Você está na página 1de 4

Physics Letters A 373 (2009) 418–421

Contents lists available at ScienceDirect

Physics Letters A

www.elsevier.com/locate/pla

Classical and quantum motion in an inverse square potential


M. Ávila-Aoki a , C. Cisneros b , R.P. Martínez-y-Romero c , H.N. Núñez-Yépez d , A.L. Salas-Brito e,∗
a
Centro Universitario Valle de Chalco, Universidad Autónoma del Estado de México, Valle de Chalco, CP 56615, Estado de México, Mexico
b
Instituto de Ciencias Físicas, Universidad Nacional Autónoma de México, Apartado Postal 6-96, CP 62131, Cuernavaca, Morelos, Mexico
c
Facultad de Ciencias, Universidad Nacional Autónoma de México, Apartado Postal 21-267, CP 04000, Coyoacán DF, Mexico
d
Departamento Física, Universidad Autónoma Metropolitana-Iztapalapa, Apartado Postal 55-534, CP 09340, Iztapalapa DF, Mexico
e
Laboratorio de Sistemas Dinámicos, Departamento de Ciencias Básicas, Universidad Autónoma Metropolitana-Azcapotzalco, Apartado Postal 21-267, CP 04000, Coyoacán DF, Mexico

a r t i c l e i n f o a b s t r a c t

Article history: Classical motion in an inverse square potential is shown to be equivalent to free motion on a hyperbola.
Received 1 August 2008 The existence of a classical splitting between the q > 0 and q < 0 regions of motion is demonstrated.
Received in revised form 22 November 2008 We show that this last property may be regarded as the classical counterpart of the superselection rule
Accepted 25 November 2008
occurring in the corresponding quantum problem. We solve the quantum problem in momentum space
Available online 4 December 2008
Communicated by P.R. Holland
finding that there is no way of quantizing its energy but that the eigenfunctions suffice to describe
the single renormalized bound state of the system. The dynamical symmetry of the classical problem is
Keywords: found to be O (1, 1). Both this symmetry and the symmetry of inversion through the origin are found to
Inverse square potential be broken.
Free motion on a hyperbola © 2008 Elsevier B.V. All rights reserved.
Superselection rule

The quantum problem of motion under an attractive inverse in [23] but their results are not strictly applicable to the case of
square potential has been discussed for years [1–12]. Motion in a inverse square interaction as they studied potentials of the form
such a potential has been investigated because it is relevant for |x|− p with p greater or equal than 1 but strictly less than 2.
certain applications [13–21]. The vortex structure of the proba- In this Letter we study the one-dimensional problem showing
bility flow of a single particle is related to behaviour under an features of its quantum solution that are analogous to properties
inverse-square potential [4,10,11]. This sort of potential has also of the classical motion. We show the existence of a superselection
been used for studying oscillators coupled to dipole fields, using rule which may be regarded as a quantum extension of a classical
the potential β q2 + g /q2 , a case in which tunnelling across the feature, the independence between the right and the left sides of
barrier has been found possible [12,22]. A charged particle inter- the singularity at q = 0.
acting with an electric dipole can be described through an inverse Let us recall that classical motion under certain potentials can
square potential, so it is useful for describing the scattering of elec- be made equivalent to geodesic motion on a curved space. An out-
trons off polar molecules, and to explain the existence of a critical standing example is the hydrogen atom which is equivalent to
dipole moment for the binding of an electron to a polar molecule. geodesic flow on a sphere [25,26]. We here show that another
It has been used to argue the existence of an anomaly in molecular example is the classical 1D motion under the inverse square po-
physics. The need to having a better understanding of the problem tential
has prompted the use of field theoretic techniques, thus its renor-
malized solutions have been computed employing cut-off [18] and k
V (q) = − (1)
dimensional regularization [19]. Such schemes have established the q2
existence of a single bound state breaking the scale invariance of
the problem and providing an example of dimensional transmuta- with k > 0. The motion under (1) can be transformed to free
tion in a finite system [18,19]. A related problem has been studied motion on a branch of a hyperbola. These classical results are
then used to analyse its quantum behaviour. More features of
the classical motion under (1) are analyzed and shown to be re-
lated to peculiarities of its quantum solution. In particular we
*Corresponding author.
show that the classical splitting that occurs at the singularity
E-mail addresses: manvlk@yahoo.com (M. Ávila-Aoki), carmen@ce.fis.unam.mx
(C. Cisneros), rodolfo@dirac.fciencias.unam.mx (R.P. Martínez-y-Romero),
at q = 0 is related to both the existence of a superselection rule
nyhn@xanum.uam.mx (H.N. Núñez-Yépez), asb@correo.azc.uam.mx and to the spontaneous breaking of parity that occur in the
(A.L. Salas-Brito). quantum system. This result hints towards the possibility of us-

