Chiral limit of the two-dimensional fermionic determinant in a general magnetic field

M. P. Fry

Citation: Journal of Mathematical Physics 41, 1691 (2000); doi: 10.1063/1.533204

View online: http://dx.doi.org/10.1063/1.533204
View Table of Contents: http://scitation.aip.org/content/aip/journal/jmp/41/4?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

The Euclidean scalar Green function in the five-dimensional Kaluza-Klein magnetic monopole space-time
J. Math. Phys. 47, 012302 (2006); 10.1063/1.2159071

Generalized Yang–Mills actions from Dirac operator determinants

J. Math. Phys. 42, 5238 (2001); 10.1063/1.1399297

SU (N) - and SO (N) -invariant chiral fields: One- and two-dimensional subspaces
J. Math. Phys. 40, 2500 (1999); 10.1063/1.532879

Schwinger’s method and the computation of determinants

Am. J. Phys. 66, 524 (1998); 10.1119/1.18894

Bosonized Lagrangians in higher orders of the chiral expansion

Phys. Part. Nucl. 29, 33 (1998); 10.1134/1.953059

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 JOURNAL OF MATHEMATICAL PHYSICS VOLUME 41, NUMBER 4 APRIL 2000

Chiral limit of the two-dimensional fermionic determinant

in a general magnetic field
M. P. Frya)
School of Mathematics, University of Dublin, Dublin 2, Ireland
共Received 28 September 1999; accepted for publication 15 November 1999兲
We consider the effective action for massive two-dimensional QED in flat Euclid-
ean space–time in the background of a general square-integrable magnetic field
with finite range. It is shown that its small mass limit is controlled by the chiral
anomaly. New results for the low-energy scattering of electrons in 2⫹1 dimen-
sions in static, inhomogeneous magnetic fields are also presented. © 2000 Ameri-
can Institute of Physics. 关S0022-2488共00兲00404-7兴

I. INTRODUCTION

Fermionic determinants lie at the heart of gauge field theories with fermions. They are ob-
tained by integrating over the fermionic degrees of freedom in the presence of a background
potential A ␮ , producing the one-loop effective action S eff⫽⫺ln det, where the fermionic determi-
nant, det, is formally the ratio det(P” ⫺A” ⫹m)/det(P” ⫹m) of determinants of Dirac operators. The
coupling constant e has been absorbed into A ␮ . This action is exact and appears in the calculation
of every physical process. Therefore, any truly nonperturbative calculation must deal with S eff in
its full generality. The main problem with calculating S eff is that is must be known for generic
potentials, typically tempered distributions, if it is to be part of an effective measure for A ␮ . A
summary of what is known about S eff in quantum electrodynamics in 1⫹1, 2⫹1, and 3⫹1
dimensions for general fields is given in Sec. I of Ref. 1. Recall that S eff in QED only depends on
the field strength tensor F ␮ ␯ . It is seen that there are upper and lower bounds on S eff , with some
bounds holding only for restricted fields, such as unidirectional ones. After 50 years or so there are
still no equalities in QED for general fields, except for massless QED in 1⫹1 dimensions—the
Schwinger model.2
In this paper an equality is obtained for the chiral limit of S eff in two-dimensional, Wick-
rotated Euclidean QED for a general field, hereafter referred to as a static magnetic field B(r). Of
course B is not completely unrestricted. We described elsewhere1,3 precisely how rough potentials
and fields are to be smoothed as part of the regularization process required to make the functional
integration over A ␮ well-defined. It is sufficient to assume in this paper that A ␮ is differentiable
and that B is square integrable with finite range R. Then B is guaranteed to have finite flux, ⌽,
since 储 B 储 ⭓ 兩 ⌽ 兩 / 冑␲ R, where 储 B 储 2 ⫽ 兰 d 2 rB(r) 2 and ⌽⫽ 兰 d 2 rB(r). The author knows of no defi-
nition of a determinant that can handle infinite flux fields; there is simply too much degeneracy,4
resulting in volumelike divergences 共which are ignored兲 as in the constant field case. Furthermore,
finite flux and range are consistent with the need to introduce a volume cutoff to define QED2
before taking the thermodynamic limit.
With the foregoing restrictions on B our result is

⳵ 兩⌽兩
lim m 2 ln det⫽ , 共1.1兲
⳵m2 4␲
m 2 →0

where m is the fermion mass. Together with the exact scaling relation

a兲
Electronic mail: mpfry@maths.tcd.ie

0022-2488/2000/41(4)/1691/20/$17.00 1691 © 2000 American Institute of Physics

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1692 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

ln det共 ␭ 2 B 共 ␭r兲 ,m 2 兲 ⫽ln det共 B 共 r兲 ,m 2 /␭ 2 兲 , 共1.2兲

Eq. 共1.1兲 implies the strong field limit

兩⌽兩
ln det共 ␭ 2 B 共 ␭r兲 ,m 2 兲 ⬃ ⫺ ln ␭⫹R 共 ␭ 兲 , 共1.3兲
␭Ⰷ1 2␲

where lim␭→⬁ (R(␭)/ln ␭)⫽0. Note that the chiral limit in Eq. 共1.1兲 implies that QED2’s fermionic
determinant behaves like ( 兩 ⌽ 兩 /4␲ )ln m2 as m→0, which does not coincide with that of the
Schwinger model.
For nonwinding background fields with ⌽⫽0 one can prove continuity at m⫽0. As a result,
massive QED2’s fermionic determinant does coincide with that of the Schwinger model at m
⫽0:

lim ln det⫽
m→0
1
4␲2
冕 d 2 r d 2 r ⬘ B 共 r兲 B 共 r⬘ 兲 ln兩 r⫺r⬘ 兩 . 共1.4兲

This follows from results of Seiler5 and Simon6 as will be shown in a future paper.
It is reasonable to ask what is the relevance of QED2’s fermionic determinant, and its mass
dependence in particular, to physics? The answer is that the integral of this determinant over the
fermion mass fully determines QED4’s fermionic determinant for the same magnetic field B(r). 7
This determinant is still unknown except for a constant field8,9 and a sech2(x/a) varying unidirec-
tional field.10
From the input parameters to the QED2 determinant one can form the dimensionless ratios
e 储 B 储 /mc 2 and ប/mcR, where e has been temporarily restored. This paper deals with the nonper-
turbative, small mass region e 储 B 储 /mc 2 , ប/mcRⰇ1. The large mass region can be dealt with by a
derivative expansion of ln det.11 What remains in order to estimate QED4’s fermionic determinant
for general unidirectional fields are optimal upper and lower bounds on QED2’s determinant for
intermediate values of the mass.
The derivation of Eq. 共1.1兲 is really just a problem in quantum mechanics dealing with a
particle confined to a planar surface with an inhomogeneous magnetic field normal to it. The
proportionality of the limit in Eq. 共1.1兲 to the two-dimensional chiral anomaly, ⌽/2␲ , as well as
its sign and its connection with paramagnetism, are discussed in Ref. 12. Equation 共1.1兲 was
established in Ref. 12 in finite volume for a unidirectional field B(r)⭓0. There is a missing
volume factor in Eq. 共2.7兲 that was corrected in Ref. 13. These restrictions are dropped in this
paper.
Finally, the chiral limit of QED2’s continuum fermionic determinant should provide a non-
trivial test of algorithms for the determinant on large lattices. The reason is that chiral limits and
topological invariants—the chiral anomaly in this instance—are notoriously difficult to calculate
on a lattice.14 Many of the results here on low-energy scattering in static, inhomogeneous mag-
netic fields are new and are relevant to the physical case of electrons in such fields in 2⫹1
dimensions.
In Sec. II we discuss how we will demonstrate Eq. 共1.1兲. Section III develops the essentials of
low-energy scattering in inhomogeneous magnetic fields that will be required. In Sec. IV the
crucial argument that central symmetry is sufficient to establish Eq. 共1.1兲 is given. Finally, Sec. V
gives the fine points of the limit in Eq. 共1.1兲.

II. PRELIMINARIES

We adopt Schwinger’s proper time definition9 of the fermionic determinant for Euclidean
QED2:

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1693

ln det⫽
1
2
冕
0
⬁ dt
t
2 2
Tr共 e ⫺ P t ⫺exp兵 ⫺ 关共 P⫺A兲 2 ⫺ ␴ 3 B 兴 t 其 兲 e ⫺tm . 共2.1兲

Then

⳵ 1 2 ⫺1
2 ln det⫽ Tr关共 D ⫺ ␴ 3 B⫹m 兲
2
⫺ 共 P 2 ⫹m 2 兲 ⫺1 兴 , 共2.2兲
⳵m 2

where D 2 ⫽(P⫺A) 2 and ␴ 3 is the Pauli matrix. Now introduce the sum rule15

⌽
Tr关共 D 2 ⫺B⫹m 2 兲 ⫺1 ⫺ 共 D 2 ⫹B⫹m 2 兲 ⫺1 兴 ⫽ , 共2.3兲
2␲m2

where the trace is over space indices only, and assume without loss of generality that ⌽⬎0. Then
Eqs. 共2.2兲 and 共2.3兲 give

⳵ ⌽
m2 2 ln det⫽ ⫹m 2 Tr关共 D 2 ⫹B⫹m 2 兲 ⫺1 ⫺ 共 P 2 ⫹m 2 兲 ⫺1 兴 . 共2.4兲
⳵m 4␲

The continuum part of the spectrum of the negative chirality operator D 2 ⫹B stretches down to
zero in the case of open spaces. Because ⌽⬎0 the square-integrable zero modes are confined to
the spectrum of D 2 ⫺B. 16 Although B has no definite sign, its flux does, and ⌽, chirality and the
number of square-integrable zero modes of the supersymmetric pair of operators D 2 ⫾B are
correlated by the Aharonov–Casher theorem. At this stage the minor modifications one has to
make to deal with the case when ⌽⬍0 are already clear.
It might seem that Eq. 共2.4兲 makes Eq. 共1.1兲 self-evident. But if Eq. 共2.3兲 is multiplied by m 2
and the limit m 2 ⫽0 taken, then the fractional part of the chiral anomaly is given by a difference
of zero-energy phase shifts of opposite chirality,17 demonstrating that the trace difference in Eq.
共2.3兲 develops a 1/m 2 -type singularity at the bottom of the continuum. How, then, does one know
a priori that such a singularity is absent from the trace in Eq. 共2.4兲?
Our definition of the determinant in Eq. 共2.1兲 leads us to define the trace in Eq. 共2.4兲 by a
difference of diagonal heat kernels,

