states are proportional to eik r [see Eq. (3.75)],
where
k =
p
.
Eyal Buks
(16.74)
458
16.7. Momentum Representation

These wavefunctions can be normalized when the volume of the system is
taken to be finite. For simplicity, consider the case where the particles are
confined within a volume V = L3 having a cubic shape. The normalized wave
functions are taken to be given by

1
r |k = k (r ) = eik r ,
V
(16.75)
where |k labels a momentum eigenvector having an eigenvalue k . The

requirement that the wavefunctions k (r ) satisfy periodic boundary conditions, i.e. k (r ) = k (r + L
x) = k (r + L
y) = k (r + L
z), yields a
discrete set of allowed values of the wave vector k
k=
2
(nx , ny , nz ) ,
L
(16.76)
where nx , ny and nz are all integers. The orthonormality condition reads

1
d3 r k (r ) k (r ) =
d3 r ei(k k )r
V V
V
= k ,k .
(16.77)
In the momentum representation the many-particle kinetic energy T is
given by [see Eq. (16.59)]
1 2
T =
k | pSP |k ak ak
2m
k ,k
2 2
=
k ak ak ,
2m
k
the many-particle potential energy U is given by [see Eq. (16.63)]

U=
Uk k ak ak ,
(16.78)
(16.79)
k ,k
where
Uk k = k | USP (r ) |k
1
=
d3 r USP (r ) ei(k k )r ,
V V
(16.80)
and the many-particle interaction operator is given by [see Eq. (16.56)]

V =
1
2
Eyal Buks
k ,k ,k ,k
k , k | VTP |k , k ak ak ak ak ,
(16.81)
459

where
1
k , k | VTP |k , k = 2
V
d r
d3 r VTP (r , r ) ei(k
k )r i(k k )r
(16.82)
The assumption that VTP (r , r ) is a function of the relative coordinate r =

r r only, together with the coordinates transformation
r + r
,
2
r = r r ,
(16.83)
r0 =
(16.84)
yields (note that r = r0 + r/2 and r = r0 r/2)
k , k | VTP |k , k

i(k k k +k )r
1
2
d3 r0 ei(k k +k k )r0
d3 r VTP (r0 + r/2, r0 r/2) e
= 2
V V
V

i(k k k +k )r
1
2
= k +k ,k +k
d3 r vTP (r) e
,
V V
(16.85)
where
vTP (r) = VTP (r0 + r/2, r0 r/2) .
(16.86)
Thus the only allowed processes for this case are those for which the total
momentum is conserved, i.e. k + k = k + k . Using the notation
q = k k = k k ,
(16.87)
one can express V as

V =
1
vq ak ak ak q ak +q ,
2
(16.88)
k ,k ,q
where
1
vq =
V
d3 r vTP (r) e
iqr
2
(16.89)
16.8 Spin
In addition to spatial (orbital) degrees of freedom, the particles may have
spin. We demonstrate below the inclusion of spin for the case of momentum
representation. The basis for single-particle states is taken to be {|k , }k , ,
Eyal Buks
460
16.9. The Electron Gas

where the quantum number indicates the spin state. The single-particle
orthonormality condition reads
k , |k , = k ,k , .
(16.90)
The commutation (for Bosons) and anti-commutation (for Fermions) relations [see Eqs. (16.6), (16.7), (16.8) and (16.9)] become
3
4
=0,
(16.91)
[ak , , ak , ] = ak , , ak ,
3
4
ak , , ak ,
= k ,k , ,
(16.92)
For the example above, the Hamiltonian becomes

2 2
H=
k ak , ak , +
Uk k ak , ak ,
2m
k ,
1
+
2
k ,k ,
k ,k ,q, ,
vq ak , ak , ak q, ak +q, .
(16.93)
16.9 The Electron Gas

Consider a free (i.e. noninteracting) gas of N 1 electrons occupying volume
V. The Hamiltonian is given by [see Eq. (16.93)]
H=
2 2
k ak , ak , .
2m
(16.94)
k ,
In the momentum representation the single particle state |k , has a wavefunction given by [see Eq. (16.75)]

1
r |k , = eik r ,
V
(16.95)
and thus the quantized field operator (r ) is given by [see Eq. (16.37)]
1 ik r
e
ak , .
(r ) =
V k
(16.96)
The single particle state |k , has energy given by

k =
2 k2
,
2m
(16.97)
where m is the electron mass [see Eq. (16.94)].

Eyal Buks
461

The allowed values of k are determined by boundary conditions. Consider
for simplicity the case where the gas is confined in a cube (having edge length
of V 1/3 ). Imposing periodic boundary conditions on the wavefunction of the
single particle states |k , leads to the requirement (16.76). Thus, the density
of states per spin in k space is V/8 3 .
In the ground state |0 all single particle states for which |k | kF are
singly occupied, whereas all single particle states for which |k | > kF remain
empty, i.e.
#
ak , |0 .
(16.98)
|0 =
|k |kF ,
The Fermi wave vector is chosen such that the number of single particle states
for which |k | kF is N. Since the density of states per spin in k space is
V/83 one finds that
2
V 4 3
k = N ,
83 3 F
(16.99)
3 2 N
.
V
(16.100)
2 kF2
.
2m
(16.101)
thus
kF3 =
The Fermi energy F is the corresponding energy

F =
The density of states D () per spin and per unit volume is given by
D () =
1
( k ) .
V
(16.102)
where k is given by Eq. (16.97). By replacing the sum by an integral one

finds that

1
2 k2
D () =

V
2m
k
1 V
4
=
V 83
1
= 2
4

0
2m
2

2 k2
dk k2
2m
3/2
m
= 2 3 2m .
2
( )
(16.103)
Eyal Buks
462
16.10. Problems
The ground state energy is given by
E0 = 2V
F
5/2
d D ( ) =
23/2 m3/2 VF
52 3
(16.104)
or [see Eq. (16.100) and (16.101)]

E0 =
3N 2 kF2
.
5 2m
(16.105)
16.10 Problems
1. Find the many-particle interaction operator V for the case where the
two-particle potential is a constant VTP (r1 , r2 ) = V0 .
2. The same for the Coulomb interaction
VTP (r1 , r2 ) =
e2
.
|r1 r2 |
(16.106)
3. Show that
d
+ J = 0 ,
dt
(16.107)
where (r ) = (r ) (r ) is the the number density operator [see Eq.

(16.43)] and where the current density operator J is given by
J (r ) =

,
(r ) (r ) (r ) (r ) .
2im
(16.108)
4. Consider two identical Bosons having mass m in a one dimensional potential U (x) well given by
+
0 if 0 x L
.
(16.109)
U (x) =
else
The particles interact with each other via a two-particle interaction given
by VTP = V0 L (x1 x2 ), where V0 is a constant. Calculate the ground
state energy to lowest nonvanishing order in V0 .
5. By definition, an ideal gas is an ensemble of non-interacting identical
particles. The set of single particle eigenenergies is denoted by {i }. Calculate the average energy H and the average number of particles N in
thermal equilibrium as a function of the temperature T and the chemical
potential for the case of
a) Fermions.
b) Bosons.
Eyal Buks
463

6. Bogoliubov transformation - Consider the transformation
bk = uk ak + vk ak ,
(16.110)
where ak (ak ) is the annihilation (creation) operator corresponding to

the single particle state |k (|k), and where uk and vk are real coefficients. The state |Vb is defined by the condition
bk |Vb = 0 .
(16.111)
a) For the case of Fermions, under what conditions the operators bk

and bk can be considered as annihilation and creation operators?
Evaluate the expectation value Vb | ak ak |Vb .
b) The same for Bosons.
7. Find the eigenenergies of the Hamiltonian
3

4

k ak ak + ak + ak
,
(16.112)
H=
k
where ak and ak are Boson annihilation and creation operators corresponding to the single particle state |k , and where k and are real
coefficients.
8. Consider a system of identical spinless Bosons, whose Hamiltonian is
given by

k
k ak ak +
H=
ak ak + ak ak ,
(16.113)
2
where summation is over momentum single particle states having wave

vector k , and ak and ak are annihilation and creation operators, respectively. The coefficients k and k are assumed to be even functions
of k , i.e. k = k and k = k . Calculate the eigenenergies of H.
9. Find eigenvectors and eigenvalues of the quantized field operator (r)
for the case of Bosons. Evaluate the expectation values with respect to
the number operator N and with respect with the Hamiltonian of the
many body system (with one-particle and two-particle interactions).
10. Consider a neutral helium atom having 2 electrons and a nucleus having
charge +2e. Calculate the energy of the ground state. Assume that the
Coulomb interaction between the electrons can be considered as small
and calculate the energy correction due to this interaction to lowest nonvanishing order in perturbation theory. Ignore spinorbit coupling and
hyperfine interaction.
11. Consider the state

| = dr
dr F (r , r ) (r ) (r ) |0 ,
(16.114)
where (r ) is the Bosonic quantized field operator, |0 represents the
state where all occupation numbers are zero, and F (r , r ) is complex.
Eyal Buks
464
16.11. Solutions
a) Find a condition that the function F (r , r ) must satisfy in order to
ensure that the state | is normalized.
b) Consider the case where F (r , r ) can be expressed as F (r , r ) =
Af1 (r ) f2 (r ), where A is a normalization constant (which is chosen
such that | = 1) and where both functions f1 () and f2 () are
normalized according to

2
2
1 = dr |f1 (r )| = dr |f2 (r )| .
Evaluate the function
g (r ) = | (r ) | ,
where (r ) = (r ) (r ).
c) Calculate the total number of particles

N = dr g (r ) .
(16.115)
(16.116)
12. Consider a free (i.e. noninteracting) gas of N 1 electrons occupying

volume V. Calculate the correlation function
C (r r ) = 0 | (r ) (r ) |0 ,
(16.117)
where |0 is the ground state of the N electrons gas, (r) is the quantized field operator and stands for a spin state.
13. Calculate the ground state energy of electron gas containing N 1
electrons filling a volume V. Consider the Coulomb interaction between
electrons as weak and calculate the energy shift due to this interaction to
lowest non-vanishing order in perturbation theory. Assume that the volume V contains a uniform background of positive charge density +eN/V
(without the positive background the system is expected to be unstable
due to the repulsive nature of the Coulomb interaction).
16.11 Solutions
1. In general V is given by Eq. (16.88) where for this case
vq = V0 q,0 ,
(16.118)
thus
V0
a a ak ak
2 k k
k ,k

4
V0 3
=
a ak ak , ak + ak ak ak .
2 k
V =
k ,k
Eyal Buks
(16.119)
465

With the help of Eq. (16.69) one finds that [see also Eqs. (16.6), (16.7),
(16.8) and (16.9)]
4
4
3
3
ak ak , ak
= ak [ak , ak ] ak , ak ak
= k ,k ak ,
(16.120)
[for Fermions the upper sign (anti-commutation) is taken, whereas for

Bosons the lower one is taken (commutation)], thus
V = V0
N (N 1)
,
2
(16.121)
where N is the total number of particles operator. Note that N (N 1) /2

is the number of interacting pairs in the system.
2. For this case the Fourier transform f (q) of the function 1/ |r| is needed

1
(16.122)
= d3 q f (q) eiqr .
|r|
Applying the Laplace operator 2 and using the identity
2
1
= 4 (r)
|r|
(16.123)
yield
4 (r) =
d3 q f (q) |q|2 eiqr ,
(16.124)

dk eikx = 2 (x) ,
(16.125)
one finds that

f (q) =
1
,
22 q 2
where q = |q|, and therefore

1
1
1
= 2 d3 q 2 eiqr
|r|
2
q
(16.126)
(16.127)
With the help of this result one finds that V is given by [see Eqs. (16.88)
and (16.89)]
V =
1 4e2
a a ak q ak +q .
2V
q2 k k
(16.128)
k ,k ,q
Eyal Buks
466
16.11. Solutions
3. With the help of Eq. (16.67) and its Hermitian conjugate one finds that
d
d (r )
d (r )
=
(r ) + (r )
dt
dt
dt

1 2 ,
2
=
(r ) (r , t) 2 (r , t) (r ) ,
i 2m
(16.129)
where the assumptions USP

(r ) = USP (r ) and VTP
(r , r ) = VTP (r , r )
have been made, thus
d
+ J = 0 .
dt
(16.130)
Note the similarity between this result and the continuity equation that
is satisfied by a single-particle wavefunction [see Eq. (4.73)].
4. For the unperturbed case, i.e. when V0 = 0, the single-particle wavefunctions of the normalized eigenstates are given by
0
2
jx
j (x) =
sin
,
(16.131)
L
L
where j = 1, 2, , and the corresponding single-particle eigenenergies
are
j =
2 2 j 2
.
2mL2
(16.132)
For this case the ground state is the many-particle state |GS = |n1 = 2, n2 = 0, n3 = 0, ,
i.e. the state for which both particles are in the j = 1 single-particle state.
In perturbation theory to first order in V0 the energy of this state is given
by [see Eq. (9.32)]

E = 21 + GS| V |GS + O V02 ,
(16.133)
where the many-particle interaction operator V is given by Eq. (16.56).
The matrix element GS| V |GS is given by
1
1, 1| VTP |1, 1 GS| a1 a1 a1 a1 |GS
2

3
4
1
= 1, 1| VTP |1, 1 GS| a1 a1 a1 a1 , a1 a1 |GS
2
N1 (N1 1)
= 1, 1| VTP |1, 1 GS|
|GS
2
= 1, 1| VTP |1, 1 ,
(16.134)
GS| V |GS =
where the two-particle matrix element 1, 1| VTP |1, 1 is given by

Eyal Buks
467

1, 1| VTP |1, 1 =
dx1
= V0 L
0
L
dx2 1 (x1 ) 1 (x2 ) VTP (x1 , x2 ) 1 (x1 ) 1 (x2 )
dx1 41 (x1 )
3
= V0 ,
2
(16.135)
thus

2 2 3
(16.136)
V0 + O V02 .
2
mL
2
5. The grandcanonical partition function [see Eq. (8.450)] is evaluated by
summing over all many-particle states

Zgc = Tr eH+N

n1, n2 , , ni , | eH+N |n1, n2 , , ni , ,
=
E=
n1, n2 ,
(16.137)
where
H=
N=
i ai ai ,
(16.138)
ai ai ,
(16.139)

i
and = 1/kB T , thus one finds that

#
Zgc =
eni (i ) .
i
and
log Zgc =
(16.140)
ni
log
!

ni
ni (i )
"
(16.141)
a) In this case the summation over ni includes only two terms ni = 0

and ni = 1, thus

log Zgc =
log 1 + e(i ) .
(16.142)
i
The average energy

is found

using
Eq. (8.451)

log Zgc
log Zgc
+
H =
i e(i )
=
,
1 + e(i )
i
Eyal Buks
(16.143)
468
16.11. Solutions
whereas the average number of particles is found using Eq. (8.454)
N =
log Zgc e(i )

=
,
1 + e(i )
i
(16.144)
In terms of the Fermi-Dirac function fFD (), which is given by

fFD () =
1
,
exp [ ( )] + 1
these results can be rewritten as

i fFD (i ) ,
H =
(16.145)
(16.146)
and
N =
fFD (i ) .
(16.147)
b) In this case the summation over ni includes all integers ni =

0, 1, 2, , thus

1
log Zgc =
log
.
(16.148)
1 e(i )
i
The average energy is found using Eq. (8.451)

i fBE (i ) ,
H =
i
whereas the average number of particles is found using Eq. (8.454)

N =
fBE (i ) ,
(16.149)
i
where
fBE () =
1
exp [ ( )] 1
(16.150)
is the Bose-Einstein function .

6. The operators ak and ak satisfy [see Eqs. (16.6), (16.7), (16.8) and (16.9)]
3
4
[ak , ak ] = ak , ak
=0,
(16.151)
3
4
ak , ak
= k ,k .
(16.152)
Similarly, The operators bk and bk can be considered as annihilation and

creation operators provided that they satisfy
Eyal Buks
469

3
4
[bk , bk ] = bk , bk
=0,
3
4
bk , bk
= k ,k .
(16.153)
(16.154)
Using the definition (16.110) together with Eqs. (16.6) and (16.8) these
conditions become
3
4
3
4
vk uk ak , ak + uk vk ak , ak
=0,
(16.155)
3
4
3
4
vk uk ak , ak + uk vk ak , ak
=0,
(16.156)
3
4
3
4
uk uk ak , ak + vk vk ak , ak
= k ,k .
(16.157)
Note that by inverting the transformation between the operators ak , ak ,

ak and ak and the operators bk , bk , bk and bk , which can be expressed
in matrix form as [see Eq. (16.110)]
bk
ak
uk 0 0 v k
bk 0 uk vk 0 ak
=
(16.158)
bk 0 vk uk 0 ak ,
vk 0 0 uk
bk
ak
one finds that
ak
ak
1
=
ak uk uk vk vk
ak
uk
0
0
vk
0
0
uk vk
vk uk
0
0
bk
vk
0
bk
.

bk
0
uk
bk
(16.159)
This result together with Eq. (16.111) imply that the expectation value
Vb | ak ak |Vb is given by

2
vk
Vb | ak ak |Vb =
Vb | bk bk |Vb , (16.160)
uk uk vk vk
thus for both Bosons and Fermions [see Eq. (16.154)]