0375-9601/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.physleta.2008.11.054
M. Ávila-Aoki et al. / Physics Letters A 373 (2009) 418–421 419

ing this kind of interactions for quantum information processing


[28,29].
To show the equivalence of motion under the potential (1) with
geodesic motion of a particle on a hyperbola, let us begin with the
autonomous classical Hamiltonian
p2 k
H= − , (2)
2m q2
so H is a constant of the motion and the energy, E, is clearly con-
served. Other conspicuous feature of Hamiltonian (2) is its scale
invariance, any change q → λq and p → p /λ2 leaves the equations
of motion unchanged if the energy is rescaled as E → λ2 E. Thence
the problem has no characteristic dimensional scales. Quantum
mechanically this scaling symmetry also holds [24] but such quan-
tum symmetry can be broken at a certain value of the coupling
constant [17].
To classically regularize the problem the first step is to perform
a π /2 rotation in phase space, p → q and q → − p, to transform
(2) to

q2 k
− = E. (3)
2m p2
Fig. 1. Phase portrait (p vs q) of the system with Hamiltonian (2) in arbitrary units.
We analyze the case of negative energies so we set  = k/ p 2 − Notice how the phase trajectories avoid crossing from negative to positive values
q2 /2m ≡ − E > 0 and rewrite (3) as of q and vice versa. Classically, the particle remain confined in just one side of the
  2 singularity for ever.
q2 p 1
+1 = . (4)
2m k  free particle moving on a compact region, the motion is necessar-
Let us define ily unbounded. The two possible ways of moving on the hyperbola
 correspond to the bouncing motions that can occur in the original
q2 problem, the particle can come towards  the origin starting from
P = p 1+ , (5) the right at an initial position q0 = + (k/ ) > 0 with zero speed
4α 2
then move towards the origin and, instead of getting to it, change
where we have used α 2 ≡ m (q, p )/2 — where we are making ex- its velocity from −∞ to +∞, bouncing back towards its starting
plicit that  is merely a shorthand for a function of p and q. We position in a finite time. This can be explicitly proved just noting
need to choose a Q as to complete a canonical transformation. To that the following Poisson bracket
this end, we use the generating function

H , sgn(q) = 0, (11)
√   
F (q, P ) = 2km log 2P q + 2 2km + P 2 q2 , (6) always vanishes. Hence, S = sgn(q) is a classical constant of the
motion. The two motions with the same energy are both parity-
from which we get, through p = ∂ F (q, P )/∂ q, the momentum (5), related and independent from one another. This feature can be ap-
and the coordinate preciated from the phase portrait, in the original coordinates [used
 in (2)] shown in Fig. 1, which clearly exhibits how the system
∂ F (q, P ) p 2 q2 avoids crossing through the singularity. For proving the constancy
Q = = −q 1 − . (7)
∂P 2km of S we used the original Hamiltonian (2), therefore the constant
is valid both for the E  0 case and for the E < 0 case and does
The transformed Hamiltonian, H  , is then
not depend on the canonical transformation or the regularization
P2 performed.
H = , (8) The dynamical invariance group of both the original classical

