Você está na página 1de 4

Volume 146B.

number 5

PHYSICS LETTERS

18 October 1984

INDUCED FERMION CURRENT IN THE e MODEL IN (2 + 1) DIMENSIONS T. JAROSZEWICZ

Centre de Physique ThPorique 2, CNRS Luminy, Case 907, 13288 Marseille Cedex 9, France Received 25 June 1984

The topological current in the (2 + 1)-dimensional O(3) nonlinear D model is shown to be identical to the fermionic current induced by a Yukawa-type interaction with the classical field of the D model. Alternatively, the system can be described in terms of two species of fermions interacting with an abelian gauge field, and hence related to the recently discovered three-dimensional anomaly. The role of fermion zone modes is also discussed.

Several nonlinear classical field theories are known to possess topologically non-trivial solutions. One of the first examples studied [l] was the (2 t l)-dimensional nonlinear O(3) u model. In this model the dy namical variable is a 3-component field n(x) of unit length n2=n n = 1, (a = 1,2,3), and topological classesof solu(ti&s are characterized by the homotopy group Il,(S*) = Z (assuming a boundary condition n + const when x + -). The topological charge Q can be defined as: Q=ld3xJo, where Jp = (l/h)

we show that this is indeed possible, and discuss several peculiar properties of fermions coupled to the field n(x). More details and a discussion in a wider context will be given elsewhere [4]. The first step is to find an appropriate interaction. A natural guess (shown below to be correct), is the lagrangian

(2)
describing a Yukawa-type coupling of an isodoublet fermion to n(x) (which plays the role of a Higgs field). Here the 9 are isospin Pauli matrices, and we use two-dimensional Dirac matrices, y = u3, yk = -iuk (k = 1 2). The induced current is defined by

~pvvh~ab,naa92,ahn,

is an identically conserved topological current. A closely analogous expression arises in the (3 t I)-dimensional model (Skymre model) in which the topological charge is interpreted as the baryon (fermion) number [2]. Moreover it was shown in this case [3] that the topological current really is the induced fermion current, i.e. the current carried by fluctuations of a quantized fermion field coupled to the classical CT model field. This suggests that perhaps the current (1) in (2 t 1) dimensions can also be obtained as an induced fermion current. In this letter On leave from the Institute of Nuclear Physics, Cracow,
Poland. 2 Laboratoire propre du Centre National de la Recherche Scientifique.

v,(x) = (T @x>rp , WY
with the expectation value taken in the presence of the field n(x). Using the method of Goldstone and Wilczek [3], it can be easily calculated under the assumption that n(x) varies slowly on distances m-l, on which the fermion effectively propagates. Then one can expand n around its value, say N, at a given point, i.e. define the unperturbed propagator S(p) = i@-mN*f)-l andconsider-$m(n--N)-rJ/asa perturbation. In the limit m -+00it is sufficient to calculate the fermion loop with the vertex J/y,, J/ and two insertions of the perturbation, i.e. in momentum space, the correlation function

0370-2693/84/$03.00 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

337

Volume

146B. number

PHYSICS

LETTERS

18 October

1984

fi

bc(k,k,, k2) = i(2n)-3

Expressing uh in terms of a this can also be written as

= (m/lml)(i/471)Epvh where k t kl t k2 = 0. For small external momenta lI P bC(k k, k2) = -(m/lml)(l/87r)N


+ a%bcepvvhkikFi

Tr{n.zavCZ

s2-raAs2

(6) a-}.

0(k3/m3) ,

and v,(x) = (m/lml)(l/8n)E,,^E,b,n,anbahn, (3)

The last form clearly shows that VP is independent of the choice of the constant vector N, indeed eq. (6) is exactly equivalent to eq. (3). Let us now come back to eq. (5). It states that VP only depends on one component N,uz of $. Therefore, taking for definiteness N = (0, 0, l), we can replace the lagrangian (4) by XA =jL (iTraP t
~w,T~ VZT~)$,

turns out to be identical to JP (for m > 0; for m < 0 the sign of the coupling in (2) changes and so changes the sign of V,). There is another way of describing the fermion interaction: in (2) we can make a local gauge transformation $(x) + a- (x)$(x) to rotate n(x) to any constant vector N, N . T = !Z- l(x)n(x)
l

(7)

where, with an appropriate wc( = u; = - :(l

choice of !& . (8)

-I-n3)- lQ13 n&pl

z Q(x) .