2
vk
Vb | ak ak |Vb =
.
uk uk vk vk
(16.161)
a) For the case of Fermions one finds using Eq. (16.9) that the conditions (16.155), (16.156) and (16.157) become (recall that [A, B]+ =
[B, A]+ )
(vk uk + uk vk ) k ,k = 0 ,
(16.162)
(16.163)
(vk uk + uk vk ) k ,k = 0 ,
(uk uk + vk vk ) k ,k = k ,k ,
(16.164)
Eyal Buks
470
16.11. Solutions
thus
vk uk + uk vk = 0 ,
(16.165)
2
2
(16.166)
uk + vk = 1 .
These conditions are guarantied to be satisfied provided uk and vk
are expressed using a single real parameter k as
uk = cos k , vk = sin k ,
(16.167)
uk = cos k , vk = sin k .
(16.168)
For this case Eq. (16.159) becomes
ak
bk
cos k
0
0 sin k
ak 0
bk
cos k sin k
0
=
, (16.169)
ak 0 sin k cos k
bk
0
sin k
0
0
cos k
ak
bk
and Eq. (16.161) becomes
Vb | ak ak |Vb = sin2 k .
(16.170)
b) For the case of Bosons one finds using Eq. (16.7) that the conditions (16.155), (16.156) and (16.157) become (recall that [A, B] =
[B, A] )
vk uk + uk vk = 0 ,
(16.171)
2
2
uk vk = 1 .
(16.172)
These conditions are guarantied to be satisfied provided that uk and
vk are expressed using a single real parameter k as
uk = cosh k , vk = sinh k ,
(16.173)
uk = cosh k , vk = sinh k .
(16.174)
For this case Eq. (16.161) thus becomes
Vb | ak ak |Vb = sinh2 k .
(16.175)
7. Consider the unitary transformation [see for comparison Eq. (9.49)]

k = eLk Hk eLk ,
H
(16.176)
where
3

4
Hk = k ak ak + ak + ak
,
(16.177)

Lk = ak ak .
(16.178)
and where

eL AeL = A + [L, A] +
Eyal Buks
1
1
[L, [L, A]] + [L, [L, [L, A]]] + , (16.179)
2!
3!
471

and the identities
4
3
ak , ak ak = ak ,
3
4
ak , ak ak = ak ,
3
4
ak ak , ak + ak = 2 ,
(16.180)
(16.181)
(16.182)
one finds that

k = Hk k ak + a 2k 2 + k 2
H
k

2
= k ak ak .
(16.183)
Thus, the unitary transformation

= U HU ,
H
(16.184)
where
!
U = exp

k
Lk
"
which yields

=
H
k ak ak 2 ,
(16.185)
(16.186)
can be employed for diagonalization of H. Let |

n be a number state,
which satisfy
ak ak |
n = nk |
n ,
(16.187)
where nk is the number of particles in single-particle state |k . The

following holds
HU |
n = UU HU |
n
|
= UH
n

=
k nk 2 U |
n ,
k
(16.188)
thus the eigenvectors of H are the vectors U |

n and the corresponding
eigenenergies are given by

k nk 2 .
(16.189)
En =
k
Eyal Buks
472
16.11. Solutions
8. By employing the Bogoliubov transformation [see Eqs. (16.159), (16.173)
and (16.174)]
ak
ak
1
=
ak uk uk vk vk
ak
bk
uk
0
0 vk
0

uk vk 0
bk ,
0 vk uk
0 bk
vk 0
0
uk
bk
(16.190)
where
uk = uk = cosh k ,
vk = vk = sinh k ,
(16.191)
(16.192)
the identities
sinh (2 k ) = 2 sinh k cosh k ,
cosh (2 k ) = sinh2 k + cosh2 k ,
cosh (2k ) + 1
cosh2 k =
,
2
cosh (2k ) 1
sinh2 k =
,
2
3
4
and the commutation relation bk , bk = 1, one finds that

H=
(k cosh (2k ) k sinh (2k )) bk bk
(16.193)
(16.194)
(16.195)
(16.196)
cosh (2k ) 1 k sinh (2k )
2
2
k

cosh (2 k ) k sinh (2k )
k
+
bk bk + bk bk .
2
(16.197)
To eliminate the mixed terms bk bk and bk bk the factors k are

chosen such that
k cosh (2k ) k sinh (2k ) = 0 .
(16.198)
Using the identities

cosh2 (2k ) =
sinh2 (2k ) =
1
,
1 tanh2 (2k )
tanh2 (2 k )
,
1 tanh2 (2k )
(16.199)
(16.200)
one finds that for that choice H becomes diagonal

Eyal Buks
473

H=
k
2
k
k
2

bk bk
k
k
2
1 ,
(16.201)
and thus, the eigenenergies are given by

:
2

k
En =
k 1
nk
k
k
:
2
k
1 k
+
1 ,
2
k
(16.202)
where the nonnegative integer nk is the number of so-called quasi particles in state k .
9. Consider the state | (r ), which is defined by
| (r ) = D(r ) |0 ,
(16.203)
where (r ) C and where the operator D(r ) is given by [see for comparison Eq. (5.36)]
/
D(r ) = e
dr ((r ) (r ) (r ) (r ))
(16.204)
For general operators A and B the following holds [see Eq. (2.175)]
1
eA+B = eA eB e 2 [A,B] = eB eA e 2 [A,B] ,
(16.205)
provided that
[A, [A, B]] = [B, [A, B]] = 0 .
Moreover, with the help of Eq. (16.39) one finds that

dr (r ) (r ) , dr (r ) (r )

,
= dr dr (r ) (r ) (r ) , (r )

2
= dr | (r )| ,
(16.206)
(16.207)
thus [see for comparison Eq. (5.39)]

Eyal Buks
474
16.11. Solutions
/
dr (r ) (r )
D(r ) = e
= e
dr (r ) (r )
dr (r ) (r ) 12
dr (r ) (r )
e2
dr |(r )|
2
dr |(r )|
.
(16.208)
Using the last result (16.208) it is easy to show that D(r ) is unitary
D(r
) D(r ) = D(r ) D(r ) = 1 ,
(16.209)
and thus | (r ) is normalized. With the help of Eq. (16.208) together

with the relation (r) |0 = 0 one finds that
1
| (r ) = e 2
dr |(r )|
dr (r ) (r )
|0 .
(16.210)
To show that | (r ) is an eigenvector

,
/ of the quantized
field -operator (r)
the commutation relation exp dr (r ) (r ) , (r) is evaluated
below. For general operators A and B and for a smooth function f (A)
the following holds
[f (A) , B] =
df
[A, B] ,
dA
(16.211)
provided that [[A, B] , A]/ = 0 [see Eq. (2.174)]. Using this general result
[with f (A) = eA , A = dr (r ) (r ) and B = (r)] together with
Eq. (16.39) yields

3 /
4
/
e dr (r ) (r ) , (r) = e dr (r ) (r )
dr (r ) (r ) , (r)

/
dr (r ) (r )
= e
dr (r ) (r r )
/
= e
dr (r ) (r )
(r) ,
(16.212)
The last result together with the relation (r) |0 = 0 can be used to
show that the state | (r ) is an eigenvector of (r) with eigenvalue (r)
(r) | (r )
/
1
= (r) e 2 dr |(r )| e dr (r ) (r ) |0
/
3
/
/
1
= e 2 dr |(r )| e dr (r ) (r ) (r) + (r) , e
12
= (r) e
dr |(r
)| e
dr (r )
(r ) |0 ,
dr (r ) (r )
4
|0
(16.213)
that is
(r) | (r ) = (r) | (r ) .
Eyal Buks
(16.214)
475

The expectation value with respect to the number operator N [see Eqs.
(16.42) and (16.43)] is given by
(r )| N | (r ) = d3 r (r )| (r ) (r ) | (r )

= d3 r | (r)|2 ,
whereas the expectation value with respect to the Hamiltonian H [see
Eq. (16.66)] is given by

2
(r )| H | (r ) =
d3 r (r) (r)
2m

+ d3 r USP (r ) | (r)|2

1
2
2
3
+
d3 r VTP (r , r ) | (r )| | (r )| .
d r
2
(16.215)
10. First consider the unperturbed problem where the Coulomb interaction
between the electrons is disregarded. The single-electron Hamiltonian is
obtained by substituting the factor e2 in the Hamiltonian of a hydrogen atom by Ze2 , where for helium Z = 2. The single electron energy
eigenstates |n, l, m, are chosen to be also eigenvectors of the single
electron angular momentum operators Lz and L2 [see Eqs. (7.42), (7.43)
and (7.44)]. While n, l and m are orbital quantum numbers, labels the
spin state. The single electron eigenenergies are given by [see Eq. (7.84)]
En =
Z 2 EI
,
n2
(16.216)

EI =
me e4
,
22
(16.217)
and where me is the electrons mass. The position wavefunction n,l,m (r)
of a single-electron energy eigenstates having orbital quantum numbers
n, l and m is given by [see Eq. (7.92)]
(Z)
nlm (r, , ) = Rnl (r) Ylm (, ) ,
(16.218)
(Z)
where the radial wavefunction Rnl (r) is obtained by substituting e2

by Ze2 in the radial wave function Rnl (r) of hydrogen [see Eqs. (7.115),
(7.116) and (7.117)]. The ground state | (when the Coulomb interaction
between the electrons is disregarded) is given by [see Eq. (16.21)]
Eyal Buks
476
16.11. Solutions
| = an=1,l=0,m=0,= an=1,l=0,m=0,=+ |0 ,
(16.219)
where an,l,m, are creation operators and where |0 represents the state
where all occupation numbers are zero. The energy of the unperturbed
ground state is 2 22 EI = 8EI [see Eq. (16.216)]. The Coulomb interaction between the electrons is described by the two-particle operator
[see Eq. (16.106)]
VTP (r1 , r2 ) =
e2
.
|r1 r2 |
(16.220)
In the many-particle case the two-electron interaction is represented by

the operator V , which is given by Eq. (16.56). To first order in perturbation theory the energy of the ground state [see Eq. (9.32)] is given by
8EI + | V | , where [see Eq. (7.115)]

2
e2
| V | = dr1 dr2
1,0,0 (r1 ) 1,0,0 (r2 )
|r1 r2 |

R(Z) (r ) R(Z) (r )
2
e2
1 10
2
= dr1 dr2
10
4
4
|r1 r2 |
6
4(r1 +r2 )

e2 a20
e a0
=
dr1 dr2
2
|r1 r2 |
= EI ,
(16.221)
the dimensionless factor is given by
27 1
= 2 5
a0
dr1
4(r1 +r2 )
e a0
dr2
,
|r1 r2 |
(16.222)
and where
a0 =
2
me e2
(16.223)
is the Bohrs radius [see Eq. (7.64)]. The integration over r2 is performed
in spherical coordinated, where the z axis is chosen in the direction of
the vector r1
Eyal Buks
477

27 1
= 2 5
a0
dr1 e
dr2 r22 e
2
d (cos 2 )

d2
r12 + r22 2r1 r2 cos 2
1
0

4r2
a0
= 4
4r1
a0
2 1
4
2 a50
5
= ,
2

0
4r
a1
0
dr1 r12 e
1
r1
1
r1 +r2 |r1 r2 |
r1 r2
r1
4r
a2
0
dr2 r22 e
4r
a2
0
dr2 r2 e
r1
(16.224)
thus the ground state energy is 8EI + | V | = (11/2) EI . Note
that the fact the energy correction | V | is comparable with the unperturbed value of 8EI suggests that the accuracy of the first order
perturbation approximation is relatively poor.
11. With the help of the commutation relations (16.39), (16.40) and (16.41)
one finds that

| = dr
dr
dr
dr F (r , r ) F (r , r )
0| (r ) (r ) (r ) (r ) |0

= dr dr dr F (r , r ) F (r , r ) 0| (r ) (r ) |0

+ dr
dr
dr
dr F (r , r ) F (r , r ) 0| (r ) (r ) (r ) (r ) |0

2
= dr dr |F (r , r )| + F (r , r ) F (r , r ) .
(16.225)
a) The condition is

2
1 = dr dr |F (r , r )| + F (r , r ) F (r , r ) . (16.226)
b) The normalization
for this case reads

condition

2
2
1 = |A|
dr dr |f1 (r ) f2 (r )| + f1 (r ) f2 (r ) f1 (r ) f2 (r )

= |A|2 1 + | 12 |2 ,
(16.227)
where
12 =
dr f1 (r ) f2 (r ) .
(16.228)
The following holds [see Eqs. (16.39), (16.40) and (16.41)]

Eyal Buks
478
16.11. Solutions
g (r ) =
1
1 + | 12 |2
dr
dr
dr
dr f1 (r ) f2 (r ) f1 (r ) f2 (r )
0| (r ) (r ) (r ) (r ) (r ) (r ) |0
=
|f1 (r )|2 + |f2 (r )|2 + 12 f1 (r ) f2 (r ) + 12 f1 (r ) f2 (r )

1 + | 12 |2
c) The number of particles is given by

N = dr g (r ) = 2 .
(16.229)
(16.230)
12. The correlation function C (r r ) is given by

C (r r ) =
1 i(k r k r )
e
0 | ak , ak , |0 ,
V
(16.231)
k ,k
where |0 is the ground state of the free electron gas [see Eq. (16.98)],
thus
1 ik (r r )
e
C (r r ) =
.
(16.232)
V
|k |kF
For N 1 the summation can be approximately substituted by integration over the Fermi sphere having radius kF [see Eq. (16.100)]. In
spherical coordinates in which the z axis is taken to be in the direction
of the vector r r one has

V 2 kF 2 1
dk k
d (cos ) eik cos |r r | , (16.233)
C (r r ) = 3
8 V 0
1
thus
C (r r ) =
1 sin (kF |r r |) kF |r r | cos (kF |r r |)

.
22
|r r |3
(16.234)
With the help of Eq. (16.100) the result can be expressed as

C (r r ) =
3N sin (kF |r r |) kF |r r | cos (kF |r r |)

.
2V
(kF |r r |)3
(16.235)
13. First consider the unperturbed case, where the electron-electron Coulomb
interaction is disregarded. The ground state
Eyal Buks
479

|0 =
|k |kF ,
ak , |0
(16.236)

is given by Eq. (16.98), and its energy E0 = (3N/5) 2 kF2 /2m by Eq.
(16.105), where kF is the Fermi wave vector. To first order in perturbation
(1)
theory the energy of the ground state becomes EGS = E0 + E, where
the energy shift E due to electron-electron Coulomb interaction is given
by [see Eqs. (9.32), (16.65), (16.93) and (16.106)]

1
3
E =
d r
d3 r VTP (r , r ) 0 | (r ) (r ) (r ) (r ) |0 ,
2
,
(16.237)
where
VTP (r , r ) =
e2
.
|r r |
(16.238)
With the help of the expansion (16.96) and the commutation relations
(16.91) and (16.92) one finds that
0 | (r ) (r ) (r ) (r ) |0

1
ei(k k )r ei(k k )r 0 | ak , ak , ak , ak , |0
= 2
V
k ,k ,k ,k
1
= 2 ,
V
+
1
V2
ei(k
k )r i(k k )r
k ,k ,k
k ,k ,k ,k
ei(k
k )r i(k k )r
0 | ak , ak , |0
0 | ak , ak , ak , ak , |0 .
(16.239)
The only nonvanishing terms in the second line are those for which either
k = k and k = k or k = k and k = k . For the second case
the two possibilities = and = are separately considered
Eyal Buks
480
16.11. Solutions
0 | (r ) (r ) (r ) (r ) |0

1
= 2 ,
eik (r r )
eik (r r )
V
|k |kF
1
+ 2
| Nk , Nk , |0
V 0
k ,k
1
ei(k k )(r r ) 0 | ak , ak , ak , ak , |0
+ 2 ,
V

k ,k
1
+ 2 (1 , )
ei(k k )(r r ) 0 | ak , ak , ak , ak , |0 ,
V
k ,k
(16.240)
thus
0 | (r ) (r ) (r ) (r ) |0

1
= 2 ,
eik (r r )
eik (r r )
V
+
+
1
V2
|k |kF
|k |,|k |kF

1
,
1
2
V
|k |kF
ik r r ik r r
1
)
(
)
,
e (
e
2
V
k
|k |kF
1
ik (r r )
2 ,
|k |kF

1
+ 2 (1 , )
1.
V
|k |kF
(16.241)
For N 1 the single summation terms are negligibly small

0 | (r ) (r ) (r ) (r ) |0
1
1
ik (r r )
= 2
1 2 ,
,
V
V
|k |kF
|k |,|k |kF
(16.242)
or [see Eqs. (16.232) and (16.235)]
Eyal Buks
481

0 | (r ) (r ) (r ) (r ) |0
2
N
2
=
, |C (r r )| ,
2V
(16.243)
where
C (r r ) =
3N sin (kF |r r |) kF |r r | cos (kF |r r |)

.
2V
(kF |r r |)3
(16.244)
With the help of the above result one obtains

2

1 N
E =
d3 r d3 r VTP (r , r )
2 V

2
3
d r
d3 r VTP (r , r ) |C (r r )| .
(16.245)
The first term of (16.245) represents the electrostatic energy due to

electron-electron interaction. However, as is argued below, in the presence of positive charge density +eN/V this term should be disregarded.
This can be seen by noticing that the self electrostatic energy of the
positive background is identical to the first term of (16.245), whereas
the electrostatic energy due to interaction
/
/ between the electrons and the
positive background is (N/V)2 d3 r d3 r VTP (r , r ), thus these two
contributions exactly cancels the first term of (16.245). The second term,
which is commonly called the exchange energy, can be evaluated using
(1)
Eq. (16.235). Thus the energy of the ground state EGS (to first order in
perturbation theory) is given by
(1)
3N 2 kF2
5 2m

2

e2 3N
(sin (kF |r |) kF |r | cos (kF |r |))2
V d3 r
|r | 2V
(kF |r |)6
2
9Ve2 N
3N 2 kF2
(sin x x cos x)2
V
=
dx
,
2
5 2m
kF
x5
0

EGS =
1/4
(16.246)
or [see Eq. (16.100)]

2 2

3 kF 3kF e2
(1)
EGS = N
.
5 2m
4
Eyal Buks
(16.247)
482
17. Superconductivity
In this chapter two models are discussed, the Londons model, in which a
macroscopic wavefunction is introduced to describe the state of a superconductor, and the model by Bardeen, Cooper and Schrieffer (BCS), which provides an insight on the underlying microscopic mechanisms that are responsible for superconductivity.
17.1 Macroscopic Wavefunction

In this chapter the Londons equations are derived from the assumption that
the state of a superconductor can be describe using a macroscopic wavefunction.
17.1.1 Single Particle in Electromagnetic Field
Consider a single particle having charge q and mass m in electromagnetic
field characterized by the scalar potential and the vector potential A. The
electric field E and the magnetic field B are given by (in Gaussian units) [see
Eqs. (1.41) and (1.42)]
E =
1 A
,
c t
(17.1)
and
B=A,
(17.2)
8
where c = 2.99 10 m s is the speed of light in vacuum. Let r = (x, y, z)

be the position vector of the particle in Cartesian coordinates. The variable
vector canonically conjugate to the position vector r is given by [see Eq.
(1.61)]
q
p = mr+ A .
(17.3)
c
The classical equation of motion is given by [see Eq. (1.60)]

1
m
r = q E + r B .
(17.4)
c
Chapter 17. Superconductivity

The Schrdinger Equation. The Hamiltonian of the system is given by
[see Eq. (1.62)]

2
p qc A
+ q .
(17.5)
H=
2m
The Schrdinger equation for the wavefunction (r , t ) is given by [see Eq.
(4.194)]
d
1
q 2
i
=
i A + q .
(17.6)
dt
2m
c
The continuity Equation. The continuity equation expresses the probability conservation law [see Eq. (4.73)]
d
+J=0 ,
dt
where
=
(17.7)
(17.8)
is the probability distribution function and

J=
q

Im ( )
A
m
mc
(17.9)
is the current density [see Eq. (4.200)]. For a wavefunction having the form
= 1/2 ei ,
(17.10)
where is real, one has [see Eq. (6.361)]

q

A .
J=
m
c
(17.11)
Gauge Invariance. Consider the following gauge transformation [see Eqs.

