Eikonal Approximation

K. V. Shajesh

Department of Physics and Astronomy, University of Oklahoma,

440 W. Brooks St., Norman, OK - 73019, U. S. A.
This is a study report on eikonal approximations which was undertaken under the guidance
of Kimball Milton. This report was submitted to partially fulll the requirements for the general
examination.
Contents
I Introduction 1
II Scattering in quantum mechanics 3
A Formulation of the scattering problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
B Identities satised by the scattering amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1 Dynamical reversibility theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Optical theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
III Partial wave expansion 6
A Method of partial wave expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
B Scattering length of a potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
C Inverted nite spherical well potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
IV Eikonal approximation in quantum mechanics 9
A Formalism for the eikonal approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
B Validity of the eikonal approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
C Zeroth order eikonal approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
D Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1 Inverted nite spherical well potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2 Yukawa potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
V Quantum electrodynamics 15
A Schwingers functional dierential equations for quantum electrodynamics . . . . . . . . . . . . . . 15
B A formal solution to Schwingers functional dierential equations . . . . . . . . . . . . . . . . . . . 16
C Electron-electron scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
VI Approximation to electron-electron scattering 19
A Quenched or ladder approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
B Eikonal approximation to G
+
[x, y; A

] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
I. INTRODUCTION
The etymology of the word eikon traces it back to the word eikenai which is the transliteration of the word
[1] in the Greek language meaning to resemble [2]. In the Greek language it evolved into the word eik on which is
the transliteration of the word o [1] meaning image [2]. It was borrowed into the Latin language and later into
the English language as eikon. In the English language it transformed into the words icon and ikon which are the
variants of the word eikon. Notice that throughout its evolution it has meant image.
The main goal of this report is to understand the eikonal approximation in quantum mechanics and quantum
eld theories. Approximations play a very important role in the understanding of processes that cannot be solved

Email: [email protected]
1
exactly. The Born approximation in quantum mechanics is an example of an approximation that has been extensively
used for studying low energy processes. In quantum eld theories involving coupling constants smaller than one
we use the standard weak-coupling perturbation series which is parallel in its approach to the Born approximation
in quantum mechanics. In the 1950s and 1960s when high energy physics was ascending towards its peak, it was
realized among the high energy physicists of those times that the Born approximation is not a valid approximation for
studying processes involving high energies. This period in fact was the golden age in the development of the eikonal
approximation in quantum mechanics and quantum eld theories. The descendents of this era took the theory of
eikonal approximation for granted in quantum mechanics and quantum eld theories. There was prolic activity in
the application of eikonal approximation in high energy physics, especially in QCD.
The eikonal approximation was not born in the study of quantum mechanics. It originated far back in optics.
Light we know obeys Maxwells equations. In terms of Maxwells equations light is understood as a wave obeying a
wave equation. But two hundred years before Maxwell wrote down his equations scientists understood reection and
refraction of light which was extensively studied under the branch of science called ray optics. Today in elementary
school we learn ray optics without ever introducing Maxwells equations. In ray optics we assume that light travels
in a straight line. This assumption works ne as long as the size of the obstacle is large compared to the wavelength
of light. This is called the eikonal approximation in optics. Here we make contact with the meaning of the word
eikon in the sense that eikon meaning image is formed by light only in the straight line approximation. In processes
involving diraction we encounter the limit of the validity of the eikonal approximation in optics. We quickly switch
over to Maxwells equations to study diraction.
Optics is described by Maxwells equations which can be written as a wave equation with the dispersion relation
given by = kc. Similarly, quantum mechanics is described in terms of Schrodingers equation which is a diusion
equation (in imaginary time) with a dispersion relation given by =
hk
2
2m
. With this correspondence we ask the
question, can we not have a corresponding eikonal approximation in quantum mechanics? Yes we can. The eikonal
approximation in quantum mechanics works for processes involving the scattering of particles with large incoming
momentum and when the scattering angle is very small.
In the language of dierential equations, the main advantage the eikonal approximation oers is that the equations
reduce to a dierential equation in a single variable. This reduction into a single variable is the result of the straight
line approximation or the eikonal approximation which allows us to choose the straight line as a special direction.
The early steps involved in the eikonal approximation in quantum mechanics are very closely related to the WKB
approximation in quantum mechanics. The WKB approximation involves an expansion in terms of Plancks constant
h. The WKB approximation also reduces the equations into a dierential equation in a single variable. But the
complexity involved in the WKB approximation is that this variable is described by the trajectory of the particle
which in general is complicated. The advantage of the eikonal approximation is in the classical trajectory being a
straight line. Thus in this manner the eikonal approximation is a very stringent semi-classical limit.
A very comprehensive collection of work on scattering theory in general with a very extensive bibliography which
covers scattering theory in both electromagnetism and quantum mechanics is the book by Roger G. Newton [3]. A
couple of textbooks among many which I have used are [6] and [7].
An extensive list of literature on the theory of eikonal approximation in quantum mechanics and quantum eld
theories developed between the years 1950 and 1970 is available in [4]. It is unanimously accepted by everyone that
R. J. Glaubers lecture notes on eikonal approximation [5] is the best available work on the subject. Glaubers lecture
notes addresses the question of the conditions for the validity of the eikonal approximations in quantum mechanics.
In the section on eikonal approximations in this article starting out from his ideas we derive the conditions for the
validity of the eikonal approximations more concretely. We verify our conclusions by evaluating the total scattering
cross section for the simplest potential (an inverted nite spherical well) using the eikonal approximation and by
comparing it with the result obtained by the partial wave expansion.
After our study of eikonal approximation in quantum mechanics we concentrate on quantum electrodynamics. To
study the eikonal approximation in quantum electrodynamics we need to rst write the eld theoretical equations in a
suitable form. Schwingers formalism for quantum electrodynamics is the most suitable for this purpose. We develop
Schwingers formalism in detail based on [1518] in section V.
In section VI we concentrate on electron-electron scattering. We introduce the quenched approximation. We outline
the derivation of the electron-electron scattering in the eikonal approximation based on [19,20]. We point out the
similarity of the result of electron-electron scattering obtained in the eikonal approximation to the non-relativistic
scattering amplitude due to the Coulomb potential.
The terminology used in this study report is the following
f(, ) Scattering amplitude

scatt.
Scattering cross section
2

abs.
Absorption cross section

tot.