Then the lagrangian becomes LG=$[iya, trpu,(x)-mN*r]$, (4)

The lagrangian LA describes two fermion fields of masses fm coupled whit a strength ?l to an abelian (hence subscript A) gauge field wP. Unlike u:, wp is not a pure gauge field and its field strength tensor wPv = aPw, - a,w, does not vanish. In terms of its dual *W/J = &pVAw ,,h, the induced current reads v,(x) = -(m/lml)(l/2n)*w,(x) . (9)

where up = tiFra = CZ-lia,s2 is a pure gauge field. Now the current Vfi can be expressed in terms of oneloop correlation functions involving the vector coupling. For m -+00only the two- and three-point function can contribute. In particular, n,haQc& s i(2rrP3 ld3p Tr{y,iS(p)y,raiS@-k)}

This is (up to a factor 2 because of two species of fermions) the current associated with the recently found three-dimensional abelian anomalies [5,6] (see also ref. [7]). Eqs. (8) and (9) give a useful expression for the charge Q (from now on we take m > 0, so that J,, = V,). Let us now assume a time-independent configuration n,(r) such that as I + 00, n3 tends to some constant n3 = cos 0 (not necessarily equal +I). Then Q can be written as a contour integral

= -(m/lm

I) (i/2r)e,

uh kvNa + O(k*/m)

and the three-point function turns out to be O(k/m) and therefore negligible. As a result, the induced current comes entirely from the two-point correlation function and is * v,(x) = -(~/I~I)(~/~~~)E~~~~~N~u~(x).

along a large circle, taken in the positive (anti-clockwise) direction. We get Q = ;(l - cos 0)~) (10)

(5)

* This formula and eq. (9) are not valid along stringslines where &?is discontinuous and up singular (see below). However, eqs. (3) and (6) still hold.

where 27rv = Acu is the angle by which the phase cr of (n 1 t in2) rotates when going around the contour. As noted before, eq. (9) does not hold at points where n3 -+-1 and around which the phase of n 1 + in2 rotates by a nonzero amount. If such singular vortices exist, they have to be excluded by taking

338

Volume

146B, number

PHYSICS

LETTERS

18 October

1984

small contours around them and adding their contributions (-A42n) to eq. (10). [These vortices are points where the t = const. plane is punctured by the string; the string, like the Dirac string, is of course only an artifact of the gauge transformation.] A particularly interesting case is 0 = n/2 for r -+ 00. Then the charge is half-integer, which signals a possible existence of self-conjugate zero-energy solutions of the Dirac equation [8]. We are thus led to study the energy spectra of the hamiltonians corresponding to lagrangians (2) and (7): Hy = -c30kak
HA = -030k(ak

rr it is equal to the net number of energy levels changing from negative to positive values. To check whether this really happens one can write $ = (vu +faaa)u2 (here IJJis considered as a matrix whose first index is the spinor and the second the isospin index) and note that from the Dirac equation in D, for E = 0, it follows that some combinations of fp are functions of z = x + iv holomorphic in D: f3 t if,, = f+(z),f3 - if0 = f*_(z), fi - if2 = g+(z),fi + if2 = g?(z). The bag boundary conditions are

+ e3mn a ra 7
- iwk) t U3m

where nk is the external normal to the boundary

L, or

rl*f* =77*f+ -ng+, 17*8*_ =ng++qf+,


wheren=nl tirzq,n=nl +ir)2.Sincef+,g, are holomorphic in D, their phase can only change by a nonnegative amount when going around D in the positive direction. The phase of n changes by 2n and of n by 2nv. By considering the phases of the expressions above one concludes that zero-energy solutions can only exist for n3 = 0, i.e. 0 = 42. Then, if v > 0, g, = 0, f+f_/l I n/v L, and there are v different f+j_= on ways of writing f+ as a product of two independent f_ holomorphic functions (this essentially corresponds to writing zV-l = z~z~-~-~, k = 0, . . . v - 1). So, there are exactly v independent zero-energy solutions at 0 = 7r/2, and as 0 increases, the energies change sign from negative to positive values, as can be checked by considering u3mn3r3 as a perturbation. An analogous result is obtained for v < 0. The spectrum of HA is different (although its spectral asymmetry is the same as for Hy). Physically, HA describes a fermion interacting with a magnetic field * wO(r), corresponding to the total flux @, defined as @E-$ s d2 r *wO(r) = (1 - cos ep .