problem (1) and of the free particle on a hyperbola is the same,
where μ ≡ k/2. We have thus transformed the Hamiltonian of the O (1, 1), with both translations ξ → ξ + ξ0 , p ξ → p ξ and reflec-
problem to a free particle form [26]. Using in (2) the alternate co- tions ξ, p ξ → −ξ, − p ξ [30] as can be seen explicitly in Eq. (10).
ordinates The constancy of the S implies that the classical motion is wholly
 
q √ to the right or wholly to the left of the singularity at q = 0. Parity
ξ = arcsinh √ , pξ = 2m p cosh(ξ ), (9)
is already broken at the classical level. As we show below, in the
2m
quantum case the operator associated to Ŝ generates a superselec-
the original Hamiltonian is also converted to the free form (8), tion rule. This operator breaks parity.
p 2ξ We now pinpoint some quantum properties of the system that
H  = . (10) are related to the previous classical features. As one may in-

fer from equation (10), the quantum system has eigenfunctions
These coordinates (ξ, p ξ ) exhibit the phase curve on which the ψ(ξ ) = N exp(±i βξ ), where N a normalization constant and β is
particle freely moves, it is a branch of a hyperbola. The hyperbola any real number. No discrete negative energy levels are found for
branch gets traversed at a constant speed in one of the two possi- there is no way to quantize the energy in an unbounded region
ble ways, either from ξ = −∞ to ξ = +∞ or the other way round. with no singularities. The energy levels are continuous, given a
Notice that in the process of getting (8) — or (10) — we have regu- state with negative energy 0 there are infinitely more with lower
larized the problem, the singularity at q = 0 has disappeared from energies — as can be also argued from dimensional considerations.
the Hamiltonian. It is not possible to transform the problem to a This feature implies that there is no ground state in the system
420 M. Ávila-Aoki et al. / Physics Letters A 373 (2009) 418–421

[40]. If the Hamiltonian (2) is interpreted as a quantum operator, all states of the system both the negative and the positive energy
it can be factored as ones. This operator Ŝ commutes with every other observable in
 1   the system. However, it does not commute with the operator of
h̄2 d b+ 2 d b + 12
Ĥ = + − + , (12) inversion through the origin, P̂ [30,37],
2m dx x dx x
[ Ŝ , P̂ ] = 0. (16)
where h̄2 (b + 1/2)(b − 1/2)/2m = k is the constant in the potential.
Hence, the associated property does not correspond to a good
The mean value of the energy  Ĥ cannot be negative if κ  1/4,
quantum number and the symmetry under inversion through the
where κ ≡ 2mk/h̄2 ; hence κ > 1/4 to allow for negative energy
origin is broken. The eigenstates of the problem may be regarded
eigenstates in Hamiltonian (2) [18]. Thus κ = 1/4 plays the role of
as doubly degenerate. This apparent violation of the nondegener-
a critical coupling and as long as κ is larger than the critical value
acy theorem for bound states in one dimension, as in the example
the theory may admit bound states. However, it still requires of
of the 1D hydrogen atom [31,39,41], is due to fact that the sin-
renormalization for producing the bound state that the regularized
gularity at q = 0 separates the two regions q > 0 and q < 0. Even
problem supports. Renormalization breaks the O (1, 1) scale sym-
the quantum flux, j = i (h̄/2m)(ψ∂ψ ∗ /∂ q − ψ ∗ ∂ψ/∂ q), vanishes for
metry of the problem and has been shown to produce a simple
all state at q = 0 [31]. There cannot be any relation between the
realization of an anomaly, namely, the binding of an electron to a
region q > 0 and the region q < 0 as has been also established
polar molecule [17,19]. In the classical 1D case these properties are
for the hydrogen atom in one-dimension [27,31,34]. As a quantum
related to the splitting at the origin that breaks the symmetry of
analog of the classical space splitting, the Hilbert space of the sys-
the problem.
tem H is splitted in two coherent sectors H = H+ ⊕ H− , where
On the other hand, the symmetry under reflections of (1) and
H+ (H− ) is the space where the left (right) vanishing eigenfunc-
the above considerations, mean that the coordinate space eigen-
tions belong [35]. The feature has been termed a “space splitting”
functions corresponding to the two possible classical bouncing mo-
by some authors [38]. So, any physically realizable state of the
tions, namely |φ+ and |φ− , should vanish, respectively, for q < 0
system must be an eigenstate of Ŝ, and its eigenvalues +1, and
and for q > 0. Therefore the system has no eigenstates with well
−1 label the superselection sectors, H+ and H− , in which the
defined parity and thus parity is also spontaneously broken. We
Hilbert space splits. The superselection rule we have shown oper-
can prove this property as follows. Assume we were allowed to
ates, though proved for the one-dimensional case, might neverthe-
form parity states,
less be useful in some applications [29].
1 1 The first step in the classical regularization we carried out, the
|φe = √ |φ+ + |φ− and |φo = √ |φ+ − |φ− ; (13)
2 2 π /2 rotation in phase space, corresponds to a Fourier transform in
the quantum case. This suggest using the momentum representa-
we should then conclude that such states describe exactly the
tion for analysing the quantum problem. We therefore write, with
same physics. To begin with, the probability densities would have
the help of the correspondence
the same value everywhere for both the even and the odd states
|φe |2 = |φo |2 , and their Wronskian determinant would always van- p
−1 i
ish, meaning that |φe and |φo could not be regarded as describing q̂ → · · · dp  , (17)