(12.49) and (12.50)]
= A + ,
AA
(17.12)
=,
(17.13)
where = (r) is an arbitrary smooth and continuous function of r, which

is assumed to be time independent. This transformation leaves E and B
unchanged [see Eqs. (12.1) and (12.2)], however, the wavefunction is transformed according to the following rule. Given that the wavefunction (r , t )
solves the Schrdinger equation with vector A and scalar potentials, the
= A+ and scalar
transformed Schrdinger equation with vector A
=
(r , t ), which is given
potentials is solved by the transformed wavefunction
by [see Eq. (12.53)]

(r , t ) = exp iq (r ) (r , t ) .
(17.14)
c
Eyal Buks
484
17.1. Macroscopic Wavefunction

17.1.2 Drude Model
Consider a conductor containing charge carriers having charge q and mass m
in electromagnetic field. The density of charge carriers (i.e. number per unit
volume) is n. Scattering is taken into account in the Drude model by adding
a damping term to the classical equation of motion (17.4)

1
1
m
r+ r = q E + r B ,
(17.15)
tr
c
where tr is the so-called scattering time. For simplicity the applied magnetic
field is assumed to vanish, i.e. B = 0. In terms of the current density vector
J, which is related to the the velocity vector v = r by the relation
v=
J
,
qn
(17.16)
Eq. (17.15) yields

m J 1
J
=E.
+
q 2 n t tr
(17.17)
Maxwells Equations. The macroscopic Maxwells equations (in Gaussian

units) for the electric field E, electric displacement D, magnetic induction
B and magnetic field H in the presence of external charge density ext and
external current density Jext are given by
1 D
4
Jext +
,
c
c t
1 B
E =
,
c t
D = 4ext ,
B = 0 .
H =
(17.18)
(17.19)
(17.20)
(17.21)
For an isotropic and linear medium the following relations hold

D = E+4P ,
D = E ,
P = e E ,
B = H + 4M ,
B = H ,
M = m H ,
(17.22)
(17.23)
(17.24)
(17.25)
(17.26)
(17.27)
where P is the electric polarization, = 1 + 4e is the permittivity (dielectric constant of the medium), e is the electric susceptibility, M is the
magnetization, = 1 + 4m is the permeability and m is the magnetic
susceptibility.
Eyal Buks
485

Fourier Expansion. In general, a scalar function f (r, t) can be Fourier
expanded as

f (r, t) = dq d ei(qrt) f (q, ) ,
(17.28)
and a vector function F (r, t) can be Fourier expanded as

F (r, t) = dq d ei(qrt) F (q, ) .
(17.29)
A vector function F (r, t) can be decomposed into longitudinal and transverse parts with respect to the wave vector q according to
F = FL + FT ,
(17.30)
nq F) n
q , the transverse one
where the longitudinal part is given by FL = (
is given by FT = (
nq F) n
q , and where n
q = q/ |q| is a unit vector in
the direction of q. The following holds F = FL and F = FT .
Recall that for a general scalar and a vector A the following holds
(A) = A + A ,
(A) = A A ,
thus
F =
dq
d ei(qrt) iq FL (q, ) ,
(17.31)
and
F=
dq
d ei(qrt) iq FT (q, ) .
(17.32)
With the help of the above relations the Maxwells equations (17.18),
(17.19), (17.20) and (17.21) can be Fourier transformed into
4
i
Jext (q, ) D (q, ) ,
c
c
q ET (q, ) = B (q, ) ,
c
iq DL (q, ) = 4ext (q, ) ,
q BL (q, ) = 0 .
iq HT (q, ) =
(17.33)
(17.34)
(17.35)
(17.36)
Dielectric Constant. While the external charge density ext is related to

D by the relation D = 4ext [see Eq. (17.20)], the induced charge
density ind , which is defined as the change in charge density with respect to
the unperturbed case, is related to the electric polarization by the relation
P =ind , and the total charge density ind +ext is related to the electric
field E by the relation
Eyal Buks
486

E = 4 (ind + ext ) ,
(17.37)
which can be Fourier transformed into

iq EL (q, ) = 4 (ind (q, ) + ext (q, )) .
(17.38)
The longitudinal dielectric function (q, ), which is defined by [compare

with Eq. (17.23)]
(q, )
|DL (q, )|
,
|EL (q, )|
(17.39)
is thus given by [see Eqs. (17.35) and (17.38)]

(q, ) =
ext (q, )
.
ext (q, ) + ind (q, )
(17.40)
For a general longitudinal field FL the following holds FL = 0, and thus

both DL and EL can be expressed in terms of scalar potentials as
DL (r, t) = ext (r, t) ,
EL (r, t) = (r, t) .
(17.41)
(17.42)
Consequently, one finds that the longitudinal dielectric function (q, ) can
alternatively be expressed as [see Eq. (17.39)]
(q, ) =
ext (q, )
.
(q, )
(17.43)
For the case where the induced charge is carried by electrons having mass
me , charge qe , density ne and scattering time tr,e Eq. (17.17) becomes

1
me
Je
+
Je = E ,
(17.44)
qe2 ne t tr,e
where the electron current density Je is related to the induced electron charge
density ind = e by the continuity equation (17.7)
de
+ Je = 0 .
dt
(17.45)
Applying to Eq. (17.44) and using Eqs. (17.45) and (17.37) lead to
d2 e
1 de
+
= 2p,e (e + ext ) ,
2
dt
tr,e dt
(17.46)
where p,e , which is given by

2p,e =
Eyal Buks
4qe2 ne
,
me
(17.47)
487

is the so-called electron plasma frequency. Consider the
case where ext is
taken to oscillate according to ext = Re ext,0 eit , where the angular
frequency is a real constant.
Seeking a solution to Eq. (17.46) having the

form e = Re e,0 eit yields
e,0 =
2p,e
,
1
2 + i tr,e
2p,e ext,0
(17.48)
thus the frequency dependent dielectric constant e () is given by [see Eq.

(17.40)]
e () = 1
2p,e i tr,e
.
2 i tr,e 1
Alternatively, in terms of the skin depth e , which is given by

:
c
2
e =
,
p,e tr,e
(17.49)
(17.50)
the dielectric constant e () can be expressed as

e () = 1 +
2ic2
1
.
2 2 1 i
e
tr,e
(17.51)
Complex Conductivity. Consider

case
the
where the applied electric field
it
E is taken to be given byE = Re E
e
. Seeking a solution of Eq. (17.44)
0

having the form Je = Re Je,0 eit yields
Je,0 = e () E0 ,
(17.52)
where
e () =
e,0
,
1 i tr,e
(17.53)
is the complex conductivity () and where

e,0 =
qe2 ne tr,e
.
me
(17.54)
In terms of the complex conductivity e () the dielectric constant e () can

be expressed as [see Eq. (17.49)]
e () = 1 + i
Eyal Buks
4 e ()
.
(17.55)
488

17.1.3 The Macroscopic Quantum Model
The macroscopic quantum model is based on the hypothesis that some properties of a superconducting media can be described by a single wavefunction
s (r , t ). It is assumed that the local density of superconducting charge carriers ns is related to the wavefunction by
2
ns = |s (r , t )| .
(17.56)
In the presence of an electromagnetic field the time evolution of s (r , t ) is

governed by the Schrdinger equation [see Eq. (17.6)]
i
d s
1
=
dt
2ms

2
q
i s A s + qs s .
c
(17.57)
where ms and qs are the mass and charge respectively of a superconducting

charge carrier. Furthermore, it is assumed that the current density carried
by a superconductor having a macroscopic wavefunction given by [see Eq.
(17.10)]
s (r , t ) =
is given by
Js =

ns (r , t )ei(r ,t ) ,
qs ns (r , t )
ms

q
s A .
c
(17.58)
(17.59)
Note that while J in Eq. (17.11) represents probability current density, Js in

a superconductor represents charge current density.
17.1.4 London Equations
London equations can be derived from the macroscopic quantum model by
assuming that the superconducting charge carriers density ns is constant.
2nd London Equation. By taking the curl of Eq. (17.59) and employing
Eq. (17.2) one obtains the second London equation, which reads
Js =
qs2 ns
B.
ms c
(17.60)
Taking the curl of Eq. (17.18), assuming = = 1 and employing Eqs.

(17.60), (17.19) and (17.21) together with the general vector identity
B = ( B) 2 B
(17.61)
lead to
Eyal Buks
489

2 B =
1
1 2B
2 B + c2 t2 ,
L
(17.62)
ms c2
4ns qs2
(17.63)
where
L =

is the London penetration depth in Gaussian units (L = ms /0 ns qs2 in
SI units). In terms of the superconducting plasma frequency p,s , which is
given by
2p,s =
4ns qs2
,
ms
(17.64)
the London penetration depth can be expressed as

L =
c
.
p,s
(17.65)
For time independent B the solution of Eq. (17.62) yields an exponential

decay of B with characteristic decay length given by the London penetration
depth L . Thus, except of a region having characteristic width L near the
surfaces the magnetic field inside a superconductor vanishes (even in the
presence of an externally applied magnetic filed). This expulsion of a magnetic
field from a superconductor, which is called the Meissner effect , represents the
perfect diamagnetism of superconductors. As can be seen from Eq. (17.18), in
the absence of time dependent electric field (i.e. when displacement currents
vanish) the expulsion of a magnetic field also implies that the supercurrent
density Js also vanishes deep inside a superconductor.
1st London Equation. When the superconducting density of charge carriers ns is assumed to be a constant Eq. (17.57) becomes [see Eq. (17.58)]

2
d
1
qs

(17.66)
=
A + qs ,
dt
2ms
c
or [see Eq. (17.59)]
d
m
= 2 s 2 J2s + qs .
dt
2qs ns
(17.67)
Applying to the above leads to

d
m
= 2 s 2 J2s + qs .
dt
2qs ns
(17.68)
Taking the time derivative of Eq. (17.59) and employing Eq. (17.1) together
with the last result yield the first London equation
Eyal Buks
490

ms
qs2 ns
Js
1
+ J2s
t 2qs ns
=E.
(17.69)
Typically in superconductors the electric field E on the right hand side

of Eq. (17.69) can be neglected in comparison with the term proportional to
J2s on the left
side of Eq. (17.69). The factor J2s can be estimated by
hand
2
the relation
J2s
2 |Js | /l0 , where l0 is a length scale that characterizes
the spacial variations of the current density Js . Moreover, the ratio |Js | /l0
can be estimated from the second London equation (17.60)
q 2 n
|Js |
s s |B| .
l0
ms c
(17.70)
Combining these results allows estimating the term proportional to J2s on

the left hand side of Eq. (17.69)
ms
2
qs ns
1
2qs ns
J2s
1 |vs | |B| ,
c
(17.71)

vs =
Js
qs ns
(17.72)
is the velocity of superconducting charge carriers. In view of the classical

equation of motion (17.4) the above estimate shows that the ratio between
|E| and the the term proportional to J2s in Eq. (17.69) represents the ratio
between electric and magnetic forces acting on the superconducting charges.
Typically in metals electric forces are strongly suppressed due to screening,
and consequently can be neglected in comparison with magnetic forces. Neglecting the E term in Eq. (17.69) leads to
Js
1
J2s = 0 .
+
t 2qs ns
(17.73)
Homogeneous solutions (i.e. position independent solutions) of the first

London equation (17.69) satisfy
ms Js
=E.
qs2 ns t
(17.74)
Comparing the above relation with Eq. (17.17) of the Drude model yields
agreement provided that the scattering time tr is assumed to be infinite.
From this point of view the first London equation represents the nullification
of resistance in superconductors. Equation (17.74) implies that the superconducting conductivity s () for the homogeneous case is given by [see Eq.
(17.53)]
s () = i
Eyal Buks
qs2 ns
.
ms
(17.75)
491

Flux Quantization. Consider a close curve C inside a superconductor. Integrating Eq. (17.59), which is given by

qs ns
qs
A ,
(17.76)
Js =
ms
c
along the curve yields
@

@
@
q n
q
dr Js = s s dr s
dr A .
ms
c C
C
C
(17.77)
The assumption that Ethe superconducting wavefunction s = ns ei is continuous implies that dr = 2n, where n is integer. The integral over
A can be calculated using Stokes theorem [see Eqs. (12.2) and (12.47)]
@
dr A = C ,
(17.78)
C
/
where C = ds B is the magnetic flux threaded through the area enclosed
by the closed path C. With these results Eq. (17.77) becomes

@
hq n
dr Js = s s n C ,
(17.79)
ms
s
C
where
s =
hc
qs
(17.80)
is the so called superconducting flux quantum (in Gaussian units).

As was shown above, the second London equation implies that the supercurrent density Js vanishes deep inside a superconductor. Consider a close
curve C inside a superconductor and assume that the distance between any
point on C and the nearest surface is much larger than the London penetration depth L . For such a curve the left hand side of Eq. (17.79) vanishes,
and consequently
C = ns ,
(17.81)
i.e. the magnetic flux is quantized in units of the superconducting flux quantum.
Two Fluid Model. In the limit of vanishing temperature only superconducting charge carriers are present in a superconductor. However, at finite
temperature also normally conducting charge carriers may be present. Consequently, the total complex conductivity () [see Eq. (17.52)] is taken to
be given by
() = n () + s () ,
Eyal Buks
(17.82)
492
17.2. The Josephson Effect

where the normal conductivity n () is assumed to be given by the Drude
formula (17.53)
n () =
qe2 nn tr,e
1
,
me
1 i tr,e
(17.83)
where nn is the density of normal charge carriers, and the super conductivity
s () is assumed to be given by Eq. (17.75)
s () = i
qs2 ns
.
ms
(17.84)
The electronic dielectric constant e () [see Eq. (17.55)] in the two fluid
model is thus given by
e () = 1 + i
4 n ()
4s ()
+i
,
(17.85)
or in terms of the skin depth [see Eq. (17.50)] and the London penetration
depth L [see Eq. (17.63)]
e () = 1 +
2i
1
1
,
1
i
tr,e
( e k)
(L k)2
2
(17.86)
where k = /c. Note that 1/ 2 nn [see Eq. (17.50)], whereas 1/2L ns

[see Eq. (17.63)]. Note also that the ratio between these characteristic length
scales is given by
:
ms nn q 2 tr
L
=
.
(17.87)
L
m ns qs2 2
17.2 The Josephson Effect

A Josephson junction is formed between two superconductors that are weakly
connected to each other. Electrons can flow between the superconductors by
crossing a barrier. In this chapter the Josephson relations are derived from
the model that was discussed in the previous chapter.
17.2.1 The First Josephson Relation
In terms of the superconducting flux quantum s [see Eq. (17.80)] Eq. (17.76)
can be rewritten as
Js =
qs ns
GI ,
ms
Eyal Buks
(17.88)
493

where
GI =
2
A.
s
(17.89)
The phase factor GI is commonly called the gauge invariant phase.

Consider an integral over Js (17.88) along a path going through a Josephson junction from point r1 on the interface between the first superconductor
and the barrier to point r2 on the interface between the second superconductor and the barrier. The path is assumed to be parallel to the direction of
the current density Js . The total current flowing through the junction I is
expected to be a function of the phase difference , which is obtained from
integrating GI

r2
2 r2
=
dr GI = (r2 ) (r1 )
dr A .
(17.90)
s r1
r1
Furthermore, by symmetry, I is expected to be an odd function of . In
addition, since the macroscopic wavefunction s = ns ei is unaffected when

is replaced by +2 [see Eq. (17.58)], I is expected to be a periodic function
of with a period of 2, i.e. I can be expressed as
I=
In sin (n) ,
(17.91)
n=1
where In are all constants. As will be shown below [see Eq. (17.313)], for the
case where the superconductors are weakly coupled all terms except for the
first one can be neglected. For that case the general current -phase relation
(17.91) becomes the so-called first Josephson relation, which is given by
I = Ic sin ,
(17.92)
where Ic is the so-called critical current of the junction.

17.2.2 The Second Josephson Relation
can be expressed
The time derivative of (17.90), which is denoted by ,
with the help of Eq. (17.67) as

1
ms 2
2 r2
A
2
=
Js (r2 ) Js (r1 ) + qs ( (r2 ) (r1 ))
dr
.
2qs2 n2
t
s
s r1
(17.93)
The assumption Js (r2 ) = Js (r1 ) leads to

2c r2
1 A
=
dr
.
s r1
c t
Eyal Buks
(17.94)
494

With the help of Eq. (17.1) this becomes
2c
=
V ,
s
(17.95)
where V , which is given by

r2
V =
dr E ,
(17.96)
r1
is the voltage across the Josephson junction. As will be shown below, the
elementary superconducting charge carrier is a pair of electrons, i.e. qs = 2e,
and consequently Eq. (17.95) becomes
2eV
=
.
(17.97)
The above result is the second Josephson relation.