scatt.
+
abs.
.
II. SCATTERING IN QUANTUM MECHANICS
A scattering process in quantum mechanics is described by the solution of the Schroedinger equation
_

h
2
2m

2
+V (r, t)
_

(r, t) = i h

t

(r, t) (1)
with the boundary conditions dictated by the requirement that the wave function

(r, t) must have a component that

involves an incident plane wave with energy E =
h
2
k
2
2m
moving in the positive z direction and another component that
involves a spherical outgoing wave. It should be emphasized that in scattering problems we do not require

(r, t) to
go to zero at r . In fact the scattering amplitude which is the quantity of interest is contained in the r
(asymptotic) part of

(r, t).
A. Formulation of the scattering problem
For the case when the initial beam can be described by a state of denite energy we can write

(r, t) = e

i
h
Et
(r) (2)
where (r) satises the dierential equation
_

2
+k
2

(r) = U(r)(r) (3)

with the boundary conditions on (r) dictated by the boundary conditions on

(r, t). We have used the notation

k
2
=
2mE
h
2
and U(r) =
2mV (r)
h
2
.
The dierential equation for (r) in eqn. (3) can be rewritten as an integral equation [8] given by
(r) = (r) +
_
d
3
r

G
0
(r, r

)U(r

)(r

) (4)
where (r) satises the potential free equation
_

2
+ k
2

(r) = 0 (5)
and the Greens function G
0
(r, r

) is the solution to
_

2
+k
2

G
0
(r, r

) =
(3)
(r r

). (6)
The boundary conditions on (r) and G
0
(r, r

) are prescribed by the boundary conditions on (r). The integral

equation in eqn. (4) is called the Lippmann-Schwinger equation [9]
1
. The solutions to (r) and G
0
(r, r

) are
(r) = A
0
e
i

kr
+B
0
e
i

kr
(7)
G
0
(r, r

) =
1
4
_
A
e
ik|rr

|
[r r

[
+B
e
ik|rr

|
[r r

[
_
(8)
1
Schwingers view of the paper was [10]: . . . I thought the importance of the paper was the variational principle . . . the so-
called Lippmann-Schwinger scattering equation is to me conventional scattering theory written in operator notation. Nothing
new. But that is what everybody paid attention to.
3
with the constraint A+B = 1. Using eqn. (7) and eqn. (8) in eqn. (4) the Lippmann-Schwinger equation takes the
form
(r) = A
0
e
i

kr
+B
0
e
i

kr

1
4
_
d
3
r

_
A
e
ik|rr

|
[r r

[
+B
e
ik|rr

|
[r r

[
_
U(r

)(r

). (9)
Imposing the boundary conditions that the wave function consists of a component that is a plane wave moving in the
positive z direction and another that is an outgoing spherical wave we get
(r) = A
0
e
i

kr

1
4
_
d
3
r

e
ik|rr

|
[r r

[
U(r

)(r

). (10)
As we emphasized before the information related to the scattering amplitude is contained in the asymptotic region
of the wave function. In most of the problems of interest the potential V (r) is conned to a nite volume in space
and the detectors are very far from the region containing the potential. For these cases we can safely conclude r

r
and thus approximate
[r r

[ = r
r r

r
+O
_
_
r

r
_
2
_
. (11)
We can thus write

r
(r) = A
0
e
i

kr

1
4
_
d
3
r

1
r
e
ik(r
rr

r
)
U(r

)(r

) (12)
= e
i

kr
+f(, )
e
ikr
r
(13)
where we have set A
0
= 1 so that
f(, ) =
1
4
_
d
3
r

e
i

U(r

)(r

) (14)
can be interpreted as the scattering amplitude, where

k

= k
r
r
. The illustration of the variables involved is shown in
g. 1.

:
-

k, z

= k
r
r
x
y

r = (r sin cos , r sin sin, r cos )

= (k sin cos , k sin sin, k cos )

k = (0, 0, k)
r

= (b

cos

, b

sin

, z

)
1
FIG. 1. Illustration of the various variables used in the calculation. Note that r,

, and

k are in spherical polar coordinates
and r

is in cylindrical polar coordinates.

Often it is more useful to denote f(, ) as a function of

k and

k

and thus write f(, ) = f(

k). Observe that

even though the information related to f(, ) is contained in the asymptotic region of (r) the only contributions to
f(, ) in eqn. (14) comes from regions where the potential is not zero. To complete the formulation of the scattering
problem we dene the scattering cross section as

scatt.
=
_
d [ f(, ) [
2
(15)
4
and the total cross section as

tot.
=
scatt.
+
abs.
(16)
where
abs.
is the absorption cross section which will be explicitly dened in eqn. (27).
To summarize this section on formulation of the scattering problem, we have concluded that a prescription to
evaluate the scattering amplitude is to use
f(, ) =
1
4
_
d
3
r

e
i

U(r

)(r

) (17)
where (r) is determined by solving
_

2
+k
2

(r) = U(r)(r) (18)

under the boundary conditions described after eqn. (1).
B. Identities satised by the scattering amplitude
Starting from the Schroedinger equation we shall derive two identities satised by the scattering amplitude. The
rst, the dynamical reversibility theorem was rst derived by R. J. Glauber and V. Schomaker [11] in 1953. The
second, the optical theorem, is a statement of probability conservation in quantum mechanics written as a continuity
equation in reference to the scattering amplitude. The optical theorem was rst derived by E. Feenberg [12] in 1932.
1. Dynamical reversibility theorem
In the Schroedinger equation if we write the wave function as