Tr关共 D 2 ⫹B⫹m 2 兲 ⫺1 ⫺ 共 P 2 ⫹m 2 兲 ⫺1 兴 ⫽ 冕 0
⬁
dte ⫺tm
2
冕 d 2 r 具 r兩 e ⫺ 共 D
2 ⫹B 兲 t 2
⫺e ⫺ P t 兩 r典 . 共2.5兲

Denote the scattering states of D 2 ⫹B corresponding to outgoing radial waves by ␺ (⫹) (k,r)
⫽ 具 r兩 k,in 典 whose eigenvalues are E⫽k 2 . These satisfy the normalization condition

冕 d 2 r ␺ 共 ⫹ 兲 * 共 k,r兲 ␺ 共 ⫹ 兲 共 k⬘ ,r兲 ⫽ ␦ 共 k⫺k⬘ 兲 . 共2.6兲

Assume that B(r) is noncentral. Let ⌰ denote the direction of the incident beam with momentum
k relative to an axis fixed in the scattering center. The asymptotic behavior of ␺ (⫹) (k,r) for kr
Ⰷ1 is

1 ikr cos共 ␪ ⫺⌰ 兲 f 共 ␪ ,⌰ 兲 ikr

␺ 共 ⫹ 兲 共 k,r兲 ⫽ e ⫹ e ⫹R, 共2.7兲
2␲ 2 ␲ 冑r

where f is the scattering amplitude and R is the remainder in the large-r expansion of ␺ (⫹) .
Equation 共2.7兲 is obtained from the Lippmann–Schwinger equation

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1694 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

␺ 共 ⫹ 兲 共 k,r兲 ⫽
1 ik"r i
2␲
e ⫺
4
冕 d 3r ⬘H ⫹
0 共 k 兩 r⫺r⬘ 兩 兲 V 共 r⬘ 兲 ␺
共⫹兲
共 k,r⬘ 兲 , 共2.8兲

where H (⫹)
0 is a Hankel function of the first kind and

V⫽⫺P"A⫺A"P⫹A 2 ⫹B. 共2.9兲

As we will show later, we can choose a gauge such that

⌽
A⫽ ˆ␪ , 共2.10兲
2␲r

for r sufficiently large, where ␪ˆ is a unit vector orthogonal to r. Therefore, we are dealing with a
long range (1/r 2 ) potential V, and this is what makes the proof of Eq. 共1.1兲 nonroutine. The
completeness of the ‘‘in’’ states for D 2 ⫹B and Eq. 共2.5兲 gives

Tr关共 D 2 ⫹B⫹m 2 兲 ⫺1 ⫺ 共 P 2 ⫹m 2 兲 ⫺1 兴

⫽ 冕0
⬁
dte ⫺tm
2
冕 冕 d 2r
⬁

0
dk ke ⫺k
2t
冕
0
2␲
d⌰ 共 兩 ␺ 共 ⫹ 兲 共 k,r兲 兩 2 ⫺ 兩 ␺ 0 共 k,r兲 兩 2 兲 , 共2.11兲

where ␺ 0 (k,r)⫽e ik"r/2␲ . We are interested in the small m 2 , high t limit of Eq. 共2.11兲 which is
determined by the low-energy end of the spectrum of D 2 ⫹B. Therefore we cut off the energy
integral in Eq. 共2.11兲 at M with M RⰆ1 and consider, for m 2 →0,

Tr关共 D 2 ⫹B⫹m 2 兲 ⫺1 ⫺ 共 P 2 ⫹m 2 兲 ⫺1 兴

⫽ 冕 0
⬁
dte ⫺tm
2
冕 冕 d 2r
0
M
dk ke ⫺k
2t
冕
0
2␲
d⌰ 共 兩 ␺ 共 ⫹ 兲 共 k,r兲 兩 2 ⫺ 兩 ␺ 0 共 k,r兲 兩 2 兲 ⫹R 共 m 2 兲 . 共2.12兲

The remainder, R, can be put in the form

冕 0
⬁
e ⫺t 共 m
2 ⫹M 2 兲
冕 冕d 2r
0
⬁
dp pe ⫺p
2t
冕0
2␲
d⌰ 共 兩 ␺ 共 ⫹ 兲 共 冑p 2 ⫹M 2 ,r,⌰ 兲 兩 2 ⫺ 兩 ␺ 0 共 冑p 2 ⫹M 2 ,r,⌰ 兲 兩 2 兲 ,

which makes the energy gap between 0 and M evident so that limm 2 →0 m 2 R(m 2 )⫽0. Thus Eq.
共2.4兲 shows that Eq. 共1.1兲 will be established if the integral in Eq. 共2.12兲 multiplied by m 2 vanishes
in the limit m 2 ⫽0.
We will calculate in the Lorentz gauge ⳵ ␮ A ␮ ⫽0 which, in two dimensions, allows us to set
A ␮ ⫽ ⑀ ␮ ␯ ⳵ ␯ ␾ with

B 共 r, ␪ 兲 ⫽⫺ ⳵ 2 ␾ 共 r, ␪ 兲 , 共2.13兲

and ⑀ ␮ ␯ ⫽⫺ ⑀ ␯ ␮ with ⑀ 12⫽1. Assuming that B has range R we can calculate ␾ in a disk D of radius
aⰇR with ␾ (a, ␪ )⫽0. A unique solution of Poisson’s equation with Dirichlet boundary condi-
tions requires that we also specify ␾ as r→⬁, which we will do by requiring that ␾ approach the
potential of a flux line through the origin. The construction of the Dirichlet Green’s function for
this problem is standard, with the result

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1695

␾ 共 r, ␪ 兲 ⫽⫺
1
4␲
冕 D
d 2 r ⬘ B 共 r ⬘ , ␪ ⬘ 兲 ln
冉 r 2 ⫹r ⬘ 2 ⫺2rr ⬘ cos共 ␪ ⬘ ⫺ ␪ 兲
r 2r ⬘2
a ⫹ 2 ⫺2rr ⬘ cos共 ␪ ⬘ ⫺ ␪ 兲
2
a
,
冊 r⬍a

共2.14兲
⌽
⫽⫺ ln共 r/a 兲 , r⬎a.
2␲

This potential results in a discontinuity in ⳵ ␾ / ⳵ r at r⫽a that is of order 储 B 储 R 2 /a 2 and which

vanishes in the limit of radial symmetry. This introduces a zero-flux, zero-range magnetic field on
a ring at r⫽a that does not affect the low energy phase shifts. What is gained by this is radial
symmetry for r⬎a.
From Eq. 共2.12兲 it is evident that we will need the outgoing wave solution of

关共 P⫺A兲 2 ⫹B 兴 ␺ 共 ⫹ 兲 ⫽k 2 ␺ 共 ⫹ 兲 , 共2.15兲

for k→0. This can be solved explicitly in the exterior region r⬎a with overall normalization fixed
by Eq. 共2.7兲. We can approximate the interior solution by the exact zero-energy solution of Eq.
共2.15兲 because k 2 is a regular perturbation of D 2 ⫹B for r⬍a. Then an interior solution of Eq.
共2.15兲 can be expanded as a power series in k 2 . Following this the interior and exterior solutions
are matched at r⫽a. 17

III. LOW-ENERGY SCATTERING STATES

Since the case of noncentral potentials may be unfamiliar we will parallel our discussion with
the special case of radial symmetry in the interest of clarity.

A. Central field: r Ì a

Expand ␺ (⫹) (k,r) in partial waves,

⬁
1
␺ ⫹ 共 k,r兲 ⫽ 兺 ␺ l共 k,r 兲 e il共 ␪ ⫺⌰ 兲.
& ␲ l⫽⫺⬁
共3.1兲

Equations 共3.1兲 and 共2.6兲 give the normalization condition

冕 0
⬁
drr ␺ l* 共 k,r 兲 ␺ l 共 k ⬘ ,r 兲 ⫽ ␦ 共 E⫺E ⬘ 兲 . 共3.2兲

Substitution of Eqs. 共3.1兲 and 共2.10兲 in Eq. 共2.15兲 results in Bessel’s equation for ␺ 1 , with l
shifted to l⫺⌽/2␲ for r⬎a. In order to include the case when ⌽/2␲ is an integer we choose as
linearly independent solutions the Hankel functions H 兩(⫾)
l⫺⌽/2␲ 兩 (kr) whose asymptotic behavior for
r→⬁ is

H 共␯⫾ 兲 共 kr 兲 ⬃ 冑 2 ⫾i 共 kr⫺ ␯␲ /2⫺ ␲ /4兲

␲ kr
e . 共3.3兲

Setting W⫽ 兩 l⫺⌽/2␲ 兩 we construct ␺ l as the following linear combination:

e ⫺i ␲ W/2e i ␲ 兩 l 兩
␺ l 共 k,r 兲 ⫽ 共 H 共W⫺ 兲 共 kr 兲 ⫹e i ␲ 共 W⫺ 兩 l 兩 兲 e 2i ␦ l 共 k 兲 H 共W⫹ 兲 共 kr 兲兲 , 共3.4兲
2&

where S l ⫽e 2i ␦ l is the S-matrix for the partial phase shift ␦ l . Recalling that in two dimensions

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1696 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

e ik"r⫽e ikr cos共 ␪ ⫺⌰ 兲 ⫽ 兺

l⫽⫺⬁
i l J l 共 kr 兲 e il 共 ␪ ⫺⌰ 兲 , 共3.5兲

and noting Eq. 共3.3兲 we see that the normalization factors in Eq. 共3.4兲 ensure that Eq. 共3.1兲
assumes the asymptotic form Eq. 共2.7兲 as r→⬁.