17.2.3 The Energy of a Josephson Junction
Let I (t) and V (t) be the current through and voltage across a Josephson
junction, respectively, at time t. The energy UJ of the junction can be evaluated by integrating the work done by the source
t
UJ =
dt I (t ) V (t ) .
(17.98)
With the help of the first (17.92) and second (17.97) Josephson relation this
becomes

Ic
UJ =
d sin ,
(17.99)
2e
thus up to a constant UJ is given by
UJ = EJ cos ,
(17.100)
where
EJ =
Ic
Ic
= s .
2e
2c
The energy UJ (17.100) can be expressed as [see Eq. (17.313)]

:
2
I
UJ = EJ 1
.
Ic
(17.101)
(17.102)
To second order in I this becomes [see Eq. (17.314)]

Eyal Buks
495

LJ I 2
+ O I4 ,
2
UJ = EJ +
where
LJ =
(17.103)
s
2cIc
(17.104)
is the so-called Josephson inductance. Note, however, that an inductor-like

behavior of a Josephson junction is expected only when I Ic .
17.3 RF SQUID
A radio frequency (RF) superconducting quantum interference device (SQUID)
is made of a superconducting loop interrupted by a Josephson junction (see
Fig. 17.1). Consider a close curve C going around the loop. The requirement
that the phase of the macroscopic wavefunction is continues reads
@
2n = dr ,
(17.105)
C
where n is integer. The section of the close curve C inside the superconductor
is denoted by C and the integral through the junction is denoted as an
integral from point r1 to point r2 . With the help of Eq. (17.76) the above
condition becomes
r2

m
2
m
2n = s
dr Js + s
dr Js +
dr A . (17.106)
qs ns r1
qs ns C
s C
Consider the case where the curve is chosen such that the supercurrent density
Js vanishes everywhere on the curve C (i.e. inside the superconductor the
distance between any point on C and the nearest surface is much larger than
the London penetration depth L ). For this case Eq. (17.106) becomes
2n = +
2
,
s
(17.107)
where
=
ms
qs ns
r2
r1
dr Js =
r2
r1
dr GI
(17.108)
is the gauge invariant phase difference across the junction [see Eqs. (17.88)
and (17.90)] and where
@
= dr A
(17.109)
C
Eyal Buks
496
17.3. RF SQUID
Fig. 17.1. RF SQUID is made of a superconducting loop interrupted by a Josephson junction.
is the magnetic flux threaded through the area enclosed by the closed path
C [see Eq. (17.78)].
The junctions critical current is labeled by Ic . It is assumed that the
junction has capacitance, which is denoted by CJ . Consider the case where a
magnetic flux that is denoted by e is externally applied by a magnetic field,
which is applied perpendicularly to the plane of the loop. The total magnetic
flux threading the loop is given by
= e + Is ,
(17.110)
where Is is the circulating current flowing in the loop and is the self inductance of the loop.
17.3.1 Lagrangian
The Lagrangian of the system [see Eq. (1.16)] can be expressed as a function
of the dimensionless flux coordinate , which is defined by
=
2
,
s
(17.111)
According to Faradays law of induction the voltage

and its time derivative .
across the capacitor (in Gaussian units) is
Eyal Buks
497
V = ,
c
(17.112)
and therefore the kinetic energy of the system T is the capacitance energy
T =
2
CJ
CJ 2s 2
=
.
2c2
82 c2
(17.113)
The potential energy U has two contributions, the inductive energy (in
Gaussian units)
Is2
( e )2
2 ( e )2
=
= s
,
2c
2c
8 2 c
(17.114)
where
2e
s
e =
(17.115)
is the normalized external flux, and the Josephson energy UJ [see Eqs.
(17.100) and (17.107)]
UJ =
s Ic
cos .
2c
(17.116)
Thus the Lagrangian L = T U is given by

L=
CJ 2s 2 2s ( e )2 s Ic
+
cos ,
82 c2
8 2 c
2c
or in a dimensionless form by
!
"
2
L = E0
u (; e ) ,
LJ 2p
(17.117)
(17.118)
where the energy constant E0 is given by

E0 =
2s
,
82 c
(17.119)
the junctions plasma frequency p is given by

0
0
c
2ecIc
p =
=
,
LJ CJ
CJ
(17.120)
where LJ = s /2cIc is the Josephson inductance [see Eq. (17.104)], the

dimensionless potential u (; e ) is given by
u (; e ) = ( e )2 2 L cos ,
Eyal Buks
(17.121)
498
17.3. RF SQUID
and the dimensionless parameter L is given by
L =
2Ic
.
s
(17.122)
The resulting Euler - Lagrange equation of motion (1.8) is given by

d L
L
,
(17.123)
=
dt
thus

+ e + L sin = 0 .
LJ 2p
(17.124)
With the help of Eqs. (17.107), (17.110) and (17.112) the equation of motion
can be rewritten as
Is = Ic sin + CJ V .
(17.125)
The above equation states that the circulating current Is equals the sum of
the current Ic sin through the Josephson junction and the current CJ V
through the capacitor.
17.3.2 Flux Quantum Bit
Consider the case where the externally applied flux e is chosen to be close
to a half integer value in units of the superconducting flux quantum s . To
study this case the normalized externally applied flux e (17.115) is taken to
be given by
e = + e,r ,
(17.126)
where e,r 1. Using the notation

= + r ,
(17.127)
the dimensionless potential u (17.121) becomes

u = (r e,r )2 + 2 L cos r .
(17.128)
Consider that case where e,r = 0 (i.e. when e = ). As can be seen

from Eq. (17.110), for this case the relative coordinate r is proportional to
the circulating current Is
r =
2
Is .
s
Eyal Buks
(17.129)
499
Fig. 17.2. The LC resonator that is coupled to the RF SQUID allows readout.
Moreover, when e,r = 0 one finds to second order in r that the potential u
is given by

u = 2 L + (1 L ) 2r + O 4r .
(17.130)
Thus if L > 1 the point r = 0 becomes a local maxima point of u. The

corresponding potential barrier centered at r = 0 (i.e. at = ) separates
two symmetric potential wells on the right and on the left. At sufficiently low
temperatures only the two lowest energy levels are expected to be occupied.
In this limit the Hamiltonian of the system can be expressed in the basis of
the states | and |, that represent localized states in the left and right
well, respectively, having opposite circulating currents. Such a device can be
used as an artificial two-level system, i.e. as a quantum bit (qubit in short).
17.3.3 Qubit Readout
Readout of the state of an RF SQUID can be performed by inductively coupling the superconducting loop to an LC resonator (see Fig. 17.2), which is
made of an inductor and a capacitor in parallel having inductance L and
capacitance C respectively. The mutual inductance between the RF SQUID
and the resonator is denoted by M . Detection is performed by injecting a
monochromatic input current Iin into the LC resonator at a frequency close
to the resonance frequency and measuring the output voltage Vout (see Fig.
17.2).
The total magnetic flux threading the SQUID loop is given by [compare
with Eq. (17.110)]
= e + i ,
Eyal Buks
(17.131)
500
17.3. RF SQUID
where the term i represents the flux generated by both, the circulating
current in the RF SQUID Is and by the current in the inductor of the LC
resonator IL
i = Is + M IL ,
(17.132)
where is the self inductance of the loop. Similarly, the magnetic flux in
the inductor of the LC resonator is given by
= IL L + MIs .
Eqs. (17.132) and (17.133) can be rewritten in a matrix form

i
M
Is
.
=
M L
IL
(17.133)
(17.134)
Inverting the above relation allows expressing the currents Is and IL in terms
of i = e and
i
M
,
(1 K 2 ) L (1 K 2 )
Mi
IL =
,
2
L (1 K ) L (1 K 2 )
Is =
(17.135)
(17.136)
where the dimensionless constant K is given by

M
.
K=
L
(17.137)
Lagrangian. The Lagrangian of the system L = T U [see Eq. (1.16)] is

expressed below as a function of the coordinates and and their time
derivatives and .
The contributions to the total kinetic energy T are the
capacitance energy of the Josephson junction that is given by Eq. (17.113)
and the capacitance energy of the capacitor in the LC resonator, thus T is
given by
T =
CJ 2s 2 C 2
+ 2 .
8 2 c2
2c
(17.138)
The inductive energy UI stored in the RF SQUID loop and the lumped inductor L is calculated using Eqs. (17.135) and (17.136)
Eyal Buks
501

M
1
Is
Is IL
M L
IL
2c
1

M

1
L
i
=
M
1
L
2c (1 K 2 ) i
L
UI =
2i
2 M
i
L
+ L
2c (1 K 2 )

2
M
2
i
L
+
=
2cL 2c (1 K 2 )

2
2
2M
2 2
e
s
s L
C e
=
+
,
2
2c
8 c (1 K 2 )
(17.139)
where
1
e =
LC
(17.140)
is the LC angular resonance frequency. The total potential energy U is given

by
U = UI
Iin s Ic
cos ,
c
2c
(17.141)
where the term Iin /c is the potential energy of the current source and
(s Ic /2c) cos is the Josephson energy [see Eq. (17.116)].
With the help of the above relations one finds that the Lagrangian of the
system L = T U can be expressed as
L = L0 + L1 ,
(17.142)
where L0 , which is given by

L0 =
C 2 C2e 2 Iin
+
,
2c2
2c
c
(17.143)
is the Lagrangian of the driven LC resonator. The Lagrangian of the superconducting loop L1 is given by [see Eqs. (17.119), (17.120) and (17.122)]

2
2
2M
2 2
e
s
s L
CJ s
Ic
L1 =
+ s cos
82 c2
8 2 c (1 K 2 )
2c
!
"
2
= E0
uK (; e,eff ) ,
LJ 2p
(17.144)
Eyal Buks
502
17.3. RF SQUID
where the dimensionless potential uK (; e,eff ) is given by [compare with Eq.
(17.121)]
2
uK (; e,eff ) =
( e,eff )
2 L cos ,
1 K2
(17.145)
and where the effective external flux e,eff is given by

e,eff = e +
2M
.
s L
(17.146)
Note that L1 depends on the effective external flux e,eff , which, in turn,
depends on the coordinate of the LC resonator [see Eq. (17.146)]. This
dependence gives rise to the coupling between the LC resonator and the RF
SQUID.
The Euler - Lagrange equations (1.8) are given by

d L
L
=
,
(17.147)
dt

d L
L
=
,
(17.148)
dt
thus
CJ s
=
2c
s (e )
2
M
L
Ic sin ,
(1 K 2 )
(17.149)
and
( )
e
s
M
C
2
=
+ Iin .
c
L (1 K 2 )
(17.150)
With the help of Eqs. (17.135) and (17.136) the equations of motion can be
rewritten as
Is = Ic sin
CJ
,
c
(17.151)
and
Iin =
+ IL .
c
(17.152)
While Eq. (17.151) expresses the law of current conservation in the SQUID
loop, Eq. (17.152) expresses the same law in the LC resonator.
Eyal Buks
503

Hamiltonian. The variables canonically conjugate to and are given by
[see Eq. (1.20)]
L
2E0
=
.
LJ 2p

L
C
= 2 .
q=

c
Q=
(17.153)
(17.154)
The Hamiltonian is given by [see Eq. (1.22)]

H = Q + q L = H0 + H1 ,
(17.155)
where
H0 =
c2 q 2 C 2e 2 Iin
+
,
2C
2c
c
(17.156)
LJ 2p Q2
+ E0 uK (; e,eff ) .
4E0
(17.157)
and where
H1 =
Quantization is achieved by regarding the variables {, Q, , q} as Hermitian operators satisfying the following commutation relations [see Eqs.
(3.6), (3.7) and (3.8)]
[, Q] = [, q] = i ,
(17.158)
[, ] = [, Q] = [q, ] = [q, Q] = 0 .
(17.159)
and
Adiabatic Approximation. The dynamics of the superconducting loop is

typically much faster than the one of the LC resonator. This can be exploited
for simplifying the systems equations of motion by employing the adiabatic
approximation (see section 13.6). Similarly to L1 , the Hamiltonian H1 of the
superconducting loop depends on the effective external flux e,eff , which in
turn depends on the coordinate of the LC resonator [see Eq. (17.146)].
As a basis for expanding the general solution we use the eigenvectors of
the following Schrdinger equation
H1 |n () = n () |n () ,
(17.160)
where is treated here as a parameter (rather than a degree of freedom).

The local eigenvectors are assumed to be orthonormal
m () |n () = nm .
(17.161)
The eigenenergies n () and the associated wavefunctions n (, ) are

found by solving the following Schrdinger equation [see Eq. (4.50)]
Eyal Buks
504
17.3. RF SQUID
J
n
d2 n
+ uK (; e,eff ) n =
,
d2
E0 n
(17.162)
where
LJ
J =
p
2E0
2
(17.163)
The state of the entire system (t) at time t is expanded at any point
using the local basis {|n ()} [see Eq. (13.71)]

(t) =
n (, t) |n () .
(17.164)
n
In the adiabatic approximation the time evolution of the coefficients n is governed by the following set of decoupled equations of motion [see Eq. (13.83)]
[H0 + n ()] n = i n .
(17.165)
Note that in the present case the geometrical vector potential [given by Eq.
(13.76) for the case m = m] vanishes since the wavefunctions n (, ) can
be chosen to be real [see Eq. (13.84)].
Two level approximation. Consider again the case where the externally
applied flux e is chosen to be close to a half integer value in units of the
superconducting flux quantum s . The potential uK (17.145) can be expressed
as
uK =
(r e,eff,r )2
+ 2 L cos r ,
1 K2
(17.166)
where e,eff,r and r are defined by [see Eqs. (17.126) and (17.127)]
2M
= + e,eff,r ,
s L
= + r .
e,eff = e +
(17.167)
(17.168)

In what follows we focus on the case where |e,eff,r | 1 and L 1 K 2 >
1. In this case the potential uK given by Eq. (17.166) contains two wells separated by a barrier near r = 0 (see Fig. 17.3). As was discussed above, at low
temperatures only the two lowest energy levels are expected to contribute. In
this limit the Hamiltonian H1 can be expressed in the basis of the states |
and |, representing localized states in the left and right well respectively
having opposite circulating currents. In this basis, H1 is represented by the
2 2 matrix

e,eff,r
.
(17.169)
H1 =
e,eff,r
Eyal Buks
505

The real parameters and (having units of energy) can be determined
by solving numerically the Schrdinger equation (17.162). It is convenient to
express in terms of the parameter Icc
s Icc
.
(17.170)
2c
The physical meaning of the parameter Icc , which has units of current, will
be discussed below.
Using the notation
=
tan =
2c
=
,
e,eff,r
s Icc e,eff,r
H1 can be rewritten as
:

2
s Icc e,eff,r
cos sin
+ 2
.
H1 =
sin cos
2c
(17.171)
(17.172)
The eigenvectors and eigenenergies are denoted as

H1 | = | ,
(17.173)
where [compare with Eqs. (6.213) and (6.214)]

cos 2
sin 2
|+ =
;
|
=
,
sin 2
cos 2
(17.174)
and where
:
2
s Icc e,eff,r
=
+ 2
2c

2

1 s Icc e,eff,r
= 1 +
+ O 4e,eff,r .
2
2c
(17.175)
Exercise 17.3.1. Show that the expectation value of the circulating current
Is when the qubit is in state n is given by
n| Is |n = c
n
.
e
(17.176)
Solution 17.3.1. With the help of Eqs. (17.110), (17.132), (17.135), (17.145)
and (17.157) one finds that
c
H1
= Is ,
e
(17.177)
thus
n| Is |n = c n|
Eyal Buks
H1
n
|n = c
.
e
e
(17.178)
506
17.3. RF SQUID
Fig. 17.3. Eigenstates of H1 . (a)-(c) The first 3 lowest energy states for the case
e,eff,r = 0. (d) The energy of the two lowest states vs. e,eff,r .
Eyal Buks
507

With the help of Eqs. (17.175) and (17.176) one finds that the expectation
value of the circulating current Is corresponding to the eigenvectors |+ and
| is given by
s Icc e,eff,r
2c
| Is | = Icc 0
1+
s Icc e,eff,r
2c
2 .
(17.179)
In the limit where s Icc |e,eff,r | /2c (i.e. when e,eff is far from the
point e,eff = ) the eigenvectors of H1 become the states | and | [see
Eqs. (17.171) and (17.174)]. Thus, in view of the above result (17.179), it is
evident the the constant Icc represents the absolute value of the circulating
current associated with the states | and |.
In the adiabatic approximation the dynamics of the LC resonator is governed by the Hamiltonian H , which is given by
H = H0 + () ,
(17.180)
when the qubit is the state | [see Eq. (17.165)]. With the help of Eqs.
(17.156), (17.167) and (17.175) one finds that
H =
c2 q 2
Iin
+ U ()
,
2C
c
(17.181)
where
C 2e 2
U () =
2c
:
s Icc e,eff,r
2c
2
+ 2 ,
(17.182)
and where
e,eff,r = e +
2M
.
s L
(17.183)
For simplicity, consider the case where e = . For that case one finds to
second order in M that [see Eqs. (17.140) and (17.175)]
U () =
where
= e

C 2 2
+ O M4 ,
2c
2
M M Icc
.
L c
(17.184)
(17.185)
Thus the effective angular resonance frequency of the LC resonator depends on the state being occupied by the qubit. This dependence allows
reading out the qubit state by measuring the resonance frequency of the LC
resonator.
Eyal Buks
508
17.4. BCS Model
17.4 BCS Model

This chapter briefly discusses the BCS microscopic model of superconductivity.
17.4.1 Phonon Mediated Electron-Electron Interaction
Let (r ) be the electron density in a medium having volume V. Classically, the two-particle Coulomb interaction VTP (r1 , r2 ) = e2 / |r1 r2 | [see
Eq. (16.106)] gives rise to energy V given by [see Eqs. (16.65) and (16.43) for
comparison with the analogous second-quantization expression]

1
V =
d3 r d3 r VTP (r , r ) (r ) (r ) .
(17.186)
2
With the help of the Fourier expansion

1
(r ) =
d3 q (q ) eiq r
3/2
(2)
V
(17.187)
and Eqs. (4.47) and (16.127) one finds that [see Eq. (16.128) for comparison
with the analogous second-quantization expression]

4e2
1
(17.188)
V =
d3 q 2 (q) (q) .
2V
q
The effect of induced charges in the medium (i.e. screening) can be taken
into account by dividing by the dielectric constant of the medium [see Eq.
(17.43)]

1
4e2
V =
d3 q 2 (q) (q) .
(17.189)
2V
q
Ionic Dielectric Constant. The expression for the electronic dielectric constant e () given by Eq. (17.49) is evaluated by calculating the induced electron charge density e due to external charge density ext . Below, however,
we focus on the induced ion charge density i associated with lattice vibrations and the resultant ionic dielectric constant i . To account for the effect of
lattice vibrations on electron-electron interaction both contributions to the
total induced charge density e + i have to be taken into account. With
analogy to Eq. (17.46), the equation of motion for the ion induced charge
density i is given by
d2 i
1 di
+
= 2p,i (e + i + ext ) ,
dt2 tr,i dt
(17.190)
where p,i , which is given by

Eyal Buks
509

2p,i =
4qi2 ni
,
mi
(17.191)
is the ion plasma frequency and where mi , qi and ni are the ionic mass, charge
and density, respectively. Note that p,i p,e since the ion mass mi is much
larger than electron mass me .
Thomas-Fermi Approximation. The electron density ne is assumed to be
a constant provided that the media is homogeneous. It can be evaluated by
summing the Fermi-Dirac function fFD (i ) over all electronic states having
energies i [see Eq. (16.147)]
ne () =
1
1
1
fFD (i ) =
,
V i
V i exp [ (i )] + 1
(17.192)
where V is the volume, 1 = kB T is the thermal energy and where is

the chemical potential. The external charge density ext together with the
induced charge density e + i give rise to a spatially varying scalar potential
(r), which is related to the total charge density by the Poissons equation
2 = 4 (e + i + ext ) .
(17.193)
In the Thomas-Fermi approximation, which is valid provided that (r) is a

slowly varying function of position on the length scale of electron wavelength,
the local value of electron density is taken to be given by ne ( qe (r)),
and consequently the electron induced charge density e (r) is taken to be
given by
e (r) = qe [ne ( qe (r)) ne ()] .
(17.194)
For sufficiently small this becomes

e (r) = qe2
ne
(r) .
(17.195)
When the thermal energy kB T is much smaller than the Fermi energy F the
factor ne / is approximately the density of states at the Fermi energy F ,
ne
m2 vF
2e 3 ,
(17.196)
where vF is the so-called Fermi velocity (which is defined by the relation

k /k = vF , where k is the wave number and where the derivative is
taken at k = F ). For this case Eq. (17.195) becomes
e (r) = qe2
Eyal Buks
m2e vF
(r) .
2 3
(17.197)
510
17.4. BCS Model

Combing Eqs. (17.193) and (17.197) yields the following relation between the
electron induced charge density e and the total charge density e + i + ext
2
2 e = kTF
(e + i + ext ) ,
(17.198)
where
2
kTF
=
4qe2 m2e vF
.
2 3
(17.199)
Effective Electron-Electron Interaction. Consider the case where ext

is taken to oscillate in time and space according to Re ext,0 ei(qrt) ,
where both the real angular frequency and the real vector q are constants.
In steady state the induced
charges

oscillate according to e =
Re e,0 ei(qrt) and i = Re i,0 ei(qrt) . Substituting into Eqs. (17.190)
and (17.198) yields
2 i,0
and

i
= 2p,i e,0 + i,0 + ext,0 ,
tr,i i,0
(17.200)

2
q 2 e,0 = kTF
e,0 + i,0 + ext,0 .
(17.201)

2
kTF
e,0 =

2 1+ i
tr,i
2 )
(q 2 + kTF
2 )
kTF
2p,i
2 1+ i
tr,i
2p,i

ext,0
"
"
(17.202)
(17.203)
2
+ kTF
and
i,0 =
(q 2
q 2 ext,0

2 1+ i
tr,i
2p,i
2
kTF
thus the ionic dielectric constant i (q, ) is given by [see Eq. (17.40)]
e,0 + i,0 + ext,0
1
=
i (q, )
ext,0
2
=
2 )
(q 2 + kTF
q
!