(r, t;

k) = e

i
h
Et
(r;

k) (19)
where 2mE = h
2
k
2
, we can quickly derive the following identity
(r;

k
2
)
2
(r;

k
1
) (r;

k
1
)
2
(r;

k
2
) = (k
2
1
k
2
2
)(r;

k
1
)(r;

k
2
). (20)
Evaluating the above expression on a sphere of radius r for the case [

k
1
[=[

k
2
[ we have
_
r
d

S
_

r
(r;

k
2
)
2

r
(r;

k
1
)
r
(r;

k
1
)
2

r
(r;

k
2
)
_
= 0. (21)
Using eqn. (13) in the above expression and evaluating the surface integrals after taking the limit r we get [5]
f(

k
2
,

k
1
) = f(

k
1
,

k
2
) (22)
where we have used the notation we dened for f(, ) in the paragraph after eqn. (14). Thus the scattering amplitude
for a process going from

k
1
to

k
2
is identical to that of a process going from

k
2
to

k
1
. This is dynamical reversibility.
2. Optical theorem
In a very similar manner as we did in the earlier section (in this case we take the complex conjugate of the wave
function) we arrive at the following continuity equation
(r)
t
+

j(r) = s(r) (23)

where
5
(r) = [ (r) [
2

j(r) =
h
2im
_

(r)

(r) (r)

(r)
_
s(r) =
2
h
[Im V (r)] [ (r) [
2
. (24)
Integrating the continuity equation over a sphere of radius r and noting that the derivative of the term does
not contribute because both the states have the same energy we get
_
r
d

S
_

r
(r)
2

r
(r)
r
(r)
2

r
(r)

= lim
r
_
d
3
r s(r). (25)
Again using eqn. (13) in the above expression and evaluating the surface integrals after taking the limit r we
get [5]
_
2
0
d
_

0
sin d [ f(, ) [
2
lim
r
2m
h
2
k
_
d
3
r [Im V (r)] [ (r) [
2
=
4
k
Imf( = 0). (26)
Using eqn. (15) and dening the absorption scattering cross section as

abs.
= lim
r
2m
h
2
k
_
d
3
r [Im V (r)] [ (r) [
2
(27)
we have the optical theorem

scatt.
+
abs.
=
4
k
Imf( = 0). (28)
III. PARTIAL WAVE EXPANSION
In this article we will be concerned with approximation methods for evaluating the scattering amplitude. In
particular we will be interested in the eikonal approximation to the scattering amplitude. After we have made an
approximation we would like to be aware of how much we have deviated from the exact result. The closest we can
have to an exact result in a scattering problem is the result got by the method of partial wave expansion. We shall
thus nd it very useful to use the results obtained from the method of partial wave expansion as a benchmark.
The method of partial wave expansion breaks down the initial wavefunction into an innite sum over angular
momentum components labeled by the quantum number l. The contribution to the scattering amplitude from each l
is calculated separately and called the partial wave scattering amplitude. The complete scattering amplitude is then
obtained by summing over all the partial wave scattering amplitudes. We say it is the closest we can have to an
exact expression because the solution cannot in general be written in a closed form. Nevertheless for an incoming
beam of a given energy characterized by the parameter k we can always dene its angular momentum as ka where
a is the scattering length dened in section III B in terms of the partial wave scattering amplitude corresponding to
l = 0. It turns out that for most of the potentials the contributions from l terms very much higher than ka converge
very rapidly and thus can be neglected. Physically this is understood as, only those beams that have sucient initial
energy and those that pass through the scattering length of the potential are deected.
If the method of partial wave expansion gives us the exact solution why do we need approximation methods?
Firstly because for almost all potentials (even for the simple Gaussian potential) we need to depend on numerical
computational methods to solve the radial dierential equation. Secondly for high energy scattering problems we need
to sum over a large range of l.
For potentials involving a boundary beyond which the potential is zero it sometimes becomes possible to evade the
need for computational methods to a signicant extent. For the simplest potential of this class, that of an inverted
nite square well, the radial dierential equation mentioned above can be solved exactly. Thus we shall solve the
inverted nite square well in detail and use it for comparison with the results obtained from other approximation
methods.
6
A. Method of partial wave expansion
For a spherically symmetric potential of the form
V (r) = V (r) (29)
we can write the solution to eqn. (18) as
(r) =

l=0
(2l + 1)i
l
R
l
(r)P
l
(cos ) (30)
where where R
l
(r) is the solution to the radial dierential equation
d
2
R
l
(r)
dr
2
+
2
r
dR
l
(r)
dr
+
_
k
2

l(l + 1)
r
2
U(r)
_
R
l
(r) = 0. (31)
For potentials that die o faster than
1
r
2
we can write the solution to eqn. (31) in the r region to be
lim
r
R
l
(r) = C
l
_
cos
l
_
lim
r
j
l
(kr)
_
sin
l
_
lim
r

l
(kr)
__
. (32)
Substituting eqn. (32) in eqn. (30) and using the identity
e
ikr cos
=

l=0
(2l + 1)i
l
j
l
(kr)P
l
(cos ) (33)
in eqn. (14) we have
f(, )
e
ikr
r
=

l=0
(2l + 1)i
l
P
l
(cos )
_
(C
l
cos
l
1)
_
lim
r
j
l
(kr)
_
C
l
sin
l
_
lim
r

l
(kr)
__
. (34)
Using
lim
r
j
l
(kr) =
1
kr
cos
_
kr (l + 1)

2
_
=
1
2kr
_
i
(l+1)
e
ikr
+i
+(l+1)
e
ikr
_
(35)
lim
r

l
(kr) =
1
kr
sin
_
kr (l + 1)