B. General field: r Ì a
Although radial symmetry is present for r⬎a, the absence of rotational symmetry for r⬍a
can cause the incident particle to scatter into a final state that is a superposition of angular
momentum states. Thus the S-matrix is no longer diagonal in l:S l →S l,L , where L is the initial-
state angular momentum. Then Eq. 共3.1兲 generalizes to

1
␺ 共 ⫹ 兲 共 k,r兲 ⫽
&␲
兺
l,L
␺ l,L 共 k,r 兲 e il ␪ e ⫺iL⌰ , 共3.6兲

which, together with Eq. 共2.6兲, results in the normalization condition

兺l 冕 0
⬁
* 共 k,r 兲 ␺ l,L ⬘ 共 k ⬘ ,r 兲 ⫽ ␦ L,L ⬘ ␦ 共 E⫺E ⬘ 兲 .
drr ␺ l,L 共3.7兲

Equation 共3.4兲 now generalizes to

1
␺ l,L 共 k,r 兲 ⫽e ⫺i ␲ 共 W l ⫹W L 兲 /4e i ␲ 共 兩 l 兩 ⫹ 兩 L 兩 兲 /2 共 ␦ l,L H 共W⫺l 兲 共 kr 兲 ⫹e i ␲ 共 W l ⫺ 兩 l 兩 兲 /2S l,L e i ␲ 共 W L ⫺ 兩 L 兩 兲 /2H 共W⫹l 兲 共 kr 兲兲 ,
2&
共3.8兲

where W l ⫽ 兩 l⫺⌽/2␲ 兩 , etc. Unless ⌰ needs to be displayed, as in Eq. 共3.6兲, we suppress it in what
follows. Again, the normalization factors in Eq. 共3.8兲 are chosen so that Eq. 共3.6兲 assumes the
asymptotic form of Eq. 共2.7兲. The scattering amplitude is given by

1
冑2 ␲ k 兺
f 共 k⬘ ,k兲 ⫽ 共 S l,L ⫺ ␦ l,L 兲 e il ␪ e ⫺iL⌰ e i ␲ 共 W L ⫺W l ⫺1 兲 /4. 共3.9兲
l,L

C. General field: r Ë a
We seek zero-energy solutions of Eq. 共2.15兲 in the region r⬍a that are sufficiently regular to
maintain the Hermiticity of D 2 ⫹B. This operator factorizes to L † L so that Eq. 共2.15兲 at k 2 ⫽0
reduces to

L † L ␺ ⫽0, 共3.10兲

where

L⫽e ⫺i ␪ 冉 1 ⳵ 1 ⳵ 1 ⳵␾
⫺ ⫺
i ⳵r r ⳵␪ r ⳵␪
⫺i
⳵␾
⳵r
. 冊 共3.11兲

One set of solutions is given by

L ␺ ⫽0, 共3.12兲

whose solution by inspection is

␺ ⫽e ⫺ ␾ 共 r, ␪ 兲 g 共 re ⫺i ␪ 兲 , 共3.13兲

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1697

where g is analytic in re ⫺i ␪ in and on the disk D. Solutions of the form Eq. 共3.13兲 do not give all
of the regular solutions of Eq. 共3.10兲. This is evident in the limit of radial symmetry, for then ␺ is
a superposition of only negative or zero angular momentum states.
There are irregular solutions of Eq. 共3.12兲 and hence Eq. 共3.10兲 of the form

␺ ⫽e ⫺ ␾ 共 r, ␪ 兲 h 共 r ⫺1 e i ␪ 兲 , 共3.14兲

where h may be expanded in a power series away from the origin. These solutions can be used to
find additional regular solutions of Eq. 共3.10兲 that reduce to superpositions of positive angular
momenta in the radial symmetry limit. Thus we look for regular solutions about the origin of the
form

␺ ⫽e ⫺ ␾ h 共 r ⫺1 e i ␪ 兲 F 共 r, ␪ 兲 . 共3.15兲

Then Eq. 共3.10兲 gives

冋
L † e ⫺ ␾ e ⫺i ␪ h 共 r ⫺1 e i ␪ 兲 冉 ⳵F i ⳵F
⫺
⳵r r ⳵␪ 冊册 ⫽0. 共3.16兲

Again by inspection the solution of Eq. 共3.16兲 is

e ⫺ ␾ e ⫺i ␪ h 冉 ⳵F i ⳵F
⫺
⳵r r ⳵␪ 冊
⫽e ␾ b 共 re i ␪ 兲 , 共3.17兲

where b is analytic in re i ␪ in and on D. Actually, h is now an unnecessary complication. Letting

F⫽ f (r, ␪ )/h, we get

⳵f i ⳵f
⫺ ⫽e i ␪ e 2 ␾ b 共 re i ␪ 兲 , 共3.18兲
⳵r r ⳵␪

and hence Eq. 共3.15兲 becomes

␺ ⫽e ⫺ ␾ f . 共3.19兲

Equation 共3.18兲 indicates that f is undetermined up to a function of the form p(re ⫺i ␪ ). But this is
the same as g in Eq. 共3.13兲, and so we set p⫽0. Also, the value of ␺ at the origin can be fixed by
g. So for definiteness we require f (0)⫽0. Noting that

ⵜ 2 ⫽e ⫺i ␪ 冉 ⳵ i ⳵ i␪ 1 i ⳵
⫺
⳵r r ⳵␪
e ⫹
r r ⳵␪ 冊 冉
, 冊 共3.20兲

the solution of Eq. 共3.18兲 is, for r苸D,

f 共 r兲 ⫽
1 i␪ ⳵
2␲
e 冉
⫹
i ⳵
⳵r r ⳵␪ 冊冕D
d 2 r ⬘ ln兩 r⫺r⬘ 兩 e 2 ␾ 共 r⬘ 兲 b 共 r ⬘ e i ␪ ⬘ 兲 ⫹C, 共3.21兲

where C is a constant fixed by f (0)⫽0. Since

e i␪ 冉 ⳵
⫹
i ⳵
⳵r r ⳵␪ 冊
ln兩 r⫺r⬘ 兩 ⫽⫺e i ␪ ⬘
⳵
⫹
i ⳵
⳵r⬘ r⬘ ⳵␪⬘ 冉
ln兩 r⫺r⬘ 兩 , 冊 共3.22兲

f takes the final form

f 共 r兲 ⫽⫺
1
2␲
冕D
d 2 r ⬘ e i ␪ ⬘ e 2 ␾ 共 r⬘ 兲 b 共 r ⬘ e i ␪ ⬘ 兲 冉 ⳵
⫹
i ⳵
⳵r⬘ r⬘ ⳵␪⬘ 冊
ln共 兩 r⫺r⬘ 兩 /r ⬘ 兲 . 共3.23兲

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1698 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

Combining Eqs. 共3.13兲, 共3.19兲, and 共3.23兲, the general solution of Eq. 共3.10兲 is

␺ 共 r, ␪ 兲 ⫽e ⫺ ␾ 共 r, ␪ 兲 共 g 共 re ⫺i ␪ 兲 ⫹ f 共 r, ␪ 兲兲 . 共3.24兲

The functions b, f, and g will be determined below when we join the region r⬍a with r⬎a. They
then acquire energy-dependent normalization factors that depend on scattering data, including the
initial-state angular momentum L. Thus ␺ in Eq. 共3.24兲 has an implicit dependence on ⌰.
As discussed at the end of Sec. II, k 2 is a regular perturbation of D 2 ⫹B in Eq. 共2.15兲 in the
region r⬍a. Hence the radial wave functions ␺ l,L (k,r) in Eq. 共3.6兲 may be expanded in k 2 for
r⬍a. Inserting the implicit ⌰-dependence of ␺ (r, ␪ in Eq. 共3.24兲 we expand it in partial waves

␺ 共 r, ␪ ,⌰ 兲 ⫽ 兺 ␺ l,L 共 r 兲 e il ␪ e ⫺iL⌰ . 共3.25兲

l,L

We set

␺ l,L 共 k,r 兲 /& ␲ ⫽ ␺ l,L 共 r 兲共 1⫹k 2 ␹ l,L 共 r 兲 ⫹O 共 k 4 兲兲 , 共3.26兲

and thus

␺ 共 ⫹ 兲 共 k,r兲 ⫽ ␺ 共 r, ␪ ,⌰ 兲 ⫹
k2
共 2␲ 兲2
冕0
2␲
d␪⬘ 冕
0
2␲
d⌰ ⬘ ␺ 共 r, ␪ ⫺ ␪ ⬘ ,⌰⫺⌰ ⬘ 兲 ␹ 共 r, ␪ ⬘ ,⌰ ⬘ 兲 ⫹O 共 k 4 ␺ 兲 ,
共3.27兲

where ␹ can be expanded as in Eq. 共3.25兲. We will abbreviate Eq. 共3.27兲 as

␺ 共 ⫹ 兲 共 k,r兲 ⫽ ␺ 共 r, ␪ 兲 ⫹k 2 ␺ 夝 ␹ ⫹O 共 k 4 ␺ 兲 , 共3.28兲

where the star denotes convolution. An equation for ␹ can be obtained by substituting Eq. 共3.28兲
in Eq. 共2.15兲 and retaining terms of order k 2 . As we will see, ␹ is not required in the general field
case.
To fix b and f define the operator

⳵ i ⳵
L⫽ ⫺ , 共3.29兲
⳵r r ⳵␪

and let it act on ␺ (⫹) in Eq. 共3.28兲, using Eq. 共3.24兲:

L␺ 共 ⫹ 兲 ⫽⫺ ␺ L␾ ⫹e ⫺ ␾ Lf ⫹k 2 L␺ 夝 ␹ ⫹O 共 k 4 L␺ 兲 . 共3.30兲

Equation 共2.14兲 gives

⳵ ␪ ␾ 共 a, ␪ 兲 ⫽0, 共 ⳵ r ␾ 共 r, ␪ 兲兲 a ⫽⫺⌽/2␲ a⫹O 共 储 B 储 R 2 /a 2 兲 . 共3.31兲

From now on the arbitrarily small correction to the radial derivative of ␾ at r⫽a will be implicit
in what follows. Thus Eqs. 共3.31兲 and 共3.18兲 applied to Eq. 共3.30兲 at r⫽a give

L␺ 共 ⫹ 兲 共 k,a, ␪ 兲 ⫽⌽ ␺ 共 a, ␪ 兲 /2␲ a⫹e i ␪ b 共 ae i ␪ 兲 ⫹k 2 L␺ 夝 ␹ ⫹O 共 k 4 L␺ 兲 . 共3.32兲

Denote the wave functions on either side of a by ␺ ⬍

(⫹)
and ␺ ⬎
(⫹)
. Continuity of ␺ (⫹) and L␺ (⫹)
at r⫽a and repeated use of Eq. 共3.28兲 allow Eq. 共3.32兲 to be put in the form

e i ␪ b 共 ae i ␪ 兲 ⫽ 共 L⫺⌽/2␲ a 兲 ␺ 共⬎⫹ 兲 ⫺k 2 ⌽ ␺ 共⬎⫹ 兲 夝 ␹ /2␲ a⫹k 2 L␺ 共⬎⫹ 兲 夝 ␹ ⫹O 共 k 4 L␺ 共⬎⫹ 兲 兲 . 共3.33兲

Since b(re i ␪ ) is analytic in and on D we can make the expansion

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1699

⬁ ⬁

b 共 re i ␪ 兲 ⫽ 兺 兺
l⫽0 L⫽⫺⬁
b l,L r l e il ␪ e ⫺iL⌰ , 共3.34兲

where we have anticipated the ⌰-dependence of b. From Eqs. 共3.6兲, 共3.33兲, and 共3.34兲 we get

& ␲ b l⫺1,L a l⫺1 ⫽ 冉 d

dr
⬎
⫹ 共 l⫺⌽/2␲ 兲 /r ␺ l,L 冊
共 k,r 兲共 1⫹k 2 ␹ l,L 共 r 兲 ⫹O 共 k 4 兲兲 , 共3.35兲

with r⫽a after differentiating. Referring to Eq. 共3.8兲 it is evident from Eq. 共3.35兲 that the expan-
sion coefficients b l,L and hence b and f in Eq. 共3.23兲 will be determined to leading order in k 2 once
S l,L is known.
There now remains the function g in Eq. 共3.24兲. Equations 共3.24兲 and 共3.28兲 together with
continuity of ␺ (⫹) at r⫽a give

␺ 共⬎⫹ 兲 共 k,a, ␪ 兲 ⫽g 共 ae ⫺i ␪ 兲 ⫹ f 共 a, ␪ 兲 ⫹k 2 ␺ 夝 ␹ 共 a, ␪ 兲 ⫹O 共 k 4 ␺ 兲 . 共3.36兲

Letting

f 共 r, ␪ 兲 ⫽ 兺
l,L
f l,L 共 r 兲 e il ␪ e ⫺iL⌰ , 共3.37兲

and recalling that g is analytic in and on D so that

⬁ ⬁

g 共 re ⫺i ␪ 兲 ⫽ 兺 兺
l⫽0 L⫽⫺⬁
g l,L r l e ⫺il ␪ e ⫺iL⌰ , 共3.38兲

we obtain from Eqs. 共3.6兲, 共3.36兲–共3.38兲, for l⭓0,

⬎ ⬎ ⬎
␺ ⫺l,L 共 k,a 兲 /& ␲ ⫽g l,L a l ⫹ f ⫺l,L 共 a 兲 ⫹k 2 ␺ ⫺l,L 共 k,a 兲 ␹ ⫺l,L 共 a 兲 /& ␲ ⫹O 共 k 4 ␺ ⫺l,L 兲. 共3.39兲

This simplifies on making the expansion

冉 冊
⬁
1 r⬍ l
ln(r ⫹r ⬘ ⫺2rr ⬘
2 2
cos( ␪ ⫺ ␪ ⬘ ))⫽ln r ⬎
2
⫺2 兺
l⫽1 l r⬎
cos关 l 共 ␪ ⫺ ␪ ⬘ 兲兴 , 共3.40兲

in Eq. 共3.23兲, giving f ⫺l,L (a)⫽0, l⬎0 and

冕 冕
⬁
1 a 2␲
f 0,L 共 a 兲 ⫽
2␲ 0
dr
0
d␪e e i ␪ 2 ␾ 共 r, ␪ 兲
兺 b l,L r l e il ␪ .
l⫽0
共3.41兲

Thus g is determined to leading order in k 2 by Eqs. 共3.38兲–共3.39兲 once S l,L is known. We have
now fully determined the low-energy limit of ␺ (⫹) (k,r) for general magnetic fields in terms of the
S-matrix.

D. Central field: r Ë a

Now everything is diagonal and we may set a⫽R. Refer back to Eq. 共3.4兲 and define the
energy-dependent part, ⌬ l , of the phase shifts by

␦ l 共 k 2 兲 ⫽ ␲ 共 兩 l 兩 ⫺ 兩 l⫺⌽/2␲ 兩 兲 /2⫹⌬ l 共 k 2 兲 ⫹m ␲ , 共3.42兲

where m⫽0,⫾1,... Then Eq. 共3.4兲 reduces to, for r⬎a,

␺ l 共 k,r 兲 ⫽2 ⫺1/2共 ⫺1 兲 m i 兩 l 兩 e i ␦ l 共 J w 共 kr 兲 cos ⌬ l ⫺Y W 共 kr 兲 sin ⌬ l 兲 , 共3.43兲

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1700 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

where Y W is the Bessel function of the second kind. Then at r⫽a with kaⰆ1 and W⫽0,

␺⬎
l 共 k,a 兲 ⫽2
⫺1/2
共 ⫺1 兲 m i 兩 l 兩 e i ␦ l 关共 ka/2兲 W /⌫ 共 W⫹1 兲

⫹⌬ l ⌫ 共 W 兲共 ka/2兲 ⫺W / ␲ 兴共 1⫹O 共 k 2 ,⌬ 2l 兲兲 , 共3.44兲

共 r ⳵ r␺ ⬎
l 兲 a ⫽2
⫺1/2
W 共 ⫺1 兲 m i 兩 l 兩 e i ␦ l 关共 ka/2兲 W /⌫ 共 W⫹1 兲
⫺⌬ l ⌫ 共 W 兲共 ka/2兲 ⫺W / ␲ 兴共 1⫹O 共 k 2 ,⌬ 2l 兲兲 . 共3.45兲

It will be shown in Sec. IV that ⌬ l ⫽O(ka) 2W at least. In Eqs. 共3.44兲 and 共3.45兲 the remainder
term O(k 2 ,⌬ 2l ) should be replaced with O((ka) 2 ln ka) when W⫽1.
We can now calculate b l⫺1 in Eq. 共3.35兲. For l⬎⌽/2␲ , Eqs. 共3.35兲, 共3.44兲, and 共3.45兲 give

b l⫺1 a l ⫽ ␲ ⫺1 共 l⫺⌽/2␲ 兲共 ⫺1 兲 m i l e i ␦ l 共 ka/2兲 W /⌫ 共 W⫹1 兲共 1⫹O 共 k 2 ,⌬ 2l 兲兲 , 共3.46兲

and for 1⭐l⬍⌽/2␲ ,

b l⫺1 a l ⫽ ␲ ⫺2 共 l⫺⌽/2␲ 兲共 ⫺1 兲 m i l e i ␦ l ⌬ l ⌫ 共 W 兲共 ka/2兲 ⫺W 共 1⫹O 共 k 2 ,⌬ 2l 兲兲 . 共3.47兲

For the case W⫽0,

b l⫺1 a l ⫽⫺ ␲ ⫺2 共 ⫺1 兲 m i l e i ␦ l ⌬ l 共 1⫹O 共 ⌬ 2l 兲兲 . 共3.48兲

For g l , Eq. 共3.39兲 gives

⬎
␺ ⫺l 共 k,a 兲
g l⫽ 共 1⫺k 2 ␹ ⫺l 共 a 兲 ⫹O 共 k 4 兲兲 , 共3.49兲
&␲al

since Eq. 共3.41兲 gives f 0,L (a)⫽0 for the case of radial symmetry. Combining Eq. 共3.49兲 with Eq.
共3.44兲, we obtain, for l⭓0,

g l ⫽ 共 ⫺1 兲 m i l e i ␦ ⫺1 共 2 ␲ a l 兲 ⫺1 关共 ka/2兲 l⫹⌽/2␲ /⌫ 共 l⫹1⫹⌽/2␲ 兲

⫺ ␲ ⫺1 ⌬ ⫺l ⌫ 共 l⫹⌽/2␲ 兲共 ka/2兲 ⫺ 共 l⫹⌽/2␲ 兲 兴共 1⫹O 共 k 2 ,⌬ ⫺l
2
兲兲 . 共3.50兲

Referring to Eqs. 共3.37兲, 共3.40兲, and 共3.23兲 one finds for r⭐a,

f l 共 r 兲 ⫽b l⫺1 r ⫺l 冕 0
r
dxx 2l⫺1 e 2 ␾ 共 x 兲 , l⭓1

⫽0, l⭐0. 共3.51兲

In the radial symmetry limit ␺ l,L (r) in Eq. 共3.26兲 becomes diagonal, with Eq. 共3.24兲 now giving,
for r⭐a,