2 1+ i
tr,i
2p,i
2 1+ i
tr,i
2p,i
"
.
2
+ kTF
(17.204)
Eyal Buks
511

For the case where tr,i 1 this becomes
2
1
q2
= 2
2
2 ,
2
i (q, )
q + kTF p,i
(17.205)
where
2
p,i
= 2p,i
q2
.
2
q 2 + kTF
(17.206)
The expression for the Coulomb energy (17.189) together with the result
for the ionic dielectric constant i (q, ) (17.205) lead to the effective interaction coefficient for a pair of electrons having wave vectors k and k and
energies k and k respectively
vk ,k =
2
4e2
4e2
=
2 2 2 ,
q 2 i (q, )
q 2 + kTF
p,i
(17.207)
where q = k k and where = (k k ) /.

The fact that 1
i (q, ) becomes negative when < p,i indicates that
the effective (i.e. phonon mediated) electron-electron interaction becomes attractive in the limit of low frequencies. The characteristic energy interval
p,i in which the interaction becomes attractive is of the order of the socalled Debye energy D , which represents the largest energy of an acoustic
phonons in the lattice.
17.4.2 The Hamiltonian
In the BCS model the Hamiltonian of electrons in a superconducting metal
is taken to be given by

H=
(k F ) ak , ak , + ak , ak ,
k
g
V
F D <k ,k <F +D
k ,k
ak , ak , ak , ak , .
(17.208)
where labels spin up state, labels spin down state, k is the energy of
both single particle states |k , and |k , and F is the Fermi energy [see
Eq. (16.101)]. The coupling constant g > 0 gives rise for an effective electronelectron attracting interaction. The interaction is assume to couple pairs of
electrons whose energies are inside an energy interval of width 2D around
the Fermi energy F .
As can be seen from the comparison with the more general many-particle
interaction operator V given by Eq. (16.93), the BCS Hamiltonian contains
Eyal Buks
512
17.4. BCS Model

only interaction terms that represents annihilation (the factor ak , ak , )
and creation (the factor ak , ak , ) of electrons pairs having zero total angular momentum. Moreover, the summation is restricted only to the energy
interval of width 2D in which attractive interaction is expected, and the effective interaction coefficients are all assumed to be identical [see for comparison
Eq. (17.207)].
17.4.3 Bogoliubov Transformation
As will be seen below, the Hamiltonian can be approximately diagonalized
by employing the Bogoliubov transformation [see Eqs. (16.110), (16.153) and
(16.154)]
bk,
ak,
e k cos k
0
0
eik sin k
bk,
ak,
0
eik cos k eik sin k
0
=
,
ik
ik
bk,
ak,
0
e sin k e
cos k
0
ik
ik
0
0
e
cos k
e sin k
bk,
ak,
(17.209)
where both k and k are real. The inverse transformation is given by
ak,
bk,
0
0
eik sin k
e k cos k
ak,
bk,
0
eik cos k eik sin k
0
=
.
ik
ik
ak,
bk,
0
e sin k e cos k
0
ik
ik
e sin k
0
0
e cos k
ak,
bk,
(17.210)
The following holds

ak, ak, = sin2 k (1 Nk, ) + cos2 k Nk,

sin (2 k )
+
bk, bk, + bk, bk, ,
2
(17.211)
ak, ak, = sin2 k (1 Nk, ) + cos2 k Nk,

sin (2k )
+
bk, bk, + bk, bk, ,
2
(17.212)
and
where
Nk, = bk, bk,
(17.213)
is number operator (with respect to the bk, and bk, operators ).
Eyal Buks
513

Using the above results together with Eqs. (16.153) and (16.154) one finds
that the Hamiltonian can be rewritten as

(k F ) sin2 k
H=2
k

k
(k F ) cos (2k ) (Nk , + Nk , )

(k F ) sin (2k ) bk , bk , + bk , bk ,
k
F D <k ,k <F +D
g
V
Bk Bk ,
k ,k
(17.214)
where
Bk = ak , ak ,
=
e2ik sin (2k )

(Nk , + Nk , 1)
2
+e2ik sin2 k bk , bk , cos2 k bk , bk ,
.
(17.215)
Formally, the operator Bk can be written as

Bk = Bk + Bk ,
(17.216)
where Bk is the expectation value of Bk in thermal equilibrium and

where
Bk = Bk Bk .
(17.217)
In the mean field approximation the coupling term Bk Bk is approximated

by
'
'
(
(
(17.218)
Bk Bk Bk Bk + Bk Bk + Bk Bk ,
i.e. the term Bk , which represents the deviation from the expectation value,
is considered as small. Employing this approximation, using Eq. (17.215)
and disregarding all constant terms [the' first(line in Eq. (17.214) and terms
containing only expectation values, e.g. Bk Bk ] result in the mean field
Hamiltonian HMF , which is give by
Eyal Buks
514
17.4. BCS Model

HMF =
F D <k <F +D

k

2i

e k + e2ik sin (2k )
(k F ) cos (2k )
2
(Nk , + Nk , )
+
F D <k <F +D

k
bk , bk ,
+
F D <k <F +D
bk , bk ,
(k F ) sin (2k ) + e2ik cos2 k e2ik sin2 k
(k F ) sin (2k ) + e2ik cos2 k e2ik sin2 k
(17.219)
where
g
=
V
F D <k <F +D

k
Bk .
(17.220)
Note that all terms outside the energy interval F D < k < F + D were
disregarded in the above expression for HMF .
Diagonalization is achieved when the last two sums in Eq. (17.219) vanish,
i.e. when the terms e2ik are real and positive and when the parameters
k are chosen such that the following condition is satisfied
(k F ) sin (2k ) + || cos (2 k ) = 0 .
(17.221)
For this case

k =

1
||
tan1
,
2
k F
(17.222)
i.e.
and
||
sin (2k ) = .
,
(k F )2 + ||2
k F
,
cos (2k ) = .
2
2
(k F ) + ||
sin ( k ) =
cos ( k ) =
Eyal Buks
7
8
8 1 k F
9
(k F )2 +||2
2
7
8
8 1 + k F
9
(k F )2 +||2
2
(17.223)
(17.224)
(17.225)
(17.226)
515

and thus HMF becomes diagonal

k Nk , ,
HMF =
(17.227)
k ,
where
k =
.
2
2
(k F ) + || .
(17.228)
17.4.4 The Energy Gap

The value of the energy gap || can be determined from Eq. (17.220). Let
nk , denotes the expectation value of the operator Nk , , i.e.
Nk , = nk , .
(17.229)
In thermal equilibrium at temperature T the following holds [see Eqs.

(16.144), (16.145), (17.227) and (17.228)]
1
,
(17.230)
ek + 1
where = 1/kB T', and where( kB is Boltzmanns constant. Moreover, in thermal equilibrium bk , bk , = bk , bk , = 0 and thus Bk is given by
nk , =
[see Eq. (17.215) and recall that || = e2ik ]

Bk =
(1 nk , nk , )
.
,
2 (k F )2 + ||2
(17.231)
and thus Eq. (17.220) can be expressed as

1=
g
2V
F D <k <F +D

k
1 nk , nk ,
.
,
(k F )2 + ||2
or [see Eqs. (17.228) and (17.230)]

1
F D <k <F +D

1 2 ek + 1
g
1=
2V
k
k

F D <k <F +D tanh k

2
g
=
.
2V
k
(17.232)
Replacing the sum by an integral leads to

2 +||2
gD0 D tanh
2
1=
d .
,
2 D
2
+ ||2
(17.233)
(17.234)
where D0 is the density of states per unit volume.

Eyal Buks
516
17.4. BCS Model

Zero Temperature. For the case of zero temperature, where all occupation
numbers vanish, i.e. nk , = 0, Eq. (17.234) becomes

gD0 D
d

1=
2 D 2 + 20

D + 2D + 20
gD0

=
log
,
2
D + 2D + 20
(17.235)
where 0 stands for the value of || at zero temperature. The assumption
0 D leads to
1=
gD0
42
,
log D
2
20
(17.236)
thus

1
0 = 2D exp
.
gD0
(17.237)
Critical Temperature. The energy gap || vanishes when T = Tc , where

Tc is the critical temperature. For this case Eq. (17.234) becomes

gD0 D tanh c2
1=
d
2 D
c2 D
tanh x
= gD0
dx
,
x
0
(17.238)
where c = 1/kB Tc . Integration by parts (note that limx0 tanh x log x = 0)
yields
!
"
c2D
c D
log x
c D
1 = gD0 tanh
dx
.
(17.239)
log
2
2
cosh2 x
0
For the case of weak coupling, for which
c D
1,
2
(17.240)
one has

D
log x
1 gD0 log c
dx
.
2
cosh2 x
0
(17.241)
Using the identity
Eyal Buks
517
dx
log x
4
= log + CE ,
2
cosh x
(17.242)
where CE 0.577 is Eulers constant, one finds that [see Eq. (17.237)]
eCE
0 0.566 0 .
(17.243)
General Temperature. The energy gap || at temperature T can be numerically evaluated from Eq. (17.234). To a good approximation the solution
can be expressed by the following analytical relation
:
3
T
|| 0 1
.
(17.244)
Tc
kB Tc =
17.4.5 The Ground State

<
The ground state |0 of the mean field Hamiltonian HMF = k , k Nk ,
(17.227) is a state for which all occupation numbers vanish, i.e. Nk , |0 =
bk, bk, |0 = 0, and therefore bk, |0 = 0 for all k and . Moreover, |0
is required to be normalized, i.e. 0 |0 = 1.
Claim. The ground state |0 is given by
#
|0 =
Kk |0 ,
(17.245)
where

Kk = eik cos k eik sin k ak , ak , .
(17.246)
4
3
Proof. By employing the fact that Kk , Kk = 0 provided that k = k and
the relation
Kk Kk = cos2 k + sin2 k ak , ak , ak , ak ,

sin k cos k e2ik ak , ak , + e2ik ak , ak , ,
(17.247)
one finds that |0 is indeed normalized as required

# #
0 |0 = 0|
Kk
Kk |0
k
= 0|
= 0|
=1.
#
k
Kk Kk |0

4
#3
cos2 k + sin2 k 1 ak , ak , 1 ak , ak , |0
k
(17.248)
Eyal Buks
518
17.4. BCS Model

Moreover, using the relations [see Eq. (17.209)]

bk , Kk = eik cos k ak , eik sin k ak ,

eik cos k eik sin k ak , ak , ,
and

bk, Kk = eik cos k ak , eik sin k ak ,

eik cos k eik sin k ak , ak , ,
one finds that

bk , |0 = bk ,
#
#
k
k
=k
=
=
k
=k
k
=k
(17.250)
Kk |0
Kk bk , Kk |0
(17.249)

Kk sin k cos k ak , ak , + 1 ak , |0
Kk sin k cos k ak , ak , ak , |0
=0,
(17.251)
and similarly
bk, |0 = 0 .
Alternatively, the ground state |0 , which is given by Eq. (17.245), can
be expressed as

#
|0 = C0
1 k ak , ak , |0 ,
(17.252)
k
where C0 is a normalization constant, which is given by

#
C0 =
eik cos k ,
(17.253)
and where [see Eqs. (17.225) and (17.226)]

k = e2ik tan k
7
8
8 1 k 2F
(k F ) +||2
2ik 8
9
=e
.
1 + k 2F
2
(k F ) +||
Eyal Buks
(17.254)
519

2

Furthermore, since ak , ak , = 0 the following holds
!
|0 = C0 exp

k
k ak , ak ,
"
|0 .
(17.255)
17.4.6 Pairing Wavefunction

For a general function of position (r ) the following holds

dr
dr (r r ) (r ) (r )

1
=
ak , ak , dr dr (r r ) eik r ik r
V
k ,k

1
dr ei(k k )r
dr (r ) eik r
=
ak , ak ,
V
k ,k

=

k
ak , ak ,
=k ,k
dr (r ) eik r
(17.256)
where (r ) is quantized field operators [see Eq. (16.96)]. In view of the

above result the ground state |0 , which is given by Eq. (17.255), can be
expressed as

|0 = C0 exp
dr
dr (r r ) (r ) (r ) |0 ,
(17.257)
where the function (r r ), which is called the pairing wavefunction, satisfy

dr (r ) eik r = k ,
(17.258)
where k is given by Eq. (17.254).
The energy region near F in which k changes significantly has a characteristic width given by the energy gap 0 [see Eq. (17.254)]. The corresponding region in k space has thus a characteristic size given by 0 /vF ,
where vF is the so-called Fermi velocity (which is defined by the relation
k /k = vF , where the derivative is taken at k = F ). Consequently the
paring wavefunction (r ) is expected to have a characteristic size given
by , where
=
vF
,
||
(17.259)
is the so-called BCS coherence length.

Eyal Buks
520
17.5 The Josephson Effect

Consider the global transformation ak , ak , ei/2 and ak , ak , ei/2 ,
where is real. Such a transformation leaved the Hamiltonian (17.208) unchanged, however, the factor Bk is transformed according to Bk Bk ei
[see Eq. (17.215)] and the energy gap is transformed according to [see Eq.
(17.220)]
ei .
(17.260)
Moreover, the ground state |0 is modified [see Eq. (17.252)] and becomes
|0 | (), where
#
| () =
Kk () |0 ,
(17.261)
k
where the operator Kk () is given by

Kk () = eik cos k ei eik sin k Bk .
(17.262)
As can be seen from Eq. (17.261), the vector state | () becomes identical
to the ground state |0 (17.252) when = 2n, where n is integer. In view
of the fact that the pair creation operator Bk = ak , ak , (17.215) in Eq.
(17.262) is multiplied by the factor ei one may argue that the phase can
be considered as the phase of Cooper pairs.
Claim. The state | () (17.263) can be alternatively expressed as
| () = einP |0 ,
(17.263)
where |0 is the BCS ground state (17.252) and where

nP =
1
a ak ,
2 k ,
(17.264)
k ,
is the so-called pair number operator.

Proof. On one hand

ak , ak , + ak , ak , Kk () |0

= ei eik sin k ak , ak , + ak , ak , ak , ak , |0
3

4
= ei eik sin k ak , ak , 1 ak , ak , + ak , ak , 1 ak , ak , |0

= ei eik sin k ak , ak , + ak , ak , |0 ,
(17.265)
Eyal Buks
521

and thus
#
#
1
nP
Kk () |0 =
ak , ak , + ak , ak ,
Kk () |0
2 k
k
k
#
=
Kk () ei eik sin k ak , ak , |0 .
k k
=k
On the other hand

#
Kk
#
i
Kk () |0 = i
Kk ()
|0 ,
k
k k
=k
(17.266)
(17.267)
and therefore [see Eq. (17.262)]

nP | () = i
where | () =
>
k
| () ,
(17.268)
Kk () |0 [see Eq. (17.261)]. The above result together
with the Taylor expansion formula for the exponential function [see Eq.
(3.31)] lead to
einP | (0) = | () ,
(17.269)
which proofs the claim (17.263).

17.5.1 The Second Josephson Relation
Consider the case where a voltage V is applied to a superconductor. The
added energy of = eV per electron, where e is the electron charge, can be
taken into account by adding a term to the Hamiltonian of the system, which
becomes
H (V ) = HMF + 2nP ,
(17.270)
where HMF is given by Eq. (17.227) and where the pair number operator nP
is given by Eq. (17.264). As will be shown below, the added term 2nP gives
rise to time dependence of the complex energy gap [see Eq. (17.220)].
Claim. The following holds
i
d
= 2 .
dt
(17.271)
Proof. With the help of the Heisenberg equation of motion (4.38) one finds
that
i
d Bk
= [Bk , H (V )] .
dt
Eyal Buks
(17.272)
522

thus [see Eqs. (17.211), (17.212), (17.215) and (17.227)]
'3
4(
d Bk
= k e2ik sin2 k bk , bk , cos2 k bk , bk , , Nk , + Nk ,
i
dt
4(
e2ik sin2 (2k ) '3
+
Nk , + Nk , , bk , bk , + bk , bk ,
2
4(
'3
2ik
+e
cos (2k ) sin2 k bk , bk , cos2 k bk , bk , , Nk , + Nk ,
'3
4(
+e2ik sin (2k ) bk , bk , , bk , bk ,
.
(17.273)
With the help of the commutation relations

3
4
bk , bk , , Nk , + Nk , = 2bk , bk , ,
3
4
bk , bk , , bk , bk , = Nk , + Nk , 1 ,
(17.274)
(17.275)
one finds that

i
d Bk
= e2ik sin (2k ) Nk , + Nk , 1 ,
dt
(17.276)
and therefore [see Eqs. (17.223) and (17.231)]

i
d Bk
= 2 Bk .
dt
(17.277)
Thus, the complex energy gap , which is given by [see Eq. (17.220)]
g
=
Bk ,
(17.278)
V
k
satisfies Eq. (17.271).

For a fixed the solution of Eq. (17.271) reads
(t) = (0) ei(t) ,
(17.279)
where the phase factor (t) is given by

(t) =
2t
2eV t
=
.