2
_
=
1
2kr
_
i
(l+2)
e
ikr
i
l
e
ikr
_
(36)
and equating the coecients of e
ikr
and e
ikr
in eqn. (34) we get
f() =
1
2k

l=0
(2l + 1)i
l
P
l
(cos )i
(l+1)
_
C
l
e
+i
l
1

(37)
0 =
1
2k

l=0
(2l + 1)i
l
P
l
(cos )i
+(l+1)
_
C
l
e
i
l
1

. (38)
The second equation above further reduces to C
l
= e
i
l
which when used in eqn. (37) to eliminate C
l
gives
f() =
1
2ik

l=0
(2l + 1)P
l
(cos )
_
e
2i
l
1

. (39)
This is the standard expression for the scattering amplitude given as a sum of partial waves.
l
s are dened as the
phase shifts in Bessel functions (which are trigonometric functions in the r limit) as introduced in eqn. (32).
Observe that the phase shifts
l
s in eqn. (39) are in principle determined by solving the radial dierential equation
in the presence of U(r) which can be solved analytically only for special cases. In practice the phase shifts are
evaluated by solving the radial equation numerically. For problems involving high energies a further complication
arises because we need to sum over a suciently high number of partial waves.
For the case of a inverted square well potential it is possible to solve the radial equation exactly. In the subsequent
section we shall write down an expression for
l
for the inverted square well potential. We shall nd it useful to
compare the results later when we are doing eikonal approximations.
7
B. Scattering length of a potential
The scattering cross section is obtained by squaring eqn. (39) and integrating over all directions. Using the
orthogonality of Legendre polynomials it turns out to be

scatt.
=
4
k
2

0
(2l + 1) sin
2

l
. (40)
The contribution to the scattering cross section from the s-wave (l = 0) is

scatt.

l=0
=
4
k
2
sin
2

0
. (41)
The scattering length a is dened as the radius of the sphere that will give a contribution in a classical scattering
problem equal to the s-wave contribution above. Correspondingly we write

scatt.

l=0
= 4a
2
. (42)
Thus we get
a =
1
k
sin
0
. (43)
Observe that sin
0
will require the solution to the dierential equation in eqn. (31) for l = 0 which normally can only
be solved numerically.
C. Inverted nite spherical well potential
For a spherically symmetric potential of the form
V (r) =
_
V (r) for r < a
0 for r > a
(44)
we can write
R
l
(r) =
_
R
l<
(r) for r < a
C
l
[cos
l
j
l
(kr) sin
l

l
(kr)] for r > a
(45)
where R
l<
(r) is the solution to eqn. (31) for r < a. Requiring the wave function and its derivative to be continuous
and taking the ratio of the equations got by requiring the continuity we get
a
dR
l<
(r)
dr

r=a
R
l<
(a)
=
ka
_
d
dka
j
l
(ka) tan
l
d
dka

l
(ka)

[j
l
(ka) tan
l

l
(ka)]
. (46)
The above expression can be solved for tan
l
if R
l<
(r) is known at r = a.
For an inverted square well potential dened by
V (r) =
_
V for r < a
0 for r > a
(47)
we further have
R
l<
(r) = B
l
j
l
(r) (48)
where = k
_
1
V
E
with
V
E
< 1. For this case we can write the explicit expression for the phase shifts to be
sin
2

l
=
N
2
(ka, a)
N
2
(ka, a) +D
2
(ka, a)
(49)
where
N(ka, a) = a j
l1
(a)j
l
(ka) ka j
l
(a)j
l1
(ka)
D(ka, a) = a j
l1
(a)
l
(ka) ka j
l
(a)
l1
(ka). (50)
In g. 2 we plot
tot.
versus ka
V
E
for the inverted spherical well potential for ka = 500.
8
20 40 60 80 100
0.5
1
1.5
2
2.5
3
3.5
partial wave (ka = 500)

tot.
a
2
ka
V
E
FIG. 2. tot. versus ka
V
E
for ka = 500 using partial wave method.
IV. EIKONAL APPROXIMATION IN QUANTUM MECHANICS
In this section we shall convince ourselves that the eikonal approximation is valid for processes involving small angle
scattering and very large incoming momentum. More rigorously the conditions are
V
E
1 and
1
V/E
ka
1
(V/E)
2
.
In this stringent parameter zone the expression for scattering amplitude takes the form
f(, ) =
k
2i
_
d
2
b

e
i

_
e
i(

)
1
_
(51)
where
(

) =
1
2k
2m
h
2
_
+

dz

V (

, z

). (52)
The scattering cross section takes the form

scatt.
= 8a
2
_

0
dt t sin
2
_
k
E
_
+

dz

V (

, z

)
_
. (53)
We shall derive the above formulas in this section.
A. Formalism for the eikonal approximation
In eqns. (17) and (18) we found the prescription for calculating the scattering amplitude to be
f(, ) =
1
4
_
d
3
r

e
i

U(r

)(r

) (54)
where (r) is determined by solving
_

2
+k
2

(r) = U(r)(r). (55)