␺ l 共 r 兲 ⫽b l⫺1 r ⫺l e ⫺ ␾ 共 r 兲 冕 0
r
dxx 2l⫺1 e 2 ␾ 共 x 兲 , l⭓1

⫽g ⫺l r ⫺l e ⫺ ␾ 共 r 兲 , l⭐0. 共3.52兲

Since it will be needed in what follows, we end this section by calculating ␹ l in Eq. 共3.26兲.
Substitution of Eqs. 共3.25兲 and 共3.26兲 in Eq. 共2.15兲 and matching terms of O(k 2 ) gives

冉 ⫺
d
dr
⫹
l⫺1
r
⫹␾⬘ 冊冉 d l
冊
⫹ ⫹ ␾ ⬘ ␺ l␹ l⫽ ␺ 1 .
dr r
共3.53兲

Requiring ␹ l (0)⫽0, Eqs. 共3.52兲 and 共3.53兲 fix ␹ l for r⭐a to be

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1701

␹ l ⫽⫺ 冕0
r
dxx ⫺2 兩 l 兩 ⫺1 e 2 ␾ 共 x 兲 冕 x

0
dy y 2 兩 l 兩 ⫹1 e ⫺2 ␾ 共 y 兲 , l⭐0

⫽⫺ 冕0
r
dxx 2l⫺1 e 2 ␾ 共 x 兲 冉冕 x

0
dww 2l⫺1 e 2 ␾ 共 w 兲 冊 冕
⫺2

0
x
dy y 1⫺2l e 2 ␾ 共 y 兲 冉冕
0
y
dzz 2l⫺1 e 2 ␾ 共 z 兲 冊 2
,

l⭓1. 共3.54兲

It is important to bound the growth of ␹ l for 兩 l 兩 →⬁. In the limit of radial symmetry Eq. 共2.14兲
reduces to

␾ 共 r 兲 ⫽⫺ 冕 0
a
dr ⬘ r ⬘ B 共 r ⬘ 兲 ln共 r ⬎ /a 兲 , 共3.55兲

for r⭐a. Then some easy estimates applied to Eqs. 共3.54兲 and 共3.55兲 yield

兩 ␾ 共 r 兲 兩 ⭐ 储 B 储共 a⫺r 兲 /2冑␲ , 共3.56兲

and

e 储 B 储 a 冑␲ 2
兩 ␹共 r 兲兩 ⭐ r , l⭐0
4 共 兩 l 兩 ⫹1 兲

e 6 储 B 储 a 冑␲ 2
⭐ r , l⭓1. 共3.57兲
4 共 l⫹1 兲

IV. LOW-ENERGY PHASE SHIFTS

In order to calculate Eq. 共2.12兲 in the limit m 2 →0 the leading energy-dependent behavior of
S l,L is required. The case of central fields is dealt with first.

A. Central fields

The calculation of ⌬ l in Eq. 共3.42兲 proceeds by matching the log-derivatives ␥ l ⫽r ⳵ ln ␺l at

r⫽a, where again we can set a⫽R. Then Eq. 共3.43兲 gives

⬘ 共 ka 兲
␥ l J W 共 ka 兲 ⫺kaJ W
tan ⌬ l ⫽ , 共4.1兲
⬘ 共 ka 兲
␥ l Y W 共 ka 兲 ⫺kaY W

where ␥ l denotes ␥ l 共inside兲. For kaⰆ1 this reduces to

W⫺ ␥ l 共 ka/2兲 2W
⌬ l⫽ ␲ 共 1⫹O 共 ka 兲 2 兲 , 共4.2兲
W⫹ ␥ l ⌫ 共 W 兲 ⌫ 共 W⫹1 兲

for W⫽ 兩 l⫺⌽/2␲ 兩 ⫽0,1,... . Results for integer values of W will be given below.
Now suppose l⭓1. Then Eq. 共3.35兲 reduces to

␥ l ⫽⌽/2␲ ⫺l⫹& ␲ b l⫺1 a l / ␺ ⬍

l 共 k,a 兲 ⫹k 关共 ⌽/2␲ ⫺l⫺ ␥ l 兲 ␹ l 共 a 兲 ⫺ 共 r ⳵ r ␹ l 兲 a 兴 ⫹O 共 k 兲 . 共4.3兲
2 4

From Eqs. 共3.6兲, 共3.24兲, and 共3.26兲,

␺⬍
l 共 k,a 兲 /& ␲ ⫽ f l 共 a 兲共 1⫹k ␹ l 共 a 兲 ⫹O 共 k 兲兲 ,
2 4
共4.4兲

which, together with Eqs. 共3.51兲 and 共4.3兲, gives

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1702 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

a 2l
␥ l ⫽⌽/2␲ ⫺l⫹ ⫺k 2
a ␹ ⬘l 共 a 兲 ⫹O 共 k 4 兲 . 共4.5兲
兰 a0 drr 2l⫺1 e 2 ␾ 共 r 兲

Note that ␥ l ⬃l as l→⬁, as it should.

Next, let l⭐0. Equations 共3.6兲, 共3.24兲, and 共3.26兲 give

␥ i ⫽⌽/2␲ ⫺l⫹k 3 a ␹ l⬘ 共 a 兲 ⫹O 共 k 4 兲 , 共4.6兲

since f l ⫽0 for l⭐0 by Eq. 共3.51兲. Equations 共3.42兲, 共4.2兲, 共4.5兲, and 共4.6兲 determine the leading
energy dependence of the phase shifts for W⫽0,1,... .
Finally, let W⫽0,1,... . There is nothing new in principle here; only the expansion of Y W for
kaⰆ1 has to be modified. The whole calculation goes forward as above with the result that for
W⫽2,3,..., ⌬ l is still given by Eq. 共4.2兲; for W⫽1 replace O(ka) 2 in Eq. 共4.2兲 by
O((ka) 2 ln(ka)), and for W⫽0,

␲
⌬ l⫽ ⫹O 共 1/ln2 共 ka 兲兲 . 共4.7兲
2 ln共 ka 兲

It is interesting that the energy dependence of ⌬ l for W⫽0 specialized to l⫽0 is exactly the same
as that derived by Chandon et al.18 for a large class of nonmagnetic Schrödinger operators in 2
⫹1 dimensions. Note that there is no smooth interpolation of ⌬ l from W⫽0 to W⫽0. This case
will, therefore, have to be considered separately in what follows.

B. General fields

The S-matrix S l,L appearing in Eq. 共3.8兲 is obtained from

S l,L ⫽ ␦ l,L ⫹& ␲ i ⫺l⫺1 兺m 冕 ⬁

0
drrJ l 共 kr 兲 V l⫺m 共 r 兲 ␺ m,L 共 k,r 兲 , 共4.8兲

with ␺ l,L given by Eqs. 共3.6兲 and 共2.8兲 and where

V l⫺m 共 r 兲 ⫽
1
2␲
冕 0
2␲
d ␪ V 共 r, ␪ 兲 e ⫺i 共 l⫺m 兲 ␪ , 共4.9兲

with V as in Eq. 共2.9兲. An infinite set of coupled equations must be solved to extract the phase
shifts in the general field case. In practice, only a few off-diagonal elements of S l,L are required to
obtain the phase shifts in the low energy limit.
Consider an incident low-energy particle (kaⰆ1) with angular momentum L with respect to
the scattering center. It will encounter a high centrifugal barrier (l⫺⌽/2␲ ) 2 /r 2 to the spatially
asymmetric region r⬍a where B(r, ␪ )⫽0. For values of L⬃⌽/2␲ the barrier is minimized, and
so we expect the magnitude of the energy-dependent corrections to the Aharonov–Bohm phase
shifts, ⌬ L , will assume their maximum values, as Eqs. 共4.2兲 and 共4.7兲 illustrate in the centrally
symmetric case. The intuition is that S l,L only has significant off-diagonal elements for values of
l,L clustered about ⌽/2␲ and that otherwise S l,L can be assumed diagonal with small error.
To test this hypothesis we will assume that

⌽
L⬍ ⬍L⫹1, 共4.10兲
2␲

and take S l,L to be a 2⫻2 matrix to include the transitions L↔L⫹1, and diagonal otherwise. This
S-matrix can be calculated for all ⌽ satisfying Eq. 共4.10兲 using the results of the previous sections.
As one would expect, the mixing of angular momentum states L and L⫹1 is maximum at the
mid-interval value ⌽/2␲ ⫽L⫹1/2. Instead of reproducing this calculation it is more instructive to

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1703

set ⌽/2␲ ⫽L⫹1/2, where the Hankel functions assume a simple form, and show that ⌬ L and
⌬ L⫹1 have the same energy dependence as in the centrally symmetric case; only the numerical
coefficients are modified. In the case of higher order transitions 兩 ⌬L 兩 ⬎1 involving larger matrices,
we find that the relevant mixing parameters 共see below兲 compared to the 兩 ⌬L 兩 ⫽1 case are smaller
by factors of order k 兩 ⌬L 兩 ⫺1 .
The calculation begins by noting that the potential in Eq. 共2.9兲 is not time-reversal invariant
for a fixed magnetic field. Therefore, S l,L is not symmetric. We choose the parameterization

S⫽ 冉 e 2i ␦ L cos 2 ⑀
ie i ␤ sin 2 ⑀
ie i ␣ sin 2 ⑀
e 2i ␦ L⫹1 cos 2 ⑀
冊 , 共4.11兲

where we expect the mixing parameter ⑀ to vanish as k→0. Unitarity requires ␣ and ␤ to be real
with

␣ ⫹ ␤ ⫽2 共 ␦ L ⫹ ␦ L⫹1 兲 . 共4.12兲

The definition of the phase shifts in Eq. 共4.11兲 is that of Stapp et al.,19 generalized here to include
T-violation. Referring to Eq. 共3.42兲 we can rewrite Eq. 共4.11兲 as