two different states. But, if we cannot distinguish between |φe and −∞
|φo , it should be clear that the relative phase of the components
the Schrödinger equation of the problem in the p-representation
along the left and right states, |φ− and |φ+ , is irrelevant. There-
as
fore, any quantum state of the form

  p p
|Ψ = a− |φ− + a+ |φ+ , (14) p2 k
+  φ( p ) + φ( p  ) dp  dp  = 0, (18)
where a− and a+ are numbers, is necessarily devoid of meaning
2m h̄2
−∞ −∞
as a coherent superposition [31,32]. For (14) to represent a bona
where φ( p ) is the momentum eigenfunction. Such expression im-
fide quantum state, the coefficients should vanish or the state (14)
plies, according to (17) and (18), that the corresponding eigen-
should have to be interpreted as an incoherent superposition [32,
function in the coordinate representation, ψ(q), has to vanish the
33]. This feature is a manifestation of a superselection rule oper-
origin ψ(0) = 0. Selecting (17) corresponds to a sort of quantum
ating in the system [33–35]. Such rule forbids coherent superposi-
version of the regularization performed because it ameliorates the
tions of states |φ± on one side of the singularity with states |φ∓
effect of the singularity in q-space.
on the other side. We are also forced to acknowledge that parity
The solutions of (18) can be found by converting it to a dif-
is broken. The superselection rule may be regarded as the quan-
ferential equation. Every solution to such differential equation is
tum version of the classical property we have shown occurs in the
expressible as a linear combination of
classical problem, the particle is confined completely to the right
or completely to the left of q = 0, as proven by the conservation of    
2 F1 u − , u + , 1/2, − p 2 /2m and 2 F1 v − , v + , 3/2, − p 2 /2m p ,
S = sgn(q). The superselection rule shows that the side of the sin-
gularity in which the motion occurs remains a classical property (19)
even in the quantum case. Note that we use the term ‘classical where 2 F 1 [a, b, c ; −x ] is a hypergeometric function, and we
2
property’ meaning a property that can be always sharply and si- have introduced u ± ≡ (3 ± χ )/4 and v ± = (5 ± χ )/4, with χ ≡
multaneously measured with any other observable, as the mass in √
1/4 − κ . As it should be clear from our discussion, we cannot
nonrelativistic quantum mechanics [36]. impose any further limitation on the solutions. Therefore, there
The quantum operator associated with the classically conserved also no limitation on the possible energy values. Energy is not
quantity S and generating the superselection rule, is quantized as we have already guessed from the classical analysis,
 f  f   f  f  it can take therefore any negative value. As the classical analysis
Ŝ = φ φ  − φ φ  , (15)
+ + − −
has suggested, the system does not have a ground state.
f
Notice that the nature of the solutions must change when
where f is a label distinguishing the different states of the sys- κ attains values greater than 1/4 as χ becomes purely imagi-
tem and the sum (or integration if f is continuous) should include nary. Thus, κc ≡ 14 is a critical value in the coupling constant.
M. Ávila-Aoki et al. / Physics Letters A 373 (2009) 418–421 421