To solve for (r) let us write
(r) = e
i

kr
(r). (56)
With the above substitution and with the choice

k along the z direction eqn. (55) takes the form
9
_
2ik

z
U(

b, z)
_
(

b, z) =
2
(

b, z) (57)
where we have used the notation r (

b, z). We can write the formal solution to eqn. (57) as

(

b, z) = (

b, z)
_
d
2
b

dz

G
e
(

b, z,

, z

2
(

, z

) (58)
where (

b, z) satises
_
2ik

z
U(

b, z)
_
(

b, z) = 0 (59)
and G
e
(

b, z,

, z

) satises
_
2ik

z
U(

b, z)
_
G
e
(

b, z,

, z

) =
(2)
(

) (z z

) (60)
with boundary conditions on (

b, z) and G
e
(

b, z,

, z

) prescribed by the boundary conditions on (r). The solutions

to eqns. (59) and (60) are
(

b, z) = e
1
2ik
_
z

du U(

b,u)
(61)
where we imposed the boundary condition
0
(

b) = (

b, z = ) = 1, and
G
e
(

b, z,

, z

) =
1
2ik

(2)
(

) (z z

) e
1
2ik
_
z
z

du U(

b,u)
(62)
=
1
2ik

(2)
(

) (z z

) (

b, z)
1
(

b, z

). (63)
Using eqns. (61) and (63) in eqn. (58) we have
(

b, z) = (

b, z)
_
1
1
2ik
_
z

dz

1
(

b, z

)
_

2
b
+

2
z

2
_
(

b, z

)
_
(64)
which after iteration takes the form
(

b, z) = (

b, z)
_
1 +
_
z

dz

K(

b, z

b
,

z

) +
_
z

dz

K(

b, z

b
,

z

)
_
z

dz

K(

b, z

b
,

z

) +. . .
_
(65)
where the expression for K(

b, z,

b
,

z
) acting on an arbitrary function g(z) is given by
K(

b, z,

b
,

z
) g(z) =
1
2ik

1
(

b, z)
_

2
b
+

2
z
2
_
(

b, z) g(z). (66)
Using the above series expansion for (

b, z) in eqn. (56) and plugging it in eqn. (54) we can write the scattering
amplitude as
f(, ) = f
(0)
(, ) +f
(1)
(, ) +f
(2)
(, ) +. . . (67)
where
f
(0)
(, ) =
1
4
_
d
2
b

_
+

dz

e
i(

)r

U(

, z

) (

, z

)
f
(1)
(, ) =
1
4
_
d
2
b

_
+

dz

e
i(

)r

U(

, z

) (

, z

)
_
z

dz

K(

, z

)
f
(2)
(, ) =
1
4
_
d
2
b

_
+

dz

e
i(

)r

U(

, z

) (

, z

)
_
z

dz

K(

, z

)
_
z

dz

K(

, z

). (68)
We have used K(

b, z,

b
,

z
) K(

b, z) for compactness. The power in the exponent i(

) r

in the above
expressions can be explicitly written as
i(

) r

= ikb

sin cos(

) +ikz

2 sin
2

2
. (69)
The coordinates used are pictorially described in g. 1.
10
B. Validity of the eikonal approximation
Starting from the expression for f
(0)
(, ) in eqn. (68) and switching to dimensionless variables z at,

b a w
and V (

b, z) V v(

b, z) where a is the scattering length, we can write

f(, ) =
a
4
(ka)
2
V
E
_
g
(0)
(ka) +
V
E
g
(1)
A
(ka) +ka
_
V
E
_
2
g
(1)
B
(ka) +. . .
_
(70)
where we have suppressed the and dependence in g
(n)
s. The explicit expressions for the g
(0)
, g
(1)
A
and g
(1)
B
are
g
(0)
=
_
d
2
w

_
+

dt

e
ikaw

sin cos(

)+ikat

2 sin
2
2
v( w

, t

) ( w

, t

)
[ g
(0)
[
_
d
2
w

_
+

dt

[v( w

, t

)[ (71)
g
(1)
A
=
1
4
_
d
2
w

_
+

dt

e
ikaw

sin cos(

)+ikat

2 sin
2
2
v( w

, t

)
2
( w

, t

1
4
_
d
2
w

_
+

dt

e
ikaw

sin cos(

)+ikat

2 sin
2
2
v( w

, t

) ( w

, t

)
_
t

dt

_
t

dt

2
w
v( w

, t

[ g
(1)
A
[
1
4
_
d
2
w

_
+

dt

[v( w

, t

)[
2
+
1
4
_
d
2
w

_
+

dt

[v( w

, t

)[
_
t

dt

_
t

dt

2
w
v( w

, t

g
(1)
B
=
i
8
_
d
2
w

_
+

dt

e
ikaw

sin cos(

)+ikat

2 sin
2
2
v( w

, t

) ( w

, t

)
_
t

dt

v( w

, t

)
2

i
8
_
d
2
w

_
+

dt

e
ikaw

sin cos(

)+ikat

2 sin
2
2
v( w

, t

) ( w

, t

)
_
t

dt

_
_
t

dt

2
w
v( w

, t

)
_
2
[ g
(1)
B
[
1
8
_
d
2
w

_
+

dt

[v( w

, t

)[
_
t

dt

[v( w

, t

)[
2
+
1
8
_
d
2
w

_
+

dt

[v( w

, t

)[
_
t

dt

_
t

dt

2
w
v( w

, t

2
. (72)
It is easy to verify from the above expressions that for potentials that fall away faster than
1
t

w
the contributions to
g
(n)
s are nite. For these potentials it can thus be concluded that major contribution to the scattering amplitude
comes from the zeroth order for
V
E
1 and ka
_
V
E
_
2
1.
We shall later observe that the expression for the scattering amplitude obtained from the zeroth order eikonal
approximation satises the optical theorem only in the limit ka
V
E
. This requirement can be achieved only if we
additionally require ka
1
V/E
.
Thus to summarize, the conditions under which the eikonal approximation holds are
V
E
1 and
1
V/E
ka
1
(V/E)
2
. (73)
C. Zeroth order eikonal approximation
Thus for potentials that die out suciently fast and for conditions
V
E
1 and
1
V/E
ka
1
(V/E)
2
the zeroth
order contribution to f(, ) is the most signicant. From eqns. (67), (68) and (61) we can write
f
(0)
(, ) =
1
4
_
d
2
b