S⫽ 冉 e 2i 共 ⌬ L ⫺⌽/4兲 cos 2 ⑀
ie i 共 ⌬ L ⫹⌬ L⫹1 ⫺␭ 兲 sin 2 ⑀
ie i 共 ⌬ L ⫹⌬ L⫹1 ⫹␭ 兲 sin 2 ⑀
e 2i 共 ⌬ L⫹1 ⫹⌽/4兲 cos 2 ⑀
冊 , 共4.13兲

which introduces a real T-violating parameter ␭. From Eqs. 共3.8兲 and 共4.12兲 with ⌽/2␲ ⫽L
⫹1/2, the matching of the interior and exterior log-derivatives at r⫽a gives
⫺兲 ⫹兲
kaH 共1/2 ⬘ ⫹e 2i⌬ L kaH 共1/2 ⬘ cos 2 ⑀
␥ L,L ⫽ ⫺兲 ⫹兲 , 共4.14兲
H 共1/2 ⫹e 2i⌬ L H 共1/2 cos 2 ⑀

with ␥ L⫹1,L⫹1 ⫽ ␥ L (⌬ L →⌬ L⫹1 ). Also

⫹兲 ⫹兲
␥ L,L⫹1 ⫽ ␥ L⫹l,L ⫽kaH 共1/2 ⬘ /H 共1/2 . 共4.15兲

Recalling that

⫾兲
H 共1/2 共 z 兲 ⫽⫿i 冑 2 ⫾iz
␲z
e , 共4.16兲

Eq. 共4.13兲 becomes,

共 21 ⫹ika 兲 e ⫺2ika ⫹ 共 ⫺ 21 ⫹ika 兲 e 2i⌬ L cos 2 ⑀

␥ L,L ⫽ , 共4.17兲
e 2i⌬ L cos 2 ⑀ ⫺e ⫺2ika

␥ L,L⫹1 ⫽⫺ 21 ⫹iz. 共4.18兲

To get the interior values of ␥ l,L , refer to Eq. 共3.35兲. Then

& ␲ a L b L⫺1,L
␥ L,L ⫽⌽/2␲ ⫺L⫹ ⬍ ⫹O 共 ka 兲 2 . 共4.19兲
␺ L,L 共 k,a 兲

From Eqs. 共3.6兲 and 共3.36兲–共3.38兲,

⬍
␺ L,L 共 k,a 兲 /& ␲ ⫽ f L,L 共 a 兲 ⫹O 共 ka 兲 2 , 共4.20兲

so that

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1704 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

a L b L⫺1,L
␥ L,L ⫽⌽/2␲ ⫺L⫹ 共 1⫹O 共 ka 兲 2 兲 ⫹O 共 ka 兲 2 . 共4.21兲
f L,L 共 a 兲

Referring back to Eqs. 共3.23兲, 共3.37兲, and 共3.40兲 it follows that, for L⬎0,

冕 冕
⬁
1 a 2␲
f L,L 共 a 兲 ⫽
2␲aL 0
drr L
0
d ␪ e i 共 1⫺L 兲 ␪ e 2 ␾ 共 r, ␪ 兲 兺
m⫽0
b m,L r m e im ␪ . 共4.22兲

In the sum over b m,L in Eq. 共4.22兲, only b L⫺1,L and b L,L are nonzero as seen from Eqs. 共3.8兲,
共3.35兲, and 共4.13兲 since mixing is only assumed for L↔L⫹1. Then Eqs. 共4.21兲 and 共4.22兲 give

1
␥ L,L ⫽ ⫹2 ␲ a 2L
2 冉冕 a

0
drr 2L⫺1 冕 0
2␲
d ␪ e 2 ␾ 共 r, ␪ 兲 ⫹
b L,L
b L⫺1,L
冕 0
a
drr 2L 冕 0
2␲
d ␪ e i ␪ e 2 ␾ 共 r, ␪ 兲 冊 ⫺1

⫻ 共 1⫹O 共 ka 兲 2 兲 ⫹O 共 ka 兲 2 . 共4.23兲

Repeating the above steps we find

1
␥ L⫹1,L⫹1 ⫽⫺ ⫹2 ␲ a 2L⫹2
2 冉冕 a

0
drr 2L⫹1 冕 0
2␲
d ␪ e 2 ␾ 共 r, ␪ 兲 ⫹
b L⫺1,L⫹1
b L,L⫹1

⫻ 冕
0
a
drr 2L 冕 0
2␲
d ␪ e ⫺i ␪ e 2 ␾ 共 r, ␪ 兲 冊 ⫺1
共 1⫹O 共 ka 兲 2 兲 ⫹O 共 ka 兲 2 . 共4.24兲

We now calculate the ratios of b l,L in Eqs. 共4.23兲 and 共4.24兲. From Eq. 共3.35兲,
⬎
& ␲ a l b l⫺1,L ⫽ 共 ␥ l,L ⫹L⫺⌽/2␲ 兲 ␺ l,L 共 k,a 兲共 1⫹O 共 ka 兲 2 兲 . 共4.25兲

Then Eqs. 共4.25兲, 共4.15兲, and 共4.18兲 give

⬎
ik ␺ L⫹1,L 共 k,a 兲
b L,L /b L⫺1,L ⫽ ⬎
共 1⫹O 共 ka 兲 2 兲 . 共4.26兲
共 ␥ L,L ⫺ 1
2 兲 ␺ L,L 共 k,a 兲

Equations 共3.8兲, 共4.13兲, and 共4.16兲 give

⬎ ⬎
e i 共 ⌬ L ⫹⌬ L⫹1 ⫺␭⫺ ␲ L 兲 sin 2 ⑀
␺ L⫹1,L 共 k,a 兲 / ␺ L,L 共 k,a 兲 ⫽ . 共4.27兲
e ⫺2ika ⫺e 2i⌬ L cos 2 ⑀

Likewise,
⬎
共 ika⫺1 兲 a ␺ L,L⫹1 共 k,a 兲
b L⫺1,L⫹1 /b L,L⫹1 ⫽ ⬎
共 1⫹O 共 ka 兲 2 兲 , 共4.28兲
共 ␥ L⫹1,L⫹1 ⫹ 21 兲 ␺ L⫹1,L⫹1 共 k,a 兲

⬎ ⬎
e i 共 ⌬ L ⫹⌬ L⫹1 ⫹␭⫺ ␲ L sin 2 ⑀
␺ L,L⫹1 共 k,a 兲 / ␺ L⫹1,L⫹1 共 k,a 兲 ⫽ . 共4.29兲
e 2i⌬ L⫹1 cos 2 ⑀ ⫺e ⫺2ika

Then solving Eqs. 共4.23兲, 共4.26兲, and 共4.27兲 for ␥ L,L and matching the result with ␥ L,L in Eq.
共4.17兲 gives

共 1⫹ika 兲 e ⫺2ika I L ⫹ 共 ika⫺1 兲 I L e 2i⌬ L cos 2 ⑀

⫽e 2i⌬ L cos 2 ⑀ ⫺e ⫺2ika ⫹ikaJ L e i 共 ⌬ L ⫹⌬ L⫹1 ⫺␭⫺ ␲ L 兲 sin 2 ⑀ ⫹O 共共 ka 兲 3 , 共 ka 兲 2 ⌬ L 兲 ,
共4.30兲

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1705

where

I L ⫽ 共 2 ␲ a 2L 兲 ⫺1 冕0
a
drr 2L⫺1 冕 2␲

0
d ␪ e 2 ␾ 共 r, ␪ 兲 ,

共4.31兲
J L⫽共 2 ␲ a 2L⫹1 ⫺1
兲 冕 0
a
drr 2L
冕 0
2␲
d␪e e i ␪ 2 ␾ 共 r, ␪ 兲
.

Similarly, Eqs. 共4.24兲, 共4.28兲, 共4.29兲, and 共4.17兲 with ⌬ L replaced with ⌬ L⫹1 give

ika 共 e ⫺2ika ⫹e 2i⌬ L⫹1 cos 2 ⑀ 兲 I L⫹1 ⫽e 2i⌬ L⫹1 cos 2 ⑀ ⫺e ⫺2ika 共 1⫺ika 兲 J L쐓 e i 共 ⌬ L ⫹⌬ L⫹1 ⫹␭⫺ ␲ L 兲 sin 2 ⑀
⫹O 共共 ka 兲 3 , 共 ka 兲 2 ⌬ L⫹1 兲 . 共4.32兲

Equations 共4.30兲 and 共4.32兲 can be solved for ⌬ L , ⌬ L⫹1 , ⑀ and ␭ with

␭⫽␭ 0 ⫹␭ 1 ka⫹O 共共 ka 兲 2 兲 . 共4.33兲

The results are

⌬ L ⫽⫺ka/ 共 1⫹I L 兲 ⫹O 共共 ka 兲 3 兲 ,

⌬ L⫹1 ⫽ 关 I L⫹1 ⫺1⫺ 共 cos ␭ 0 Im J L ⫺sin ␭ 0 Re J L 兲 2 / 共 1⫹I L 兲兴 ka⫹O 共共 ka 兲 3 兲 ,

0⫽cos ␭ 0 Re J L ⫹sin ␭ 0 Im J L , 共4.34兲

␭ 1 ⫽0,

⑀ ⫽ 共 ⫺1 兲 L 共 sin ␭ 0 Re J L ⫺cos ␭ 0 Im J L 兲 ka/ 共 1⫹I L 兲 ⫹O 共共 ka 兲 2 兲 .

Note that ⌬ L and ⌬ L⫹1 have the same energy dependence as in the centrally symmetric case given
by Eq. 共4.2兲.