As we have anticipated the solutions change about κc , which is References


a threshold separating two regimes, one with no bound states
(κ < κc ) and other where bound states may exist (κ  κc ). The [1] K.M. Case, Phys. Rev. 80 (1950) 797.
functions (19) are the negative energy component functions in the [2] J.-M. Levy-Leblond, Phys. Rev. 153 (1967) 1.
[3] E.A. Guggenheim, Proc. Phys. Soc. 89 (1966) 491.
p-representation. Such components √ limit the possible
√ values the [4] G. Parisi, F. Zirilli, J. Math. Phys. 14 (1973) 243.
momentum variable can take as − 2m  p  + 2m in any [5] W.M. Frank, D.J. Land, R.M. Spector, Rev. Mod. Phys. 43 (1971) 36.
negative energy state. To finalize, we should check that our compo- [6] H. van Haeringen, J. Math. Phys. 19 (1978) 2171.
nent solutions (19) are capable of expressing the single bound state [7] C. Schwartz, J. Math. Phys. 17 (1976) 863.
[8] S.P. Alliluev, Sov. Phys. JETP 34 (1972) 8.
obtained by renormalization [17,19]. Using the expression given for
[9] B. Simon, Arch. Ration. Mech. Anal. 52 (1974) 44.
such state given in [19] we get, after Fourier transforming it, that [10] D.-Y. Song, Phys. Rev. A 62 (2000) 014103.
the only bound state of the system in the p-representation is [11] H. Wu, D.W.L. Sprung, Phys. Rev. A 49 (1994) 4305.
[12] S.-H. Dong, M. Lozada-Cassou, Am. J. Appl. Sci. 2 (2005) 376.
[13] T. Barford, M.C. Birse, J. Phys. A: Math. Gen. 38 (2005) 697.
φbs ( p ) = A 1 (3/4)2 2 F 1 3/4, 3/4, 1/2, − p 2 /2m [14] M. Bawin, S.A. Coon, Phys. Rev. A 67 (2003) 042712.
[15] B.D. Simons, P.A. Lee, B.L. Altshuler, Phys. Rev. Lett. 70 (1993) 4122.
+ A 2 p (5/4)2 2 F 1 5/4, 5/4, 3/2, − p 2 /2m , (20)
[16] C. Desfrançois, H. Abdoul-Carime, N. Khelifa, J.P. Schermann, Phys. Rev. Lett. 73
√ (1994) 2436.
√ A 1 ≡ (2/π ) (m ) (1 + −i ), A 2 ≡ i /(2π (m ) ) ×
1/ 2 1/ 4 1/ 2 1/ 4
where [17] H.E. Camblong, L.N. Epele, H. Fanchiotti, C.A. García Canal, Phys. Rev. Lett. 87
(1 + −i ) are constants. Hence, the bound state wave function of (2001) 220402.
the problem can be expressed — as it should be — in terms of [18] K.S. Gupta, S.G. Rajeev, Phys. Rev. D 48 (1993) 5940.
[19] H.E. Camblong, L.N. Epele, H. Fanchiotti, C.A. García Canal, Phys. Rev. Lett. 85
the momentum components (19) at precisely the critical case. The
(2000) 1590.
existence of the single bound state (20) breaks the scale symmetry [20] E. Braaten, D. Phillips, Phys. Rev. A 70 (2004) 052111.
of the problem. Hence an anomalous breaking ensues, as has been [21] S.R. Beane, P.F. Bedaque, L. Childress, A. Kryjevski, J. McGuire, U. van Kolck,
explicitly proved in [17,19]. Phys. Rev. A 64 (2001) 042103.