_
+

dz

e
i(

)r

U(

, z

) (

, z

)
=
1
4
_
d
2
b

_
+

dz

e
ikb

sin cos(

)+ikz

2 sin
2
2
U(

, z

) e
1
2ik
_
z

du U(

,u)
. (74)
11
For large incoming momentum and for small scattering angles most of the contribution to the scattering cross section
comes from angles less than or of order
V
E
. Thus in the zeroth order eikonal approximation we can approximate
ikb

sin cos(

) +ikz

2 sin
2

2
ikb

cos(

). (75)
Using eqn. (75) in eqn. (74) we have
f
(0)
(, ) =
1
4
_

0
b

db

_
2
0
d

e
ikb

cos(

)
_
+

dz

U(

, z

) e
1
2ik
_
z

du U(

,u)
. (76)
We notice that the approximation involved in eqn. (75) allows us to carry out the z integral by substituting
_
z

du U(

, u) for a new integration variable. After evaluating the z integral we get

f
(0)
(, ) =
k
2i
_

0
b

db

_
2
0
d

e
ikb

cos(

)
_
e
i(

)
1
_
(77)
where
(

) =
k
2
1
E
_
+

dz

V (

, z

). (78)
For the case when the potential is independent of variable such that the scattering is symmetric about the z-axis
we can further carry out the

integral and thus get

f
(0)
() =
k
i
_

0
b

db

J
0
(kb

)
_
e
i(b

)
1
_
(79)
where we have used the identity
J
0
(t) =
1
2
_
2
0
d e
it cos
(80)
and J
0
(t) = J
0
(t). For making the above expressions more illustrative we scale the integral variables with respect
to the scattering length of the potential dened in section III B by introducing the dimensionless integral variables u
and t dened as z = au and b = at, where a is the scattering length of the potential. We also use V to signify the
maximum value of the function V (r). In term of these variables scattering amplitude in eqn. (79) takes the form
f() = a
1
i
ka
_

0
t dt J
0
(t ka)
_
e
ika
V
E
(t)
1
_
(81)
where
(t) =
1
2
1
V
_
+

du V (at, au). (82)

Using eqn. (15) we have the expression for the scattering cross section to be

scatt.
a
2
= 2(ka)
2
_

0
t dt
_

0
t

dt

_
e
+ika
V
E
(t)
1
_ _
e
ika
V
E
(t

)
1
_
_

0
sin d J
0
(t ka)J
0
(t

ka). (83)
We said earlier that
V
E
is a good estimate for the upper limit of the scattering angle . Thus in the limit
V
E
1 we
can write sin and limit our range of integration in from 0 to
V
E
. Further introducing the integration variable
x = ka we have

scatt.
a
2
= 2
_

0
t dt
_

0
t

dt

_
e
+ika
V
E
(t)
1
_ _
e
ika
V
E
(t

)
1
_
_
ka
V
E
0
x dx J
0
(tx)J
0
(t

x). (84)
Once we have the above expressions for the scattering amplitude in eqn. (81) and scattering cross section in eqn.
(84) we would like to check if the optical theorem is satised in the eikonal approximation. For potentials that are
not complex the optical theorem can be stated as
12

scatt.
=
4
k
Im f(0). (85)
Using eqn. (81) in eqn. (85) we have

scatt.
a
2
= 8
_

0
t dt sin
2
_
ka
V
E
(t)
_
. (86)
Comparing eqn. (84) and eqn. (86) we conclude that under the conditions
V
E
1 and ka
_
V
E
_
2
1 alone the eikonal
approximation does not satisfy the optical theorem. The physical content of the optical theorem is the statement of
probability conservation in quantum mechanics. Can we save the eikonal approximation from violating the optical
theorem? Yes. We recognize that in eqn. (84) if we take the limit ka
V
E
and use the identity
_

0
x dx J
0
(tx)J
0
(t

x) =
(t t

)
t
(87)
we get exactly the expression required for the optical theorem to be satised. Thus we observe that even though
the expression for the scattering amplitude in the eikonal approximation is derived under the conditions
V
E
1 and
ka
_
V
E
_
2
1 alone, we still need to put an additional condition ka
V
E
1 for it to satisfy the optical theorem. Thus
we have the eikonal approximation valid under the conditions
V
E
1 and
1
V/E
ka
1
(V/E)
2
. (88)
For a given ka, the second inequality above puts an upper and lower bound on
V
E
given by
1
ka

V
E

1

ka
. This can
be rewritten in the form
1 ka
V
E

ka. (89)
D. Examples
1. Inverted nite spherical well potential
An inverted nite spherical well potential is dened as
V (r) =
_
V for r < a
0 for r > a
. (90)
Noting that b
2
+z
2
= r
2
and thus integrating z from 0 to

r
2
b
2
we have for this potential
(t) =
_
1 t
2
f() = a
1
i
ka
_
1
0
t dt J
0
(t ka)
_
e
ika
V
E

1t
2
1
_

scatt.
a
2
= 8
_
1
0
t dt sin
2
_
ka
V
E
_
1 t
2
_
. (91)
In gs. 3 and 4 we plot
scatt.
versus ka
V
E
and compare it with the plots got using the partial wave method. Using
eqn. (89) for ka = 50, we have the region of validity to be 1 ka
V
E
7, and for ka = 750, we have the region of
validity to be 1 ka
V
E
27. These estimates for the region of validity are in agreement with the region where the
curves obtained from the eikonal approximation t the curve obtained from partial wave method in gs. 3 and 4.
13
20 40 60 80 100
0.5
1
1.5
2
2.5
3
3.5
- - - partial wave (ka = 50)
eikonal

tot.
a
2
ka
V
E
FIG. 3. Comparison of eikonal and partial wave method (ka = 50) for the inverted nite spherical well.
20 40 60 80 100
0.5
1
1.5
2
2.5
3
3.5
- - - partial wave (ka = 750)
eikonal

tot.
a
2
ka
V
E
FIG. 4. Comparison of eikonal and partial wave method (ka = 750) for the inverted nite spherical well.
2. Yukawa potential
Consider the Yukawa potential
V (r) = V
e
r/a
r/a
(92)
where V =
e
2
a
and r
2
= b
2
+z
2
. For this potential we get
(t) =
1
2
_
+

du
e

t
2
+u
2

t
2
+u
2
=
_

0
d e
t cosh
= K
0
(t) (93)
where K
0
(t) is modied Bessel function of order zero. Using this in eqn. (81) we have
f() = a
1
i
ka
_