V. CHIRAL LIMIT

We concluded in Sec. IV that general magnetic fields only modify the numerical coefficients
of the energy-dependent part of the phase shifts calculated in the centrally symmetric case. It was
also noted that the off-diagonal elements of S l,L fall off as powers of k. Therefore, inhomogeneous
fields of the type considered here do not result in special cases that are not already included in the
central symmetry limit. We therefore confine our discussion to central symmetry from here on and
proceed to show that Eq. 共1.1兲 is true by demonstrating that the integral in Eq. 共2.12兲 satisfies

lim m 2
m 2 →0
冕0
⬁
dte ⫺tm
2
冕 冕
d 2r
M

0
dkke ⫺k
2t
冕
0
2␲
d⌰ 共 兩 ␺ 共 ⫹ 兲 共 k,r兲 兩 2 ⫺ 兩 ␺ 0 共 k,r兲 兩 2 兲 ⫽0. 共5.1兲

This integral can be divided between contributions from the regions r⬍a and r⬎a. We may set
a⫽R.

A. r Ë a

By inspection, the term 兩 ␺ 0 兩 2 in Eq. 共5.1兲 gives a contribution proportional to ln关(M 2

⫹m )/m2兴 and so vanishes in the indicated limit. This leaves an integral over partial waves ob-
2

tained by substituting Eq. 共3.1兲 in Eq. 共5.1兲. We can obtain a bound on ␺ l (k,r) for all l from Eqs.
共3.26兲 and 共3.52兲 with b l⫺1 ,g ⫺l fixed by Eqs. 共3.46兲–共3.47兲 and 共3.49兲, respectively. These
equations and the estimates

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1706 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

r ⫺l e ⫺ ␾ 共 r 兲 冕 r

0
dxx 2l⫺1 e 2 ␾ 共 x 兲 ⭐
a l 3a 储 B 储 /& ␲
2l
e ,

共5.2兲
冕 0
a
dxx 2l⫺1 e 2 ␾ 共 x 兲 ⭓a e 2l ⫺a 储 B 储 / 冑␲
/2l,

following from Eq. 共3.56兲 give

兩 ␺ l 共 k,r 兲 兩 ⭐
e 5a 储 B 储 /& ␲
ka
&⌫ 共 W⫹1 兲 2
冉 冊 W
共 1⫹O 共 k 2 兲兲 , 共5.3兲

where the O(k 2 ) term symbolizes a remainder term that vanishes as k→0 for all l and that falls
off as k 2 / 兩 l 兩 for 兩 l 兩 Ⰷ⌽/2␲ . Recall that W⫽ 兩 l⫺⌽/2␲ 兩 . From Eq. 共5.3兲,

冕 0
⬁
dte ⫺tm
2
冕 冕
a

0
drr
0
M
dkke ⫺k
2t a2
兺l 兩 ␺ l共 k,r 兲 兩 2 ⭐ 8 e 5a 储 B 储 / 冑␲ ln 冉 M 2 ⫹m 2
m2 冊兺
l
共 M a/2兲 2W
⌫ 2 共 W⫹1 兲
,
共5.4兲

for M aⰆ1, and hence the indicated limit in Eq. 共5.1兲 is satisfied. The special case when l
⫽⌽/2␲ ⫽1,2,... is dealt with by Eqs. 共3.48兲 and 共4.7兲. For these special values of l the integrals on
the left-hand side of Eq. 共5.4兲 contribute a term of order M 2 a 2 关 (M 2 ⫹m 2 )ln2(Ma)兴⫺1, which
vanishes in the limit indicated in Eq. 共5.1兲

B. r Ì a

Equations 共3.1兲, 共3.43兲, and the expansion 共3.5兲 substituted into Eq. 共5.1兲 result in the follow-
ing integral:

I⫽ 冕 0
⬁
dte ⫺tm
2
冕 冕a
⬁
drr
0
M
dkke ⫺k
2t
兺l 关 J W2 共 kr 兲 ⫺J 2l 共 kr 兲
⫹J W 共 kr 兲 Y W 共 kr 兲 sin 2⌬ l ⫹ 共 Y W
2
共 kr 兲 ⫺J W
2
共 kr 兲兲 sin2 ⌬ l 兴 . 共5.5兲

Consider the sum over the first two Bessel functions. Entries 5.7.11.6 of Ref. 20 and 6.538.1 of
Ref. 21 give the result

兺l 关 J W2 共 kr 兲 ⫺J 2l 共 kr 兲兴 ⫽1/2J 2f 共 kr 兲 ⫹1/2J 21⫺ f 共 kr 兲 ⫺ 冕kr dtt ⫺1 关 f J 2f 共 t 兲 ⫹ 共 1⫺ f 兲 J 21⫺ f 共 t 兲兴

⬁

⬅g 共 kr 兲 , 共5.6兲

where ⌽/2␲ ⫽N⫹ f , 0⬍ f ⬍1 and N⫽0,1,... . Next Eq. 共5.6兲 has to be integrated over k following
Eq. 共5.5兲. For this we apply the weighted mean value theorem:22 Assume f and g are continuous
on 关 a,b 兴 . If f never changes sign on 关 a,b 兴 then, for some c in 关 a,b 兴 ,

冕 a
b
f 共 k 兲 g 共 k 兲 dk⫽g 共 c 兲 冕 a
b
f 共 k 兲 dk.

2t
Let f ⫽ke ⫺k and g equal the right-hand side of Eq. 共5.6兲. Then for some ␮ satisfying 0⬍ ␮
⭐M ,

冕 0
M
dkke ⫺k
2t
兺l 共 J W2 ⫺J 2l 兲 ⫽ 共 1⫺e ⫺M t 兲 g 共 ␮ r 兲 /2t.
2
共5.7兲

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1707

The value ␮ ⫽0 is excluded since the energy integral is manifestly r-dependent. The t-integral in
Eq. 共5.5兲 can be done immediately, resulting in an overall factor of ln关(M 2⫹m2)/m2兴.
It remains to be shown that the integration over r is bounded. The definition of g in Eq. 共5.6兲
for large argument gives

sin ␲ f cos 2z 共 f 2 ⫺ f ⫺ 4 兲 sin ␲ f sin 2z

冉 冊
1
cos 2z,sin 2z
g 共 z 兲 ⫽⫺ ⫹ ⫹O . 共5.8兲
␲z ␲z2 z3

Substitution of Eq. 共5.8兲 into Eq. 共5.7兲 and performing the r-integral in Eq. 共5.5兲 results in

冕 0
⬁
dte ⫺tm lim
2

L→⬁
冕 冕 L

a
drr
M

0
dkke ⫺k
2t
兺l 关 J W2 ⫺J 2l 兴

⫽ln 冉 M 2 ⫹m 2
m2
lim ⫺
L→⬁
冊 冋
sin ␲ f
4␲␮2
sin 2 ␮ L⫹convergent as L→⬁ . 册 共5.9兲

The leading term, although oscillating, is bounded and that is all that is required to satisfy the limit
indicated in Eq. 共5.1兲. For the special case when ⌽/2␲ ⫽1,2,..., the sum in Eq. 共5.6兲 is zero.
Next we consider the J W Y W terms in Eq. 共5.5兲. Since J W
2
(kz)⫹Y W 2
(kz) is a decreasing func-
tion of z for any value of W,23 then for r⭓a,
2
YW 共 kr 兲 ⭐J W
2
共 ka 兲 ⫹Y W
2
共 ka 兲 . 共5.10兲

It follows that

冏兺 l
J W 共 kr 兲 Y W 共 kr 兲 sin共 2⌬ l 兲 ⭐ 冏 兺l 兩 sin共 2⌬ l 兲 兩 关 1⫹Y W2 共 ka 兲兴 1/2, 共5.11兲

where we used 兩 J W (z) 兩 ⭐1 for W⭓0. From Eq. 共4.2兲 together with Eqs. 共4.5兲, 共4.6兲, and 共3.56兲
one obtains for all l and ⌽/2␲ ⫽1,2,...

共 ka/2兲 2W
兩 ⌬ l 兩 ⭐ ␲ 共 e a 储 B 储 / 冑␲ ⫹1 兲 共 1⫹O 共 k 2 兲兲 , 共5.12兲
⌫ 共 W 兲 ⌫ 共 W⫹1 兲

where the O(k 2 ) terms fall off for large 兩l兩 at least like 1/兩 l 兩 . Since kaⰆ1 and

1
Y W 共 ka 兲 ⬃⫺ ⌫ 共 W 兲共 ka/2兲 ⫺W ,
␲

every term in the series on the left-hand side of Eq. 共5.11兲 is bounded by a constant

共 M a/2兲 W
2 3/2共 e a 储 B 储 / 冑␲ ⫹1 兲 ,
⌫ 共 W⫹1 兲

for all r⬎a and 0⭐k⭐M . It is, therefore, a uniformly convergent series of continuous functions
of k and can be integrated term by term. Applying the weighted mean value theorem again we
obtain

冕 0
⬁
dte ⫺tm lim
2

L→⬁
冕 冕 L

a
drr
M

0
dkk 兺l J W共 kr 兲 Y W共 kr 兲 sin 2⌬ l共 k 兲
1
⫽ ln
2 m2 冉
M 2 ⫹m 2
lim
L→⬁
冊 冕 a
L
drr 兺l J W共 ␮ r 兲 Y W共 ␮ r 兲 sin 2⌬ l共 ␮ 兲 , 共5.13兲

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1708 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

for some ␮ in the interval 0⬍ ␮ ⭐M . For 兩 l 兩 Ⰷ⌽/2␲ , J W ( ␮ r)Y W ( ␮ r)⬃⫺( ␲ 兩 l 兩 ) ⫺1 which, to-
gether with Eq. 共5.12兲, implies each term in the series on the right-hand side of Eq. 共5.13兲 is
dominated by a constant whose l-dependence is ( ␮ a/2) 2 兩 l 兩 /( 兩 l! 兩 ) 2 for all r⬎a. For all finite L
⬎a it is a uniformly convergent series of continuous functions of r that can be integrated term by
term. From entry 5.11.10 of Ref. 23,

冕 a
L
drrJ W 共 ␮ r 兲 Y W 共 ␮ r 兲 ⫽ 41 L 2 关 2J W 共 ␮ L 兲 Y W 共 ␮ L 兲 ⫺J W⫺1 共 ␮ L 兲 Y W⫹1 共 ␮ L 兲