(17.280)
Taking the time derivative (which is denoted by overdot) yields, in agreement

with Eq. (17.97), the second Josephson relation
2eV
=
.
Eyal Buks
(17.281)
523

17.5.2 The Energy of a Josephson Junction
Consider a system composed of two superconductors that are separated one
from the other by a thin insulating layer, which serves as a tunneling barrier.
The Hamiltonian of the system is assumed to be given by
H = H1 + H2 + HT ,
(17.282)
where H1 and H2 are the Hamiltonians of the two decoupled superconductors

and where the tunneling Hamiltonian HT is taken to be given by

HT =
tk ,k a1,k , a2,k , + a1,k , a2,k ,
k ,k

+tk ,k a2,k , a1,k , + a2,k , a1,k , ,
(17.283)
where the annihilation operators of the first (second) superconductor are

labeled by a1,k , (a2,k , ). With the help of Eq. (17.210) one finds that HT
can be expressed as

HT =
Hk ,k ,
(17.284)
k ,k
where

Hk ,k = Ak ,k b1,k , b2,k , b1,k , b2,k ,
+Ak ,k (b2,k , b1,k , b2,k , b1,k , )

+Bk ,k b1,k , b2,k , + b1,k , b2,k ,

+Bk ,k b2,k , b1,k , + b2,k , b1,k , ,
(17.285)
the coefficients Ak ,k and Bk ,k are given by
Ak ,k = k ,k cos 1,k sin 2,k + k ,k sin 1,k cos 2,k ,

Bk ,k = k ,k sin 1,k sin 2,k k ,k cos 1,k cos 2,k ,
(17.286)
(17.287)
and where
k ,k = ei(1,k 2,k ) tk ,k .
(17.288)
We employ below time independent perturbation theory to calculate the

correction E to the systems energy to lowest nonvanishing order in the tunneling coefficients |tk ,k |. The averaged total energy change E is evaluated
by summing over all basis states of the combined system and multiplying the
energy change of each state by its thermal occupation probability. As can be
Eyal Buks
524

seen from Eq. (9.32) E vanishes to first order in |tk ,k |. To second order in
|tk ,k | the correction E is found to be given by

(1 nk ) (1 nk )
nk nk
2
E = 2
|Ak ,k |
k + k
k + k
k ,k

(1 nk ) nk
2 nk (1 nk )
,
+2
|Bk ,k |
+
k k
k k
k ,k
(17.289)
where k is given by Eq. (17.228) and nk is given by Eq. (17.230). With the
help of Eqs. (17.223), (17.286), (17.287) and (17.288) one finds that

2
2
|Ak ,k | = | k ,k | cos2 1,k sin2 2,k + sin2 1,k cos2 2,k
!
"
t2k ,k 1 2
1
+ Re
,
2
1,k 2,k
(17.290)
and

|Bk ,k |2 = | k ,k |2 sin2 1,k sin2 2,k + cos2 1,k cos2 2,k
!
"
t2k ,k 1 2
1
Re
.
2
1,k 2,k
(17.291)
In what follows it will be assumed, for simplicity, that all tunneling amplitudes tk ,k are identical. Moreover, the two superconductors will be assumed
to be of the same type, i.e. |1 | = |2 | ||. For this case all the terms
t2k ,k 1 2 can be expressed as
t2k ,k 1 2 = T ||2 ei ,
(17.292)
where T = |tk ,k |2 and where is the relative phase difference between

the two superconductors. The energy correction E can be expressed as a
function of as
E = (E)0 EJ cos ,
(17.293)
where (E)0 is independent on and where

T ||2 1 nk nk
nk nk
EJ =
+
1,k 2,k
k + k
k k
k ,k
k ,k
Eyal Buks
T ||2 (1 2nk ) k (1 2nk ) k

.
1,k 2,k
2k 2k
(17.294)
525

Replacing the sum by an integral leads to
EJ =
||2
e2 RN

0
d1 d2 tanh
1 2
2
2 1
21
tanh 2 1 2
,
22
where VD0 is the density of states,

.
2
n = 2n + || ,
(17.295)
(17.296)
[see Eq. (17.228)] and where RN , which is given by

,
4e2 V 2 D02 T
RN =
(17.297)
is the so-called normal state resistance.

The variable transformation
n = || cosh n ,
n = || sinh n ,
(17.298)
(17.299)
leads to
EJ =
||
I
e2 RN
||
2
(17.300)
where the function I (x) is given by

tanh (x cosh 2 ) cosh 1 tanh (x cosh 1 ) cosh 2
I (x) =
d1 d 2
.
cosh2 1 cosh2 2
0
0
(17.301)
In the limit of zero temperature the integral can be evaluated using the
variable transformation
1 + 2
,
(17.302)
p =
2
1 2
m =
,
(17.303)
2
which together with the identities
cosh 1 + cosh 2 = 2 cosh p cosh m ,
cosh 1 cosh 2 = 2 sinh p sinh m ,

d
=,
cosh
(17.304)
(17.305)
(17.306)
lead to
Eyal Buks
526

||
d1 d2
2
e RN 0
cosh 1 + cosh 2
0
||
d1 d2
=
2
4e RN cosh 1 + cosh 2

||
dp
d m
=
,
2
4e RN cosh p cosh m
EJ =
(17.307)
thus
EJ =
||
.
4e2 RN
(17.308)
For arbitrary temperature the result is

EJ =
||
||
tanh
.
4e2 RN
2
(17.309)
17.5.3 The First Josephson Relation

As was shown above [see Eq. (17.293)], the energy of a Josephson junction
UJ having phase relative to the energy when the phase vanishes is given
by [compare with Eq. (17.100)]
UJ = EJ cos .
(17.310)
Let I (t) and V (t) be the current through and voltage across a Josephson
junction, respectively, at time t. Assume that initially at time t = 0 the
phase vanishes. Energy conservation leads to the requirement that
t
UJ =
dt I (t ) V (t ) .
(17.311)
0
With the help of the second Josephson relation = (2e/) V (17.281) and
Eq. (17.310) this becomes

EJ cos =
d I (t ) .
(17.312)
2e 0
Taking the derivative with respect to leads, in agreement with Eq. (17.92),
to the first Josephson relation
I = Ic sin ,
(17.313)
where the so-called critical current Ic is given by

Ic =
2eEJ
2cEJ
=
,

s
Eyal Buks
(17.314)
527

where
hc
(17.315)
2e
is the superconducting flux quantum, which is identical to the superconducting flux quantum given by Eq. (17.80) provided that the charge qs is taken to
be 2e. Note also that for the normal flux quantum 0 given by Eq. (12.48)
the charge of elementary carrier is e.
s =
17.6 Problems
1. Rotating Superconductor - Consider a superconductor rotating at
angular frequency around the z axis. In the presence of an externally
applied magnetic field B calculate the magnetic field deep inside the
superconductor.
2. Consider a conductor containing charge carriers having charge q and
mass m. The density of charge carriers at point r is n (r) and the current
density is J (r). Contrary to the case of a normal metal, it is assumed
that all charge carriers at point r move at the same velocity v, which is
related to J by the relation [see Eq. (17.16)]
v=
J
.
qn
(17.316)
Show that in steady state this assumption leads to the 2nd London equation [see Eq. (17.62)]
2 H =
1
H,
2L
where H is the magnetic field and where

:
mc2
L =
.
4nq 2
(17.317)
(17.318)
3. Consider the so-called gradiometer RF SQUID seen in Fig. 17.4. The

junctions critical current is labeled by Ic . It is assumed that the junction has capacitance, which is denoted by CJ . Consider the case where a
magnetic flux that is denoted by e1 (e2 ) is externally applied to the upper (lower) loop. Let 1 (2 ) be the self inductance of the upper (lower)
loop. Derive an equation of motion for the system.
4. Consider the so-called DC SQUID device seen in Fig. 17.5. The Josephson junctions on both arms of the DC SQUID have critical currents Ic1
and Ic2 respectively and both have the same capacitance CJ . The self
inductance of the loop is denoted as . A bias current Ib is externally
injected and a magnetic flux e is the externally applied to the loop. Find
equations of motion that govern the devices dynamics.
Eyal Buks
528
17.6. Problems
Fig. 17.4. Gradiometer RF SQUID.
Fig. 17.5. The DC SQUID.
Eyal Buks
529

'
(
2
5. Calculate nP and (nP ) with respect to the BCS ground state |0 ,
where nP is the pairs number operator (17.264).
17.7 Solutions
1. In classical mechanics a mass particle in a rotating frame experiences
a force perpendicular to its velocity called the Coriolis force. For the
present case the Coriolis force F is taken to be given by
F = 2ms v ,
(17.319)
where =
z is the rotation vector and where v = r is the velocity
vector. Additional force perpendicular to the velocity, which is acting in
q
the presence of a magnetic field B, is the Lorentz force FL = cs v B
[see Eq. (17.4)]. From this point of view the effect of rotation can be taken
into account by replacing the magnetic field B by an effective magnetic
field Beff given by
Beff = B +
2ms c
.
qs
(17.320)
With this approach Eq. (17.62) (for time independent B) becomes

2ms c
1
2
B= 2 B+ .
(17.321)
qs
L
Thus the magnetic field deep inside the superconductor is given by
(2ms c/qs ) .
2. The total energy of the system in steady state is given by E = T + UH ,
where

nmv2
T =
dV
2
V
is the kinetic energy and where

1
UH =
H2 dV
8 V
(17.322)
is the magnetic energy [see Eq. (14.38)]. With the help of the Maxwells
equation (17.18) and Eq. (17.16) E can be expressed in terms of H as
3
4
1
E=
2L ( H)2 + H2 dV .
(17.323)
8 V
Let H be an infinitesimally small change in H, and let E be the corresponding change in the energy. The requirement that E obtains a minimum value leads to
Eyal Buks
530
17.7. Solutions
0 = E =
1
4
, 2
L ( H) ( H) + H H dV .
(17.324)
With the help of the general vector identity [see Eq. (14.41)]
(F1 F2 ) = ( F1 ) F2 F1 ( F2 ) ,
(17.325)
one finds (for the case where F1 and F2 are taken to be given by F1 =
H and F2 = H) that
( H) ( H) = ( ( H)) H ( H H) .
(17.326)
The vector identity ( H) = ( H) 2 H together with the
Maxwells equation (17.21) lead to

( H) ( H) = 2 H H ( H H) . (17.327)
The volume integral over the second term on the right hand side can be
expressed as a surface integral using the divergence theorem. However,
when boundary conditions of H = 0 on the surfaces are applied the
surface integral vanishes. Thus Eq. (17.324) becomes

2 2

1
0 = E =
L H + H H dV .
(17.328)
4 V
The requirement that E vanishes for arbitrary (small) H leads to Eq.

(17.317). The assumption that B = H [see Eq. (17.26)], i.e. the assumption that the medium is isotropic and linear, implies that in steady state
Eq. (17.317) is equivalent to the 2nd London equation (17.62).
3. The total magnetic flux 1 (2 ) threading the upper (lower) loop is given
by [see Eq. (17.110) and Fig. 17.6]
1 = e1 + 1 Is1 ,
2 = e2 + 2 Is2 ,
(17.329)
(17.330)
where 1 (2 ) is the total magnetic flux in the upper (lower) loop, Is1 (Is2 )
is the circulating current flowing in the upper (lower) loop and mutual
inductance between the loops is disregarded. The requirement that the
phase of the macroscopic wavefunction is continues in both the upper
and lower loops yields the following relations [see Eq. (17.107)]
21
= 2n1 ,
s
22
+
= 2n2 ,
s
+
(17.331)
(17.332)
where is the gauge invariant phase difference across the junction, s is

the flux quantum and where both n1 and n2 are integers. The Lagrangian
Eyal Buks
531

of the system can be expressed as a function of the dimensionless flux
coordinates , which is defined by [see Eqs. (17.331) and (17.332)]

1
2
= 2
n1 = 2
n2 = ,
(17.333)
s
s
and its time derivative [see Eq. (17.117)]
L=
CJ 2s 2 2s ( e1 )2 2s ( + e2 )2 s Ic
cos , (17.334)
+
82 c2
8 2 c1
82 c2
2c
where
2e1
2n1 ,
s
2e2
=
2n2 ,
s
e1 =
(17.335)
e2
(17.336)
are the normalized external fluxes. Using the notation

1
1
1
=
+
,
0
1 2
0 e1 0 e2
e0 =
,
1
2
(17.337)
(17.338)
the Lagrangian can be expressed as

L=
CJ 2s 2 2s ( e0 )2 s Ic
+
cos + CG ,
82 c2
82 c0
2c
where the constant CG is given by

0 2
2
2
0
2s
1 e1 + 2 e2 e0
CG =
.
8 2 c0
(17.339)
(17.340)
The resulting Euler - Lagrange equation of motion (1.8) is given by

d L
L
=
,
(17.341)
dt
thus
CJ s
( e0 )
= s
Ic sin .
2c
20
(17.342)
Note that [see Eqs. (17.329) and (17.330)]

e0 =
Eyal Buks
20 (Is1 Is2 )
,
s
(17.343)
532
17.7. Solutions
Fig. 17.6. Gradiometer RF SQUID.
thus with the help of Eqs. (17.97) and (17.333) the equation of motion
(17.342) can be expressed as a current conservation law [compare with
Eq. (17.125)]
Is1 Is2 = Ic sin + CJ V ,
(17.344)
where V is the voltage across the Josephson junction.

4. The total self inductance of the loop (which is assumed to be equally
divided between both arms) is taken into account by adding an inductor
having inductance /2 to each arm (see Fig 17.7). The total magnetic
flux threading the DC SQUID loop is given by
= e +
Is
,
2
(17.345)
where e is the externally applied flux,

Is = I1 I2
(17.346)
is the circulating current in the loop, and where I1 and I2 are the total currents flowing in the upper and lower arms respectively. The bias
current is given by Ib = I1 + I2 . In general the Josephson current IJk in
each junction (k = 1, 2) is related to the critical current Ick and to the
Josephson phase k by [see Eq. (17.92)]
Eyal Buks
533

IJk = Ick sin k .
(17.347)
The voltage across each junction is given by [see (17.97)]

Vk =
s
,
2c k
(17.348)
where s is the flux quantum [see Eq. (17.80)]. The condition that the
phase around the loop is single-valued reads
2n = 1 2 +
2
,
s
(17.349)
where n is integer [see Eq. (17.107)]. By using this and Eq. (17.345) one
finds that

2e
2n .
(17.350)
Is = s 1 2 +
s
The total voltage across the DC SQUID, which is denoted as VS is given
by
VS = V1 +
dI1
dI2
= V2 +
,
2c dt
2c dt
and therefore

1
dI1 dI2
VS =
V1 + V2 +
+
2
2c dt
dt

1 d s
Ib
=
( + 2 ) +
.
2 dt 2 1
2c
(17.351)
(17.352)
To model the current biasing an inductor having inductance LB is added

in parallel to the SQUID (see Fig. 17.7). As will be shown below, in the
limit where LB the added inductor allows modeling current biasing.
The requirement that the voltage across the DC SQUID is the same as
the one across the inductor LB yields [see Eq. (17.352)]

1 d s
Ib
dB
=
( 1 + 2 ) +
,
(17.353)
dt
2 dt 2
2c
where B is the magnetic flux in the inductor LB . In what follows the bias
current Ib is assumed to be a constant. Integrating the above equation
yields
B =
Eyal Buks
s
( + 1 ) + B0 ,
4 1
(17.354)
534
17.7. Solutions
where the constant B0 is taken to be given by B0 = LB Ib . The
Lagrangian of the closed system L = T U [see Eq. (1.16)] is expresses
as a function of the coordinates 1 and 2 and their time derivatives 1
and 2 , where the kinetic energy T is given by

CJ 2s 21 + 22
T =
,
(17.355)
8 2 c2
and the potential energy U by [see Eq. (17.350)]
s (Ic1 cos 1 + Ic2 cos 2 )
2c

2
e
2s 1 2 + 2
2n
U =
+
82 c
2B
.
2cLB
(17.356)
The assumption that LB together with the assumption that

s ( 1 + 1 ) /4 LB Ib [see Eq. (17.354)] implies that the term 2B /2cLB
can be replaced by s ( 1 + 2 ) Ib /4c (note that constant terms in U
can be disregarded). The Euler-Lagrange set of two equations [see Eq.
(1.8)]

L
d L
=
,
(17.357)
dt 1
1

d L
L
=
,
(17.358)
dt 2
2
yields
CJ 2s 1
4 2 c2
CJ 2s 2
4 2 c2
s Ic1 sin 1
2c

2s 1 2 +
2e
s
s Ic2 sin 2
+
2c

2s 1 2 +
2e
s
4 2 c
4 2 c

2n

2n
s Ib
,
4c
(17.359)
s Ib
,
4c
(17.360)
or [see Eqs. (17.348) and (17.350)]

dV1
= I1 ,
dt
dV2
Ic2 sin 2 + CJ
= I2 .
dt
Ic1 sin 1 + CJ
Eyal Buks
(17.361)
(17.362)
535
Fig. 17.7. The DC SQUID. The inductor LB is added to model current biasing.
5. The following holds [see Eqs. (17.245) and (17.247)]

0 | ak , ak , |0
#
= 0|
Kk Kk Kk ak , ak , Kk |0
k
=k

= 0| eik cos k eik sin k ak , ak , ak , ak , eik cos k eik sin k ak , ak , |0
= sin2 k 0| ak , ak , ak , ak , ak , ak , |0
= sin2 k ,
(17.363)
Eyal Buks
536
17.7. Solutions
thus
0 | nP |0 =
sin2 k .
(17.364)

2
Similarly, since ak , ak , = ak , ak , one finds that
0 | n2P |0 =

1
0 | ak , ak , + ak , ak , ak , ak , + ak , ak , |0
4
k ,k
1
0 | ak , ak , + ak , ak , + 2ak , ak , ak , ak , |0

1
+
0 | ak , ak , + ak , ak , ak , ak , + ak , ak , |0
4
k
=k

1 2
sin k + sin4 k +
sin2 k sin2 k ,
=
2
k
=k
(17.365)
thus
0 | (nP )2 |0 = 0 | n2P |0 (0 | nP |0 )2

1 2
=
sin k + sin4 k +
sin2 k sin2 k
sin2 k sin2 k
2
k
k
=k
k ,k

1 2
=
sin k 1 sin2 k
2
k
1 2
=
sin k cos2 k .
2
k
Eyal Buks
537
18. Open Quantum Systems
This chapter is mainly based on the book [7].
18.1 Classical Resonator

Consider a classical mechanical resonator having mass m and resonance frequency 0 . The resonator is driven by an external force Fex that is given
by

Fex = F0 cos ( p t) = F0 Re eip t ,
(18.1)
where F0 is a real constant. The equation of motion is given by
m
x + m20 x = Fex .
In steady state we seek a solution having the form

x = Re Aeip t ,
(18.2)
(18.3)
where A is a complex constant. Substituting such a solution into the equation

of motion (18.2) yields
A=
1
F0
.
2
m 0 2p
(18.4)
This result is clearly nonphysical since it diverges at resonance p = 0 . This

can be fixed by introducing a damping term in the equation of motion
m
x + m x + m 20 x = Fex ,
(18.5)
where is the damping rate. For this case the steady state amplitude becomes
finite for any driving frequency
A=
F0
1
.
m 20 2p ip
(18.6)
However, also (18.5) is a nonphysical equation of motion. The equipartition

theorem of classical statistical mechanics predicts that in equilibrium at temperature T the following holds
Chapter 18. Open Quantum Systems

) 2 * kB T
x =
.
m20
(18.7)
However, as can be seen from Eq. (18.5), when F0 = 0 the steady state
solution is given by x (t) = 0, contradicting thus the equipartition theorem.
This can be fixes by introducing yet another term f (t) in the equation of
motion representing fluctuating force
m
x + m x + m 20 x = f (t) + Fex .
(18.8)
The fluctuating
)
* force has vanishing mean f (t) = 0, however its variance
is finite f 2 (t) > 0. In exercise 1 below the autocorrelation function of the
fluctuating force f (t) is found to be given by (18.153)
f (t) f (t + t ) = 2mkB T (t ) .
(18.9)
Similarly to the classical case, also in the quantum case nonphysical behavior is obtained when damping is disregarded. This happens not only for
the above discussed example of a driven resonator. For example, recall that
for a general quantum system driven by a periodic perturbation the time
dependent perturbation theory predicts in the long time limit constant rates
of transition between states [e.g., see Eq. (10.38)]. Such a prediction can
yield correct steady state population of quantum states only when damping
is taken into account.
Damping and fluctuation in a quantum system can be taken into account
by introducing a thermal bath, which is assumed to be weakly coupled to the
system under study. Below this technique is demonstrated for two cases. In
the first one, the system under study (also referred to as the closed system)
is a mechanical resonator, and in the second one it is taken to be a two level
system. In both cases the open system is modeled by assuming that the closed
system is coupled to a thermal bath in thermal equilibrium.
18.2 Quantum Resonator Coupled to Thermal Bath