0
t dt J
0
(t ka)
_
e
ika
V
E
K0(t)
1
_
. (94)
14
The contribution to the above expression from J
0
(t ka) dies o very fast. Thus to a good approximation we can say
that we get non-zero contribution to the integral for t ka 1. Noting that in the eikonal approximation
<

V
E
and
1 ka
V
E

1
V/E
we conclude that non zero contributions to the integral comes from t 1. In the limit t 1 the
modied Bessel function takes the form
K
0
(t) ln
2
t
(95)
where = 0.577 . . . is the Eulers constant. Further we observe that one of the terms in eqn. (94) contributes only at
= 0 and thus is a delta function. Overall after substituting x = ka we get
f() = ika
2
() ie
ika
V
E
(ln2+ln(ka))
1
k
2
_

0
dxJ
0
(x)x
1ika
V
E
. (96)
In terms of the gamma functions we have
_

0
dxJ
n
(x)x
1+2i
= 2
2i+1
(1 +
n
2
+i)
(
n
2
i)
. (97)
Thus we can write
[ f() [
=0
=
1
2k
_

2
_
2
2
ka
V
E

_
1 i
2
ka
V
E
_

_
1 +i
2
ka
V
E
_ (98)
where
1
2
ka
V
E
=
e
2
hv
using V =
e
2
a
and hk = p = mv. Observe that the above result is identical to the scattering
amplitude due to a Coulomb potential for small scattering angles using the approximation sin . It is worth
observing the interesting fact that the above result is independent of a.
V. QUANTUM ELECTRODYNAMICS
A. Schwingers functional dierential equations for quantum electrodynamics
We dene the vacuum to vacuum persistence amplitude for electrodynamics in the presence of external sources
(x) J

(x), (x), (x) to be

Z[] = 0
+
[ 0

)
J
. (99)
Using Schwingers quantum variational principle we can write
Z[] = i 0
+
[ W[/; ] [ 0

)
J
(100)
where W[/; ] is the action for quantum electrodynamics given in terms of the elds /(x)
A

(x), F

(x), (x),

(x) interacting with external sources is (x)
W[/; ] = S[/] +
_
d
4
x
_
J

(x)A

(x) + (x)(x) +

(x)(x)

(101)
where
S[/] =
_
d
4
x
_

1
2
F

(x)(

(x)

(x)) +
1
4
F

(x)F

(x) +

(x)[i

m+e

(x)](x)
_
. (102)
Gauge invariance requires us to constrain the external source J

(x) to satisfy the condition

(x) = 0. Using
Schwingers quantum variational principle to vary Z[] with respect to the elds / and the external sources we
get
15
Z[] = i (x)0
+
[
W[/, ]
(x)
[ 0

)
J
+i (x)0
+
[
W[/, ]
(x)
[ 0

)
J
+i J

(x)0
+
[
W[/, ]
J

(x)
[ 0

)
J
+ i

(x)0
+
[
W[/, ]

(x)
[ 0

)
J
+i (x)0
+
[
W[/, ]
(x)
[ 0

)
J
+ i A

(x)0
+
[
W[/, ]
A

(x)
[ 0

)
J
+ i F

(x)0
+
[
W[/, ]
F

(x)
[ 0

)
J
= i (x)0
+
[ (x) [ 0

)
J
+i (x)0
+
[

(x) [ 0

)
J
+i J

(x)0
+
[ A

(x) [ 0

)
J
+ i

(x)
_
(i

m)0
+
[ (x) [ 0

)
J
+e0
+
[

(x)(x) [ 0

)
J
+(x)0
+
[ 0

)
J

+ i
_
0
+
[

(x) [ 0

)
J
(i

m) +e0
+
[

(x)

(x) [ 0

)
J
+ (x)0
+
[ 0

)
J
_
(x)
+ i A

(x)
_

0
+
[ F

(x) [ 0

)
J
+e0
+
[

(x)

(x) [ 0

)
J
+J

(x)0
+
[ 0

)
J

+ i F

(x)
1
2
_
0
+
[ F

(x) [ 0

)
J

0
+
[ A

(x) [ 0

)
J

0
+
[ A

(x) [ 0

)
J

. (103)
In the above equation we have treated the variations in the elds to be c-number variations. Schwingers quantum
action principle states that Z[] is stationary with respect to variations in the dynamical parameters (elds /(x) in
the present case). Thus we can write
(i)
Z[]
J

(x)
= 0
+
[ A

(x) [ 0

)
J
(i)
Z[]
(x)
= 0
+
[ (x) [ 0

)
J
(i)
Z[]
(x)
= 0
+
[

(x) [ 0

)
J
0 = (i

m)0
+
[ (x) [ 0

)
J
+e0
+
[

(x)(x) [ 0

)
J
+ (x)0
+
[ 0

)
J
0 = 0
+
[

(x) [ 0

)
J
(i

m) +e0
+
[

(x)

(x) [ 0

)
J
+ (x)0
+
[ 0

)
J
0 =

0
+
[ F

(x) [ 0

)
J
+e0
+
[

(x)

(x) [ 0

)
J
+J

(x)0
+
[ 0

)
J
0 = 0
+
[ F

(x) [ 0

)
J

0
+
[ A

(x) [ 0

)
J

0
+
[ A

(x) [ 0

)
J
. (104)
Eliminating the elds in the above equations we get

_
i

m+e

(i)

J

(x)
_
(i)
Z[]
(x)
= (x) Z[] (105)
(i)
Z[]
(x)
_
i

m+e

(i)

(x)
_
= (x) Z[] (106)

_
g

_
(i)
Z[]
J

(x)
= J

(x) Z[] +e (i)

(x)