⫺J W⫹1 共 ␮ L 兲 Y W⫺1 共 ␮ L 兲兴 ⫺ 共 L→a 兲

⬅h l 共 L 兲 ⫺h l 共 a 兲 . 共5.14兲

There remains limL→⬁ 兺 l h l (L)sin 2⌬l(␮). For 兩 l 兩 Ⰷ⌽/2␲ ,

h l共 L 兲 ⫽
兩l兩
␲␮ 2 ⫹O
1
兩l兩冉 冊
, 共5.15兲

which, together with Eq. 共5.12兲, implies h l (L)sin 2⌬l(␮) is dominated by a term whose
l-dependence is ( ␮ a/2) 2 兩 l 兩 / 关 ( 兩 l 兩 ⫺1)! 兴 2 for all finite L and ␮ aⰆ1. Therefore the series
兺 l h l (L)sin 2⌬l is uniformly and absolutely convergent for all finite values of L⬎a. But it does not
necessarily converge to a function continuous at the point L⫽⬁ since h l (L) is not continuous at
L⫽⬁. In fact, for fixed l and ␮ LⰇ1,

h l 共 L 兲 ⫽⫺
sin共 2 ␮ L⫺ ␲ W 兲
2␲␮2
⫹O
1
L
. 冉冊 共5.16兲

The remedy is now clear. Consider instead the two series

兺l 冋 h l共 L 兲 ⫹
sin共 2 ␮ L⫺ ␲ W 兲
2␲␮ 2 册
sin 2⌬ l ⫺
1
2␲␮2 兺l sin共 2⌬ l 兲 sin共 2 ␮ L⫺ ␲ W 兲 .

Now the limit L→⬁ and the sum can be interchanged in the first series, giving zero. The second
series is bounded for all L⬎a, and so the limit in Eq. 共5.1兲 is true for the J W Y W terms.
The special case ⌽/2␲ ⫽1,2,... requires Eq. 共4.7兲 when l⫽⌽/2␲ . This results in a term
J 0 (kr)Y 0 (kr)/ln(ka), which is continuous for k on 关 0,M 兴 . Therefore the weighted mean value
theorem may be applied again to the k-integral, resulting in an overall factor of ln关(M 2
⫹m2)/m2兴. There remains an integration over rJ 0 ( ␮ r)Y 0 ( ␮ r) between a and L, giving an oscil-
lating but bounded term sin(2␮L) as L→⬁. Again, Eq. 共5.1兲 is satisfied for this special term.
Finally, consider the last two terms in Eq. 共5.5兲. Note that

兩Y W
2
共 kr 兲 ⫺J W
2
共 kr 兲 兩 ⭐Y W
2
共 ka 兲 ⫹J W
2
共 ka 兲 , 共5.17兲

2 2
for r⭓a since J W (z)⫹Y W (z) is a decreasing function of z.23 Hence, for kaⰆ1, 0⭐k⭐M ,

共 M a/2兲 2W
兩Y W
2
共 kr 兲 ⫺J W
2
共 kr 兲 兩 sin2 ⌬ l ⭐ 共 e a 储 B 储 / 冑␲ ⫹1 兲 2 共 1⫹O 共共 M a 兲 2 , 共 M a 兲 2W 兲兲 , 共5.18兲
⌫ 2 共 W⫹1 兲

so that the sum over l of the terms on the left-hand side of Eq. 共5.18兲 converges uniformly for all
r⭓a, 0⭐k⭐M . As it is also a sum of continuous functions of k for 0⭐k⭐M it can be integrated
term by term over k. Application of the weighted mean value theorem gives

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 J. Math. Phys., Vol. 41, No. 4, April 2000 Chiral limit of the two-dimensional fermionic . . . 1709

冕0
⬁ 2
dte ⫺tm lim
L→⬁
冕 冕
L

a
drr
M

0
dkk 兺l 关 Y W2 共 kr 兲 ⫺J W2 共 kr 兲兴 sin2 ⌬ l共 k 兲
1
⫽ ln
2 冉
M 2 ⫹m 2
m2
lim
L→⬁
冊 冕 a
L
drr 兺l 关 Y W2 共 ␮ r 兲 ⫺J W2 共 ␮ r 兲兴 sin2 ⌬ l共 ␮ 兲 , 共5.19兲

for some ␮ in the interval 0⬍ ␮ ⭐M . For 兩 l 兩 Ⰷ⌽/2␲ each term in the series Eq. 共5.19兲 is domi-
nated by a r-independent constant whose l-dependence is ( ␮ a/2) 2 兩 l 兩 /( 兩 l! 兩 ) 2 , for r⭓a and ␮ a
Ⰶ1. It is therefore a uniformly convergent series of continuous functions for all finite L⬎a that
can be integrated term by term. Entry 5.54.2 in Ref. 21 gives

冕a
L
2
drr 关 Y W 共 ␮ r 兲 ⫺J W
2
共 ␮ r 兲兴 ⫽
L2 2
2
关 Y W 共 ␮ L 兲 ⫺Y W⫺1 共 ␮ L 兲 Y W⫹1 共 ␮ L 兲 ⫺J W
2
共␮L兲

⫹J W⫹1 共 ␮ L 兲 J W⫺1 共 ␮ L 兲兴 ⫺ 共 L→a 兲 ⬅k l 共 L 兲 ⫺k l 共 a 兲 . 共5.20兲

Next, consider limL→⬁ 兺 l k l (L)sin2 ⌬l(␮). For ␮ LⰇ1,

k l共 L 兲 ⫽
cos共 2 ␮ L⫺ ␲ W 兲
␲␮2
⫹O
1
L
, 冉冊 共5.21兲

and hence consider the series

兺l 冋 k l共 L 兲 ⫺
cos共 2 ␮ L⫺ ␲ W 兲
␲␮ 2 册
sin2 ⌬ l ⫹
1
␲␮2 兺l cos共 2 ␮ L⫺ ␲ W 兲 sin2 ⌬ l .

The first series is a sum of continuous functions for all L⬎a. It is also a uniformly and absolutely
convergent series for L⬎a, first because 兩 J W 兩 ⭐1 and, second, the combinations Y W 2
sin2 ⌬l and
Y W⫺1 Y W⫹1 sin ⌬l are dominated by L-independent constants whose l-dependence is
2

( ␮ a2) 2 兩 l 兩 /( 兩 l 兩 !) 2 for 兩 l 兩 Ⰷ⌽/2␲ . Hence the limit L→⬁ and the sum can be interchanged in the
first series, giving zero. The second series is bounded for all L⬎a, verifying Eq. 共5.1兲 for the last
series of terms in Eq. 共5.5兲.
The special case ⌽/2␲ ⫽1,2... is dealt with in the same way as in the case of J W Y W sin ⌬l and
gives a contribution that vanishes in the limit indicated in Eq. 共5.1兲. Thus Eq. 共1.1兲 is demon-
strated.

ACKNOWLEDGMENTS
The author has benefited from discussions with C. Adam concerning the relation between
massive QED2 and the Schwinger model. He also gratefully acknowledges support from a Forbairt
Basic Research Grant.

1
M. P. Fry, Phys. Rev. D 54, 6444 共1996兲.
2
J. Schwinger, Phys. Rev. 128, 2425 共1962兲.
3
M. P. Fry, Phys. Rev. D 51, 810 共1995兲.
4
J. E. Avron and R. Seiler, Phys. Rev. Lett. 42, 931 共1979兲.
5
E. Seiler, Phys. Rev. D 22, 2412 共1980兲.
6
B. Simon, Adv. Math. 24, 244 共1977兲.
7
M. P. Fry, Phys. Rev. D 45, 682 共1992兲.
8
W. Heisenberg and H. Euler, Z. Phys. 98, 714 共1936兲; V. Weisskopf, K. Dan. Vidensk. Selsk. Mat. Fys. Medd. 14, No.
6 共1936兲.
9
J. Schwinger, Phys. Rev. 82, 664 共1951兲.
10
G. Dunne and T. Hall, Phys. Lett. B 419, 322 共1998兲.
11
H. W. Lee, P. Y. Pac, and H. K. Shin, Phys. Rev. D 40, 4202 共1989兲; V. P. Gusynin and I. A. Shovkovy, Can. J. Phys.
74, 282 共1996兲.
12
M. P. Fry, Phys. Rev. D 55, 968 共1997兲.
13
M. P. Fry, Phys. Rev. D 56, 6714 共1997兲.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18
 1710 J. Math. Phys., Vol. 41, No. 4, April 2000 M. P. Fry

14
C. R. Gattringer, I. Hip, and C. B. Lang, Nucl. Phys. B 508, 329 共1997兲.
15
M. P. Fry, Phys. Rev. D 47, 2629 共1993兲.
16
Y. Aharonov and A. Casher, Phys. Rev. A 19, 2461 共1979兲.
17
R. Musto, L. O’Raifeartaigh, and A. Wipf, Phys. Lett. B 175, 433 共1986兲.
18
K. Chandon, N. N. Khuri, A. Martin, and T. T. Wu, Phys. Rev. D 58, 025014 共1998兲.
19
H. P. Stapp, T. J. Ypsilantis, and N. Metropolis, Phys. Rev. 105, 302 共1957兲.
20
A. P. Prudnikov, Yu. A. Brychkov, and O. I. Marichev, Integrals and Series 共Gordon and Breach, New York, 1988兲, Vol.
2.
21
I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and Products 共Academic, New York, 1965兲.
22
E. T. Whittaker and G. N. Watson, A Course of Modern Analysis, 4th ed. 共Cambridge University Press, Cambridge,
England, 1958兲.
23
G. N. Watson, A Treatise on the Theory of Bessel Functions, 2nd ed. 共Cambridge University Press, Cambridge, England,
1958兲.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
84.88.136.149 On: Fri, 05 Dec 2014 10:11:18