Consider a mechanical resonator having mass m and resonance frequency 0 .
The resonator is coupled to a thermal bath, which is modeled as an ensemble
of harmonic oscillators.
18.2.1 The closed System
First, we consider the isolated resonator. The Hamiltonian is given by [see
Eqs. (5.9), (5.10), (5.11), (5.12), (5.13) and (5.16)]
Eyal Buks
540
18.2. Quantum Resonator Coupled to Thermal Bath

p2
1
+ m20 x2
2m 2

1
,
= 0 a a +
2
H0 =
where
(18.10)

ip
a=
x+
,
m 0
0

m0
ip
x
,
a =
2
m 0
m0
2
and where
, a, a = 1 .
(18.11)
(18.12)
(18.13)
The inverse transformation is

0

x=
a +a ,
2m 0
0

m 0
p=i
a a .
2
(18.14)
(18.15)
18.2.2 Coupling to Thermal Bath

Damping is taken into account using a model containing a reservoir of harmonic oscillators interacting with the resonator. The total Hamiltonian is
given by
Ht = H0 + Hr + V ,
(18.16)
where H0 is given by Eq. (18.10), Hr is the Hamiltonian of the thermal bath,

which is assumed to be a dense ensemble of harmonic oscillators

1
k bk bk +
Hr =
,
(18.17)
2
k
and V is a coupling term

V = a
k bk + a
k bk ,
k
(18.18)
where k are coupling constants. The bath operators satisfy regular harmonic
oscillator commutation relations
3
4 ,
4
- 3
[a, bk ] = a, bk = a , bk = a , bk = 0 ,
(18.19)
Eyal Buks
541
and
3
4
bk , bl = k,l ,
(18.20)
3
4
[bk , bl ] = bk , bl = 0 .
(18.21)
a + (i 0 + ) a = F (t) ,
(18.22)
where is a constant and where the fluctuating term F (t) is given by

k exp (i k t) bk (0) .
(18.23)
F (t) = i
k
Solution 18.2.1. In general, the Heisenberg equation of motion of an operator O is given by Eq. (4.37)
i
O
O = [O, H] +
.

t
(18.24)

a = i0 a i
k bk ,
(18.25)
k bk ,
(18.26)
a = i0 a + i

k
and
b k = i k bk ik a ,
(18.27)
b k = ik bk + ik a .
(18.28)
The solution of Eq. (18.27) is given by

bk (t) = exp [i k (t t0 )] bk (t0 )
t
ik
dt exp [i k (t t )] a (t ) .
t0
(18.29)
Choosing the initial time to be given by t0 = 0 and substituting Eq. (18.29)

into Eq. (18.25) yield
t

a + i 0 a +
dt a (t )
|k |2 exp [ik (t t )]
0
= i
Eyal Buks
exp (i k t) bk (0) .
(18.30)
542

The states of the thermal bath are assumed to be very dense, thus one can
replace the sum over k with an integral

2
|k | exp [i k (t t )]
k
d | ()| exp [i (t t )] ,
(18.31)
where () is the density of states. Assuming () is a smooth function

near = 0 one finds that
t

dt a (t )
|k |2 exp [ik (t t )]
0
dt a (t ) | ( 0 )|
2
= | ( 0 )| a (t) .
d exp [i (t t )]

2(tt )
(18.32)
Thus using the notation

= | ( 0 )|2 ,
(18.33)
one has
a + (i 0 + ) a = F (t) ,
(18.34)
a + (i 0 + ) a = F (t) ,
(18.35)
where
F (t) = i
F (t) = i
k exp (i k t) bk (0) ,
(18.36)
k exp (i k t) bk (0) .
(18.37)
The fluctuation terms F (t) and F (t) represent noisy force acting on the
resonator.
From Eqs. (18.34), (18.35), (18.14), and (18.15) one finds that
p + p + m 20 x = f (t) ,
(18.38)
where
Eyal Buks
543

0
f (t) = i
m 0 ,
F (t) F (t) .
2
(18.39)
In classical mechanics the momentum p is given by p = mx.

Using this
substitution the equation of motion for the quantum operator p (18.38) takes
a form analogues to the classical equation of motion of a mechanical resonator
having damping rate and influenced by a force f (t)
m
x + m x + m 20 x = f (t) .
(18.40)
18.2.3 Thermal Equilibrium

)
*
0 (t ) ,
F (t) F (t + t ) = 2 n
)
*
n0 + 1) (t ) ,
F (t) F (t + t ) = 2 (
(18.41)
(18.42)
and
)
*
F (t) F (t + t ) = F (t) F (t + t ) = 0 ,
(18.43)
where
n
0 =
1
e0
(18.44)
and where = 1/kB T .

Solution 18.2.2. The modes of the thermal bath are assumed to be in thermal equilibrium. In general, thermal averaging of an operator Ok , associated
with mode #k in the thermal bath, is given by [see Eqs. (8.8) and (8.42)]
Ok = Tr (k Ok ) ,
(18.45)
where the density operator k is given by

k =
1 Hr,k
,
e
Z
(18.46)
where

Z = Tr eHr,k ,

1
,
Hr,k = k bk bk +
2
(18.47)
(18.48)
and = 1/kB T . Using these expressions one finds that [see Eq. (8.152)]
'
(
1
(18.49)
bk (t) bk (t) = k
n
k .
e
1
Eyal Buks
544

Using Eq. (18.20) one finds that
'
(
bk (t) bk (t) = n
k + 1 .
In a similar way one also finds that

' ( ) * ' (
bk = bk = b2k = b2
=0.
k
(18.50)
(18.51)
Moreover, using the full bath Hamiltonian Hr one can easily show that
'
(
bk bl = bk bl = 0 ,
(18.52)
'
(
bk (t) bl (t) = kl n
k ,
(18.53)
and
'
(
nk + 1) .
bk (t) bl (t) = kl (
(18.54)
The fluctuating forces are given by Eqs. (18.36) and (18.37). We calculate
below some correlation functions of these forces. Using Eq. (18.51) one finds
)
*
F (t) = F (t) = 0 .
(18.55)
Using Eq. (18.53) one finds that
*
)
F (t) F (t + t ) =
|k |2 exp (i k t ) n
k .
(18.56)
Replacing the sum over k with an integral, as in Eq. (18.31), and taking into
account only modes that are nearly resonant with the cavity mode one finds
*
)
F (t) F (t + t ) = 2 n
0 (t ) ,
(18.57)
where
n
0 =
1
.
e0 1
Similarly
*
)
n0 + 1) (t ) ,
F (t) F (t + t ) = 2 (
(18.58)
(18.59)
and
)
*
F (t) F (t + t ) = F (t) F (t + t ) = 0 .
(18.60)
) *
Exercise 18.2.3. Show that the expectation value a a in steady state is
given by
) *
a a =n
0 .
(18.61)
Eyal Buks
545

Solution 18.2.3. Multiplying Eq. (18.34) by the integration factor e(i0 +)t
yields
d (i0 +)t
= F (t) e(i0 +)t .
ae
dt
(18.62)
(i0 +)(t0 t)
a (t) = a (t0 ) e
dt F (t ) e(i0 +)(t t) .
(18.63)
t0
Steady state is established when (t t0 ) 1. In this limit the first term

becomes exponentially small (recall that is positive), i.e. effect of initial
condition on the value of a at time t0 becomes negligible. Thus in steady
state the solution becomes
t
a (t) =
dt F (t ) e(i0 +)(t t) ,
(18.64)
t0
and the Hermitian conjugate is given by

t
a (t) =
dt F (t ) e(i0 +)(t t) .
(18.65)
t0

t
t
*
) *
)
a a =
dt
dt F (t ) F (t ) e(i0 +)(t t) e(i0 +)(t t)
t0
t0
t
= 2
n0
dt e2 (t t)
t
0

=n
0 1 e2(tt0 ) .
(18.66)
The assumption (t t0 ) 1 allows writing this result as

) *
a a =n
0 .
(18.67)
) *
The last result a a = n
0 verifies that the resonator reached thermal
equilibrium in steady state. Similarly, the next exercise shows that in the
classical limit the equipartition theorem of classical statistical mechanics is
satisfied.
) *
Exercise 18.2.4. Calculate x2 in steady state.
,
Solution 18.2.4. According to Eq. (18.14) and a, a = 1 the following
holds
Eyal Buks
546
18.3. Two Level System Coupled to Thermal Bath

) 2*
x =

2m0

=
2m0

=
2m0
)

*
a + a a + a
)
)
a2 + a2 + a a + aa
*
a2 + a2 + 2a a + 1 .
(18.68)
) 2 * ) 2 *
As can be seen from Eq. (18.60), a
= a = 0. Thus, with the help of
Eq. (18.67) one has
) 2*

x =
(2
n0 + 1)
2m0
0

=
coth
,
2m0
2
(18.69)
in agreement with Eq. (8.160). In the classical limit where kB T 0 one
has
) 2 * kB T
x =
,
(18.70)
m20
in agreement with the classical equipartition theorem.
18.3 Two Level System Coupled to Thermal Bath

In this section we discuss a two level system (TLS) coupled to thermal baths,
and obtain the Bloch equations.
18.3.1 The Closed System
The Hamiltonian Hq of the closed system can be represented by a 22 matrix
[see Eq. (15.36)], which in turn can be expressed in terms of Pauli matrices
(6.136)

Hq =
(t) ,
2
(18.71)
where (t) is a 3D real vector, and where the components of the Pauli matrix
vector are given by

01
0 i
1 0
x =
, y =
, z =
.
(18.72)
10
i 0
0 1
Let P = be the vector of expectation values P = ( x , y , z ). We
refer to this vector as the polarization vector. With the help of Eq. (4.38),
which is given by
Eyal Buks
547

d A
1
=
[A, H] +
dt
i
A
t
(18.73)

dPz
1
=
[ z , Hq ]
dt
i
1
=
(x [ z , x ] + y [z , y ])
2i
= (x y y x )
= ( P)
z.
(18.74)
Similar expressions are obtained for Px and Py that together can be written
in a vector form as [see also Eq. (6.162)]
dP
= (t) P .
dt
(18.75)
The time varying effective magnetic field (t) is taken to be given by

(t) = 0
z + 1 (t) .
(18.76)
While 0 , which is related to the energy gap separating the TLS states by
0 = /, is assumed to be stationary, the vector 1 (t) is allowed to vary
in time, however, it is assumed that | 1 (t)| 0 .
18.3.2 Coupling to Thermal Baths
As we did in the previous section, damping is taken into account using a model
containing reservoirs having dense spectrum of oscillator modes interacting
with the TLS. Furthermore, since the ensembles are assumed to be dense,
summation over modes is done with continuos integrals. The Hamiltonian H
of the entire system is taken to be given by
H = Hq

+ d a1 () a1 ()

+ d a2 () a2 ()
0

1 i1
+ d
e + a1 () + ei1 a1 ()
2
0

i2
+ d
e z a2 () + ei2 a2 () z ,
4
Eyal Buks
(18.77)
548

where Hq is the Hamiltonian for the closed TLS,

01
00
+ =
, =
,
00
10
(18.78)
and the real coupling parameters 1 , , 1 and 2 are assumed to be frequency independent. The bath modes are boson modes satisfying the usual
Bose commutation relations
3
4
ai () , ai ( ) = ( ) ,
(18.79)
[ai () , ai ( )] = 0 ,
(18.80)
where i = 1, 2. While the coupling to the first bath (with coupling constant
1 ) gives rise to TLS decay through spin flips, the coupling to the second
bath (with coupling constant ) gives rise to pure dephasing.
dz
1
=
[z , Hq ] 1 (1 + z )
dt
i

2
+
i+ V1 + iV1 ,
(18.81)

1
1
d+
=
[+ , Hq ]
+ +
dt
i
2

4
i3
+
V1 z + 2 + V + V ,
(18.82)
and
where
0
V1 =
1 i1
e
2
dei(tt0 ) a1 (t0 , ) ,
(18.83)
dei(tt0 ) a2 (t0 , ) .
(18.84)
and
0
V =
i2
e
4
Solution 18.3.1. With the help of the identities

[z , + ] = 2 + ,
[z , ] = 2 ,
[+ , ] = z ,
Eyal Buks
(18.85)
(18.86)
(18.87)
549

one finds that the Heisenberg equation of motion (4.38) for z is given by
dz
1
=
[z , Hq ]
dt
i0

1
d ei1 + a1 ()
2i
2
0
1
+ 2i
d ei1 a1 () ,
2
(18.88)
for + by
1
d+
=
[+ , Hq ]
dt
i
0
1
d ei1 a1 () z
i
2
0
+ 2i
d ei2 + a2 ()
4
0
+ 2i
d ei2 a2 () + ,
4
(18.89)
for a1 () by
da1 ()
= ia1 () i
dt
1 i1
e
,
2
(18.90)
i2
z .
e
4
(18.91)
and for a2 () by
da2 ()
= ia2 () i
dt
Integrating the equations of motion for the bath operators a1 () and a2 ()

yields
a1 () = ei(tt0 ) a1 (t0 , )
0

1 i1 t i(tt )
i
e
dt e
(t ) ,
2
t0
(18.92)
and
a2 () = ei(tt0 ) a2 (t0 , )
0

i2 t i(tt )
i
e
dt e
z (t ) .
4
t0
Eyal Buks
(18.93)
550

We now substitute these results into the Eqs. (18.88) and (18.89) and make
use of the following relations
(18.94)
d ei(tt ) = 2 (t t ) ,
t0
1
(t t ) f (t ) dt = sgn (t t0 ) f (t) .
2
where sgn(x) is the sign function

+
+1 if x > 0
sgn (x) =
,
1 if x < 0.
(18.95)
(18.96)
to obtain
dz
1
=
[z , Hq ]
dt
i0

1
2i
dei1 + ei(tt0 ) a1 (t0 , )
2
1 +
0
1
dei1 ei(tt0 ) a1 (t0 , )
+ 2i
2
1 + ,
(18.97)
and
d+
1
=
[+ , Hq ]
dt
i
0
1
i
dei1 ei(tt0 ) a1 (t0 , ) z
2
1
+ + z
20
+ 2i
dei2 + ei(tt0 ) a2 (t0 , )
4
+
+ z
20
+ 2i
dei2 ei(tt0 ) a2 (t0 , ) +
4
z + .
2
(18.98)
Thus, by making use of the following relations
Eyal Buks
551

1
(1 + z ) ,
2
1
+ = (1 z ) ,
2
z + = + z = + ,
+ =
one finds that

dz
1
=
[ z , Hq ] 1 (1 + z )
dt
i

2
+
i+ V1 + iV1 ,

and

d+
1
1
=
[ + , Hq ]
+ +
dt
i
2

4
i3
+ V1 z + 2 + V + V + .
(18.99)
(18.100)
(18.101)
(18.102)
(18.103)
18.3.3 Thermal Equilibrium

' (
) *
V1 = V1 = V = V = 0 .
Using Eqs. (18.53) and (18.54), the relation

d ei(tt ) = 2 (t t ) ,
(18.104)
(18.105)
and assuming the case where the dominant contribution to the TLS dynamics
comes from the bath modes near frequency 0 (recall that 0 = /, where
is the energy gap separating the TLS states), one finds that
'
(
V1 (t ) V1 (t)

'
(
1
d d ei (tt ) a1 (t0 , ) a1 (t0 , )
= 2
2
1
= 2
dei(tt ) n ()
2
2 1 n
0 (t t ) ,
(18.106)
where n
0 is given by [see Eq. (18.58)]
Eyal Buks
552

n
0 =
1
.
e0 1
(18.107)
Similarly
'
(
V1 (t) V1 (t ) = 2 1 (
n0 + 1) (t t ) ,
and
)
*
V (t ) V (t) = 2 n
0 (t t ) ,
2
)
*
(
n0 + 1) (t t ) ,
V (t) V (t ) = 2
2
'
(
V1 (t ) V1 (t) = V1 (t ) V1 (t)
)
*
V (t ) V (t) = V (t ) V (t) = 0 .
(18.108)
(18.109)
(18.110)
(18.111)
18.3.4 Correlation Functions

Equation of motion for the polarization vector P can be obtained by taking the expectation value of Eqs. (18.102) and (18.103). Using the identity = (1/2) (x iy ) and the notation P = (1/2) (Px iPy ) and
u
= (1/2) (
x i
y) one finds that
Pz = ( (t) P)z 1 (1 + Pz )
'
(
2
+
i + V1 + i V1
,
and

1
+ P+
2
(

)
*4
i3 '
+ V1 z + 2 + V + V +
,
P+ = ( (t) P)+
(18.112)
(18.113)
where the subscripts z and + denote the components of the vector (t) P
in the
z and u
+ directions respectively. However, Eqs. (18.112) and (18.113)
contain product terms between bath operators and TLS operators (e.g. the
term + V1 in Eq. (18.112). To lowest order such terms can be evaluated
by assuming that these operators are independent, e.g. + V1 + V1 .
However, this approach, which yields vanishing contribution of all such terms
is too crude. Below we employ a better approximation to evaluate the expectation value of such terms. In the first step Eqs. (18.102) and (18.103) are
formally integrated. This yields the following results
Eyal Buks
553

z (t) = 1 + (1 + z (0)) e1 t

t

2
1
[ z , Hq ] +
i+ V1 + iV1 e1 (t t) ,
+
dt
i

0
(18.114)
and
1
+ (t) = + (0) e( 2 + )t

t
( 1 + )(t t)

1
i
e 2
+
dt
[ + , Hq ] +
V1 z + 2 + V + V +
.
i

0
(18.115)
In the second step these expressions for the TLS operators are substituted
into Eqs. (18.112) and (18.113). In this final step, correlations are disregarded
(e.g. the expectation value
form + V1 V1 is evaluated
' of a term
( having
' the (
using the approximation + V1 V1 + V1 V1 ). The expectation values
of bath operators are calculated with the help of the results of the previous
section. This approach yields the following results

(
1 t ( 21 + )(t t) '
dt e
+ V1 =
V1 (t ) V1 (t) z (t )
i 0
i1 n
0
=
Pz ,
2
(18.116)
'
( i n
1 0
V1 =
Pz ,
(18.117)
2
'
(
V1 z = i1 n
0 P+ ,
(18.118)
and
)
*
+ V + V + = i n
0 P+ ,
(18.119)
thus
Pz = ( (t) P)z 1 [1 + (2
n0 + 1) Pz ] ,
and
P+ = ( (t) P)+

1
+ (2
n0 + 1) P+ .
2
(18.120)
(18.121)
A similar equation can be obtained for P , which together with Eq. (18.121)
can be written as