(the latter is written for the isospin + f component; for the other component the signs of Wk and m change). There are two situations when the spectrum of Hy is symmetric about E = 0. First, when y2 = n2 = 0, 1 the energy changes sign under the charge conjugation.
Hy + a'H;a'

Secondly, when n3 = 0, another type of charge conjugation


Hy + u3r3Hyo3r3

changes the sign of energy. Both situations can be realized in the following setup: Take a certain region D on the plane, for simplicity singly connected and containing no singular vortices, and set m = 0 there. Next assume n3 = cos 0 = const. on the boundary L of D, and give the fermion a large mass m + 00 outside D. This is just the two-dimensional analog of the chiral bag model [9,10] (see ref. [IO] for the discussion of the limit m + -), and also in analogy it is possible to show that the charge defined as a spectral asymmetry of Hy coincides with that given by eq. (lo), where v is defined on the contour L. Let us now see how eq. (10) is related to zeroenergy eigenstates of fermions in our bag. The energy spectra are identical for 19= 0 and I!?= rr, and in both cases are symmetric about E = 0. However, an energy eigenvalue, say, negative at 0 = 0, may change sign when 6 increases to rr. If it is consistently interpreted [lo] as an antiparticle energy eigenvalue (as it was at f3 = 0), then the charge will increase by one unit. Therefore, if the charge vanishes at f3 = 0, at 0 =

This problem was studied in the literature, but only for @ = integer [ 1 I] and/or m = 0 [ 12,6]. A straightforward generalization [ 131 leads to the following picture: the induced charge clearly vanishes for @ = e = 0 (in particular, if wfi = 0, HA is invariant under the charge conjugation HA + olHial). When 0 and @ increase, there arises a nonzero charge from the continuum spectrum (or quasi-continuum, if the space is suitably compactified [6]). It grows until @ reaches 339

Volume

146B. number

PHYSICS

LETTERS

18 October

1984

unity, then drops to zero, grows again, etc. In addition, at every positive integer @ a new discrete energy level at E = -m appears [ 1 I], so the charge (for one species of fermions) is simply @/2. When the limit m + 0 is taken, these discrete energy levels become zeroenergy levels [ 12,6]. However, the continuum contribution to the charge does not vanish, in spite of the fact that HA for m = 0 changes sign under the conjugation HA + a3HAa3. The reason is that the continuum charge density is -m 2, but spreads over large distances -m - 1; this is analogous to what happens for the monopole (see ref. [14] and in particular ref. [15]). Our analysis can be extended [ 131 to discuss the spin and statistics of topological excitation in the u model with heavy fermions added. The angular momentum can be then regarded as induced by fermions in the electromagnetic field, and is simply equal to Q2/2 (modulo integer). Equivalently, the topological term (Hopf invariant) [ 161 to be added to the action of the u model is actually generated by the (anomalous) fermion loops and appears with the coefficient equal to rr. The author thanks the Centre de Physique Theorique, CNRS, for the hospitality in Marseille, and Dr. Robert Coquereaux and Dr. Klaus Fredenhagen for helpful discussions.

111 A.A. Belavin and A.M. Polyakov,

Pisma Zh. Eksp.

Teor. 22 (1975) 503 [JETP Lett. 22 (1975) 2451. 121 T.H.R. Skyrme, Proc. R. Sot. (London) A260 (1961) 127. 131 J. Goldstone and F. Wilczek, Phys. Rev. Lett. 47 (1981) 986. to be published. [41 T. Jaroszewicz, 151 A. Niemi and G. Semenoff, Phys. Rev. Lett. 51 (1983) 2077; A.N. Redlich, Phys. Rev. Lett. 52 (1984) 18;Phys. Rev. D29 (1984) 2366; L. Alvarez-Gaume and E. Witten, Harvard preprint HUTP-83/A039 (1983). [61 R. Jackiw, Phys. Rev. D29 (1984) 2375. Ann. Phys. 171 S. Deser, R. Jackiw and S. Templeton, (NY) 140 (1982) 372; I. Affleck, J. Harvey and E. Witten, Nucl. Phys. B206 (1982) 413. 181 R. Jackiw and C. Rebbi, Phys. Rev. D13 (1976) 3398. and R.L. Jaffe, Phys. Rev. Lett. 51 (1983) 191 J. Goldstone 1518. Lett. 143B (1984) 217. 1101 T. Jaroszewicz,Phys. [l I] C.R. Nohl, Phys. Rev. D12 (1975) 1840; H.J. de Vega, Phys. Rev. D18 (1978) 2932. [ 121 R. Jackiw and P. Rossi, Nucl. Phys. B 190 [FS3] (1981) 681. [ 131 T. Jaroszewicz, to be published. [14] C.G. CaIIan Jr., Phys. Rev. D25 (1982) 2141; D26 (1982) 2058; Nucl. Phys. B212 (1983) 391. [ 151 B. Grossman, Phys. Rev. Lett. 51(1983) 959; H. Yamagishi, Phys. Rev. D27 (1983) 2383. [16] F. Wilczek and A. Zee, Phys. Rev. Lett. 51 (1983) 2250.

340

Você também pode gostar