We have shown that the classical motion under an inverse [22] H. Miyasaki, I. Tsutsui, Ann. Phys. 299 (2002) 78.
[23] V.B. Gostev, A.R. Frenkin, Theor. Math. Phys. 74 (1988) 247.
square potential is equivalent to free motion on a branch of a
[24] H.E. Camblong, C.R. Ordoñez, Mod. Phys. Lett. A 17 (2002) 817.
hyperbola. We then uncovered the O (1, 1) hidden symmetry of [25] R.J. Finkelstein, J. Math. Phys. 8 (1967) 443.
the problem. The impenetrability of the origin also follows since [26] J. Moser, Commun. Pure Appl. Math. 23 (1980) 609.
S = sgn(q) is a classically conserved quantity, hence the system [27] L.J. Boya, M. Kmiecik, A. Bohm, Phys. Rev. A 37 (1988) 3567.
exhibits the so-called “space splitting” phenomenon [38]. We have [28] S.D. Bartlett, T. Rudolph, R.W. Spekkens, Rev. Mod. Phys. 79 (2007) 555.
[29] S.J. Jones, H.M. Wiseman, S.D. Bartlett, J.A. Vaccaro, D.T. Pope, Phys. Rev. A 74
also proved that the quantum operator associated to the classi- (2006) 062313.
cal quantity S induces a superselection rule in the corresponding [30] G.B. Wybourne, Classical Groups for Physicists, Wiley–Interscience, New York,
quantum problem. We solved the quantum problem in the mo- 1974.
mentum representation exhibiting that its general solution is able [31] H.N. Núñez-Yépez, C. Vargas, A.L. Salas-Brito, Eur. J. Phys. 8 (1987) 189.
[32] C. Cisneros, R.P. Martínez-y-Romero, H.N. Núñez-Yépez, A.L. Salas-Brito, Eur. J.
to reproduce the previously reported regularized solution in the
Phys. 19 (1998) 237.
coordinate representation. We point out that the breaking of the [33] H.N. Núñez-Yépez, C. Vargas, A.L. Salas-Brito, J. Phys. A: Math. Gen. 21 (1988)
scale symmetry of the problem has been interpreted as a manifes- L651.
tation of a physicochemical anomaly [17,19,24]. [34] H.N. Núñez-Yépez, C.A. Vargas, A.L. Salas-Brito, Phys. Rev. A 39 (1989) 4306.
[35] C.G. Wick, A.S. Wightman, E.P. Wigner, Phys. Rev. 88 (1952) 101.
[36] H. Primas, Perspectives in Theoretical Chemistry, Lectures Notes in Chemistry,
Acknowledgements vol. 24, Springer, Berlin, 1981.
[37] R.P. Martínez-y-Romero, A.L. Salas-Brito, J. Math. Phys. 33 (1992) 1831.
This work has been partially supported by a PAPIIT-UNAM re- [38] U. Oseguera, M. de Llano, J. Math. Phys. 34 (1993) 4575.
[39] P.M. Platzman, M.I. Dykman, Science 284 (1999) 1967.
search grant (number IN 115406-3). We dedicate this work to the [40] H.E. Camblong, L.N. Epele, H. Fanchiotti, C.A. Garcia Canal, Ann. Phys. 287
loving memories of Holda Yolanda Brito-Ancona, Darío Moreno, (2001) 14.
Oralia Vela-Carolán, M. X’Sac, and P.A. Koshka. [41] W. Fischer, H. Leschke, P. Müller, J. Math. Phys. 36 (1995) 2313.

Você também pode gostar