1
Px = ( (t) P)x
+ (2
n0 + 1) Px ,
(18.122)
2

1
Py = ( (t) P)y
+ (2
n0 + 1) Py .
(18.123)
2
Eyal Buks
554
18.4. Problems
18.3.5 The Bloch Equations
z [see Eq. (18.76)]. For
Consider the case where 1 (t) = 0, i.e. (t) = 0
this case Eqs. (18.120) and (18.121) become
Pz = 1 [1 + (2
n0 + 1) Pz ] ,

1
+ (2
n0 + 1) P .
P = i0
2
(18.124)
(18.125)
In the long time limit the solution is given by P (t ) = 0 and

Pz (t ) = Pz0 , where [see Eq. (18.58)]

1
0
= tanh
.
(18.126)
Pz0 =
(2
n0 + 1)
2
Note that Eq. (18.126) is in agreement with the Boltzmann distribution law
of statistical mechanics, according to which in thermal equilibrium the probability to occupy a state having energy is proportional to exp () (recall
that Pz is the probability to occupy the upper state of the TLS minus the
probability to occupy the lower one). In terms of the decay times T1 and T2 ,
which are defined by
T1 = 11 (2
n0 + 1)1 ,

1
1
T2 =
(2
n0 + 1)1 ,
+
2
(18.127)
(18.128)
the equations of motion for the general case, which are known as optical
Bloch equations, are given by
Px
Px = ( (t) P)x
,
T2
Py
Py = ( (t) P)y
,
T2
Pz Pz0
P z = ( (t) P)z
.
T1
(18.129)
(18.130)
(18.131)
18.4 Problems
1. Calculate the autocorrelation function f (t) f (t + t ) of the classical
fluctuating force f (t), which was introduced into the classical equation
of motion (18.8) of a mechanical resonator. The autocorrelation function
should yield a result consisting with the equipartition theorem.
2. Calculate the autocorrelation function f (t) f (t + t ), where the quantum operator f (t) is given by Eq. (18.39).
Eyal Buks
555

3. Consider a one-dimensional mechanical resonator having mass m, resonance frequency 0 and damping rate in thermal equilibrium at temperature T . Calculate the expectation value of the autocorrelation function
g () of the resonators coordinate x, which is defined by
g ( ) =
1
x (t) x (t + ) + x (t + ) x (t) ,
2
(18.132)
in steady state.
4. Consider a TLS having energy gap . A perturbation, which is externally
applied, induces transitions between the states having rate T . Calculate
the polarization vector P in steady state.
5. Magnetic resonance - A time dependent magnetic field given by
B (t) = B0
z + B1 (
x cos (t) + y
sin (t))
(18.133)
is applied to a spin 1/2 particle.

a) Use the Bloch equations to determine the polarization Pz in steady
state.
b) The polarization Px in steady state can be expressed as
Px = 2 1 [cos (t) () + sin (t) ()] ,
(18.134)
where () and () are respectively the real and imaginary parts

of the magnetic susceptibility () (i.e. () = () + i ()).
Note that the term proportional to () is in phase with respect
to the driving magnetic filed in the x direction [recall that Bx =
B1 cos (t)], whereas the second term, which is proportional to ()
is out of phase [i.e. proportional to sin (t)] with respect to Bx .
Calculate ().
18.5 Solutions
1. In the absence of any externally applied driving force, i.e. when Fex = 0,
the classical equation of motion is given by (18.8)
m
x + m x + m20 x = f (t) ,
(18.135)
where f (t) represents a random force acting on the resonator due to the
coupling with the thermal bath at temperature T . Bellow we consider
statistical properties of the fluctuating function x(t). However, since some
of the quantities we define may diverge, we consider a sampling of the
function x(t) in the finite time interval ( /2, /2), namely
+
x(t) /2 < t < /2
x (t) =
.
(18.136)
0
else
Eyal Buks
556
18.5. Solutions
The equipartition theorem requires that
) * 1
1
m 20 x2 = kB T ,
2
2
(18.137)
where
) 2*
1
x lim

dt x2 (t) .
Introducing the Fourier transform (FT)

1
x (t) =
d x ()eit ,
2
(18.138)
(18.139)
one finds that

) 2*
1
x =
2
d x ()
1
lim

= lim

d x ( )
dtei(+ )t

2(+ )
d x ()x () .
(18.140)
Since x(t) is real x () = x (). In terms of the power spectrum Sx (),

which is defined as
1
|x ()|2 ,
Sx () = lim
(18.141)
one has
) 2*
x =
d Sx () .
(18.142)
Next, we take the FT of Eq. (18.135)

m 2 im + m20 x() = f() ,
(18.143)
where
1
f(t) =
2
df ()eit .
(18.144)
Taking the absolute value squared yields

Sx () =
Eyal Buks
m2
Sf ()
3
4 .
2
2
() + ( 20 2 )
(18.145)
557

Integrating Eq. (18.145) leads to

1
Sf ()
d Sx () = 2
d
2 .
2
m
() + ( 20 2 )
(18.146)
Assuming that in the vicinity of 0 , i.e. near the peak of the integrand,
the spectral density Sf () is a smooth function on the scale of the width
of the peak , and also assuming that 0 , one approximately finds
that

1
d
d Sx () Sf ( 0 ) 2
2
2
2
m
() + ( 0 2 )

Sf ( 0 )
d
=
3
2
2
m 0 (/ 0 ) + (1 2 )2

Sf ( 0 )
d
3
2
m 0 (/ 0 )2 + 1

0 /
= 2 2 Sf (0 ) .
m 0
(18.147)
This together with Eqs. (18.137) and (18.142) yields
Sf (0 ) =
mkB T
,
(18.148)
thus Eq. (18.145) becomes

Sx () =
kB T
1
.
m ()2 + ( 20 2 )2
(18.149)
At the peak = 0 one finds

Sx (0 ) =
kB T
.
m 20
(18.150)
The correlation function C (t ) of the fluctuating force f is defined as:

1
C (t ) = f (t) f (t + t ) lim

dt f (t) f (t + t ) . (18.151)
Eyal Buks
558
18.5. Solutions

1
1 +
lim
dt
df()eit
2
i (t+t )
d f ( )e
1
1
=
lim
deit f ()
2

+
d f ( )
dtei(+ )t

C (t ) =
2(+ )
1
= lim
deit f()f()

deit Sf () .
=
(18.152)
Using Eq. (18.148) and assuming as before that Sf () is smooth function

near = 0 allow determining the coefficient C (t )
mkB T
C (t ) =
deit = 2mkB T (t ) .
(18.153)

2(t )
2. Using the definition (18.39) and Eqs. (18.57), (18.59) and (18.60) one has
m0
f (t) f (t + t ) =
-,
-*
), 2
F (t) F (t) F (t + t ) F (t + t )
= m0 (2
n0 + 1) (t )
e0 + 1
(t )
e0 1

0
= m0 coth
(t ) .
2
= m0
(18.154)
In the classical limit where kB T 0 one finds that

f (t) f (t + t ) = 2mkB T (t ) ,
(18.155)

3. With the help of Eqs. (5.11), (18.64) and (18.65) the autocorrelation
function can be expressed in terms of the noise operator F (t ) [see
Eq. (18.36)], which satisfy the correlation relations (18.57), (18.59) and
(18.60)
Eyal Buks
559

g () = Re x (t) x (t + )
)

*

=
Re a (t) + a (t) a (t + ) + a (t + )
2m
)
*

=
Re a (t) a (t + ) + a (t) a (t + )
2m
t
t
*
)
=
Re
dt
dt e(i0 +)(t t) e(i0 +)(t t ) F (t ) F (t )
2m
t0
t0
t
t
)
*

+
Re
dt
dt e(i0 +)(t t) e(i0 +)(t t ) F (t ) F (t ) .
2m
t0
t0
t
(
n0 + 1)
Re
=
dt e(i0 +)(t t) e(i0 +)(t t )
m
t0
t
n
0
Re
+
dt e(i0 +)(t t) e(i0 +)(t t ) ,
m
t0
(18.156)
n
0 =
1
,
e0 1
(18.157)
thus
1 e2(tt0 )
(2
n0 + 1)
cos ( 0 ) e
m
2
2(tt0 )
coth
1
e
2
=
cos ( 0 ) e
.
m
2
g ( ) =
(18.158)
In steady state, i.e. for (t t0 ) 1, the autocorrelation function g ( )
becomes
g ( ) =

coth
cos ( 0 ) e .
2m
2
(18.159)
4. The Bloch equation (18.131) for this case becomes

Pz Pz0
P z = T Pz
,
T1
(18.160)
thus in steady state

Pz =
Pz0
.
1 + T T1
(18.161)
Clearly, by symmetry, Px = Py = 0 in steady state.

5. The Hamiltonian of the closed system is given by
Eyal Buks
560
18.5. Solutions

Hq =
(t) ,
2
(18.162)
where
(t) = 0
z+1 (cos (t) x
+ sin (t) y
) ,
(18.163)
and where [see Eq. (4.22)]

|e| B0
,
me c
|e| B1
1 =
.
me c
0 =
(18.164)
(18.165)
In terms of the vectors u

= (1/2) (
x i
y) the vector (t) is given by
it

z+1 e
u
+ + eit u
.
(18.166)
(t) = 0
In terms of T1 , T2 and Pz0 Eqs. (18.120) and (18.121) become
Pz Pz0
P z = ( (t) P)z
,
T1
(18.167)
P+
P + = ( (t) P)+
.
T2
(18.168)
and
With the help of the identities

u ,
zu
= i
u
+ u
+ = u
u
= 0 ,
u
+ u
= i (1/2)
z,
Eqs. (18.167) and (18.168) become

i 1 eit P+ eit P
Pz Pz0
Pz =
,
2
T1
(18.169)
(18.170)
(18.171)
(18.172)
and
P+
P + = i0 P+ + i 1 eit Pz
.
T2
(18.173)
By employing the transformation into the rotating frame [see for comparison Eq. (6.240)]
P+ (t) = eit PR+ (t) ,
P (t) = eit PR (t) ,
(18.174)
(18.175)
Eqs. (18.172) and (18.173) become

Eyal Buks
561

i 1 (PR+ PR ) Pz Pz0
P z =
,
2
T1
(18.176)
PR+
.
P R+ = i ( 0 ) PR+ + i 1 Pz
T2
(18.177)
and
a) In steady state, i.e. when Pz = P R+ = 0, one has

i 1 (PR+ PR )
Pz Pz0
=
,
2
T1
PR+
i ( 0 ) PR+ + i1 Pz =
,
T2
thus (recall that PR+ = PR )

Pz =
1 + T22 ( 0 )2
Pz0 .
1 + T22 ( 0 )2 + 21 T1 T2
(18.178)
(18.179)
(18.180)
b) In steady state one has

PR+ =
iT2 1 [1 + iT2 ( 0 )]
Pz0 ,
1 + T22 ( 0 )2 + 21 T1 T2
(18.181)
iT2 1 [1 + iT2 ( 0 )]
(18.182)
thus
P+ =
1+
T22 (
0 ) +
21 T1 T2
Pz0 eit .
Using the relations

Px = 2 1 [cos (t) () + sin (t) ()]

= 1 eit () + eit () ,
(18.183)
and
Px = P+ + P ,
(18.184)
one finds that

() =
Eyal Buks
iT2 [1 + iT2 ( 0 )]
1 + T22 ( 0 )2 + 21 T1 T2
Pz0 .
(18.185)
562
References
1. Claude Cohen-Tannoudji, Bernard Diu, Franck Laloe, Quantum Mechanics, Wiley, New York (1977).
2. J.J.Sakurai, Modern Quantum Mechanics, Addison-Wesley, New York (1994).
3. L. D. Landau and L. M. Lifshitz, Quantum Mechanics Non-Relativistic Theory,
Oxford: Pergamon Press (1977).
4. L. S. Schulman, Techniques and Applications of Path Integration, , Wiley, New
York (1981).
5. M. V. Berry, Quantal phase-factors accompanying adiabatic changes, Proc. Roy.
Soc. London A 392, 45-57 (1984).
6. Alexander L. Fetter and John Dirk Walecka, Quantum Theory of Many-Particle
Systems, Mcgraw-Hill (1971).
7. Howard Carmichael, An open systems approach to quantum optics, Springer
(August 1993).
Index
action, 1
adiabatic approximation, 397
Aharonov-Bohm effect, 374
angular momentum, 149
Bogoliubov transformation, 464, 513
Bohrs magneton, 30, 72
Bohrs radius, 210
Bohr-Sommerfeld quantization rule,
358
Bose-Einstein function, 469
Boson, 446, 448
bra-vector, 17
canonically conjugate, 5
central potential, 203
chemical potential, 285
closure relation, 18
coherence length, 520
collapse postulate, 29
commutation relation, 34
commuting operators, 35
conservative system, 4
continuity equation, 353
Coulomb gauge, 415, 431
current density, 353
DC SQUID, 528
degeneracy, 24
density operator, 223
Diracs notation, 17
Drude model, 485
dual correspondence, 19
Ehrenfests theorem, 76
eigenvalue, 23
eigenvector, 23
equipartition theorem, 546
Euler-Lagrange equations, 2
expectation value, 29
Fermis golden rule, 343
Fermi-Dirac function, 469

Fermion, 446
Feynmans path integral, 373
fine-structure constant, 434
flux quantum, 377, 492, 528
fugacity, 285
gauge invariance, 377, 484
gauge transformation, 399
generalized force, 4
geometrical phase, 398
gyromagnetic ratio, 30
Hamiltons formalism, 1
Hamilton-Jacobi equations, 5
Heisenberg representation, 73
Hermitian adjoint, 22
Hydrogen atom, 208
ideal gas, 463
identical particles, 445
inner product, 15
Jaynes-Cummings model, 303
Josephson effect, 493
Josephson inductance, 496
ket-vector, 17
kinetic energy, 4
Lagrangian, 1
Larmor frequency, 72
linear vector space, 15
London Equations, 489
London penetration depth, 490
macroscopic quantum model, 489
magnetic moment, 29
matrix representation, 20
Maxwells equations, 415, 485
Meissner effect, 490
momentum representation, 54
Index
momentum wavefunction, 55
norm, 16
normal ordering, 113
number density operator, 453
number operator, 102
observable, 22
operator, 17
optical Bloch equations, 555
orbital angular momentum, 149, 157
orthogonal, 16
orthonormal basis, 16
outer product, 18
path integration, 369
Paulis exclusion principle, 449
Plancks constant, 3
plasma frequency, 488, 498
Poissons brackets, 7
position representation, 50
position wavefunction, 51
positive-definite, 38
potential energy, 4
principle of least action, 2
projector, 25
pure ensemble, 226
quantized field operator, 452
quantum bit, 499
quantum measurement, 28
quantum statistical mechanics, 227
scattering time, 485

Schr dinger equation, 3, 69
Schwartz inequality, 37, 38
second quantization, 457
semiclassical limit, 387
shell, 213
sperical harmonics, 161
spin, 29
spin 1/2, 71, 157
SQUID, 496
state vector, 15
stationary state, 71
Stern-Gerlach, 30
superconductivity, 483
symmetric ordering, 74
thermal bath, 540
thermal equilibrium, 544
Thomas-Fermi approximation, 510
Thomas-Reiche-Kuhn sum rule, 78
time evolution operator, 69
time-dependent perturbation theory,
337
tomography, 234
trace, 33
transformation function, 55
translation operator, 50
tunneling, 360
turning point, 354
uncertainty principle, 36, 233
unitary, 32
vector potential, 415, 431
radial equation, 206

reduced mass, 208
rotation, 150
Eyal Buks
Weyl transformation, 75
WKB approximation, 351
566

QM Lecture Notes PDF

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

QM Lecture Notes PDF

Enviado por

Direitos autorais:

Formatos disponíveis

Eyal Buks

Quantum Mechanics - Lecture

The dynamics of a quantum system is governed by the celebrated Schrdinger

where i = 1 and = 1.05457266 1034 J s is Plancks h-bar constant.

Hamiltons Formalism of Classical Physics . . . . . . . . . . . . . . . .

State Vectors and Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The Position and Momentum Observables . . . . . . . . . . . . . . . .

The Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Time Independent Perturbation Theory . . . . . . . . . . . . . . . . . .

Quantum Mechanics - Lecture Notes

11. WKB Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12. Path Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13. Adiabatic Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Quantum Mechanics - Lecture Notes

15. Light Matter Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

16. Identical Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Quantum Mechanics - Lecture Notes

18. Open Quantum Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Quantum Mechanics - Lecture Notes

1. Hamiltons Formalism of Classical Physics

In this chapter the Hamiltons formalism of classical physics is introduced,

1.1 Action and Lagrangian

where L is the Lagrangian function of the system. In general, the Lagrangian

Chapter 1. Hamiltons Formalism of Classical Physics

1.2 Principle of Least Action

Quantum Mechanics - Lecture Notes

1.2. Principle of Least Action

is assumed to be infinitesimally small. To lowest order in Q the change in

Integrating the second term by parts leads to

The last term vanishes since

The principle of least action implies that

Quantum Mechanics - Lecture Notes

Chapter 1. Hamiltons Formalism of Classical Physics

When the potential energy can be expressed as a function of the coordinates

Quantum Mechanics - Lecture Notes

Definition 1.3.2. The Hamiltonian of a physical system is a function of

where L is the Lagrangian.

(qn dpn pn dqn )

Chapter 1. Hamiltons Formalism of Classical Physics

Corollary 1.3.1. The following holds

Proof. Using Eqs. (1.23) and (1.24) one finds that

The last corollary implies that H is time independent provided that H

where nm are constants. Then,the Hamiltonian of the system is given by

where L = T U is the Lagrangian. The Hamiltonian is thus given by

Quantum Mechanics - Lecture Notes

1.4. Poissons Brackets

1.4 Poissons Brackets

The Poissons brackets are defined by

Quantum Mechanics - Lecture Notes

Chapter 1. Hamiltons Formalism of Classical Physics

Let r = (x, y, z) be the Cartesian coordinates of the particle.

by showing that the corresponding Euler-Lagrange equations are

Quantum Mechanics - Lecture Notes

Fig. 1.3. LC resonator.

2. Consider an LC resonator made of a capacitor having capacitance C in

4. Show that the Lagrange equations are coordinate invariant.

Quantum Mechanics - Lecture Notes

Chapter 1. Hamiltons Formalism of Classical Physics

The Hamiltonian is thus given by

Quantum Mechanics - Lecture Notes

The Hamilton-Jacobi equation for the canonically conjugate variable

Quantum Mechanics - Lecture Notes

Chapter 1. Hamiltons Formalism of Classical Physics

| R and | 0. Equality holds iff | = 0 .

The vectors | F and | F are said to be orthogonal if | = 0.

While the ket-vector | is a vector in F, the bra-vector | represents a

where | F and | F . What is the dual of the ket-vector | =

Similarly, the matrix representation of the outer product | | is given

Note that this result implies that if X = X then | X | = | X | .

|an,i an,i |an,i an,i | =