(i)
Z[]
(x)
. (107)
These are the functional dierential equations for Z[]. This was rst written down in this form by Julian Schwinger
[13,14]. Since Z[] is the generating functional for the Green functions it has the information regarding all possible
processes in quantum electrodynamics. Conversely a solution to the above functional dierential equations solves
quantum electrodynamics completely.
B. A formal solution to Schwingers functional dierential equations
Using Schwingers quantum variation principle in eqn. (100) for the variation of Z[] with respect to e we have
(i)
Z[]
e
= 0
+
[
W[/, ]
e
[ 0

)
J
=
_
d
4
x 0
+
[

(x)

(x)(x) [ 0

)
J
. (108)
16
Using eqn. (104) to eliminate the elds we get the rst order dierential equation
Z[]
e
= i
_
d
4
x (i)

(x)

(i)

J

(x)
(i)

(x)
Z[] (109)
Which can be integrated to yield the result
Z[] = e
ie
_
d
4
x (i)

(x)

(i)

J

(x)
(i)

(x)
Z
e=0
[] (110)
Where Z
e=0
[] is the generating function for the case when the photon eld is not coupled to the fermion eld. It is
fairly straight forward to show [19] that the solution to eqns. (105), (106) and (107) for the e = 0 case is
Z
e=0
[] = e
i
2
_
d
4
x
_
d
4
x

J(x)D

+
(xx

)J(x

)
e
i
_
d
4
x
_
d
4
x

(x)S+(xx

)(x

)
(111)
where the Green functions D

+
(x x

) and S
+
(x x

) satisfy the dierential equations

(g

)D

+
(x x

) = g

(4)
(x x

) (112)
(

m)S
+
(x x

) =
(4)
(x x

) (113)
with the boundary conditions on the Green functions prescribed by the requirement [Z[
2
1. Using eqn. (111) in
eqn. (110) we have
Z[] = e
e
_
d
4
x
_
d
4
x

(x)

(4)
(xx

)

J

(x)

(x

)
e
i
_
d
4
x
_
d
4
x

(x)S+(xx

)(x

)
e
i
2
_
d
4
x
_
d
4
x

J(x)D

+
(xx

)J(x

)
. (114)
For x
i
, y
i
being Grassmann variables we have the identity
e

ym
amn

xn
e
xib
ij
yj
= e
x
T

B(

1+

A

B)
1
y
e
Tr ln(

1+

A

B)
(115)
where we have used the symbolic notation, a (vector) for a
i
and

A (matrix) for A
ij
and Tr is the trace over the i, j
indices. Generalizing the above result for the case of functionals dependent on Grassmann variables we have
e
_
d
4
x
_
d
4
x

(x)
M(xx

)

(x

)
e
_
d
4
x
_
d
4
x

(x)N(xx

)(x

)
= e
_
d
4
x
_
d
4
x

(x)[N(+MN)
1
(x,x

)](x)
e
Tr ln(+MN)
(116)
where we have used the corresponding symbolic notation to suppress the integrals, Tr is the trace over both the
spinor and coordinate index, and is the Dirac delta function. Using the above identity, the expression in eqn. (114)
simplies to
Z[] = e
i
_
d
4
x
_
d
4
x

(x)G+[x,x

;(i)

J
](x)
e
L[(i)

J
]
e
i
2
_
d
4
x
_
d
4
x

J(x)D

+
(xx

)J(x

)
(117)
where
L[A] = Tr lnG
+
[x, x

; A

] + Tr lnS
+
(x x

) (118)
with Tr denoting the trace over both spinor and coordinate index and the Green function G
+
[x, x

; A

] satises the
dierential equation
[(i

m) +e

(x)] G
+
[x, x

; A

] =
(4)
(x x

) (119)
with the boundary condition prescribed by the initial condition
lim
e0
G
+
[x, x

; A

] = S
+
(x x

). (120)
To get prepared for the next step we write down the following identity involving the Gaussian function and an arbitrary
function F(x) in terms of the real variable x
F(

x
) e
1
2
xbx
= e
1
2
xbx
e
1
2

bx
b

bx
F(bx). (121)
Generalizing the above identity for functionals we have
17
F[(i)

J

] e
i
2
_
d
4
x
_
d
4
x

J(x)D

+
(xx

)J(x

)
= e
i
2
_
d
4
x
_
d
4
x

J(x)D

+
(xx

)J(x

_
e
i
2
_
d
4
x
_
d
4
x

A

(x)
D

+
(xx

)

A

(x)
F[A]
_
A =
_
DJ
(122)
where A =
_
DJ stands for A

(x) =
_
d
4
x

+
(x x

)J

(x

). Using the above identity in eqn. (117) we have

Z[] = e
i
2
_
d
4
x
_
d
4
x

J(x)D

+
(xx

)J(x

_
e
i
2
_
d
4
x
_
d
4
x

A

(x)
D

+
(xx

)

A

(x)
e
i
_
d
4
x
_
d
4
x

(x)G+[x,x

;A

](x)
e
L[A]
_
A =
_
DJ
. (123)
We observe that our problem of determining Z[] for quantum electrodynamics reduces to the evaluation of the Green
function G
+
[x, x

; A

] for an arbitrary A

(x). Any kind of approximate solution to G

+
[x, x

; A

] suggests a possible
non-perturbative approximate solution to Z[].
C. Electron-electron scattering
We can expand Z[] around Z[0] and thus write
Z[] = Z[0] +
_
d
4
x (x)
_

(x)
Z[]
_
J=0
+
_
d
4
x
_

(x)
Z[]
_
J=0
(x)
+
_
d
4
x J

(x)
_

J

(x)
Z[]
_
J=0
+. . . +T +. . . (124)
where T is the quantity of interest in the electron-electron scattering process and is given by the expression
T =
(i)
4
2! 2!
_
d
4
x
1
_
d
4
x
2
_
d
4
y
1
_
d
4
y
2

A
(x
2
)
B
(x
1
)
C
(y
2
)
D
(y
1
) G
ABCD
(x
1
, y
1
, x
2
, y
2
) (125)
where we have explicitly introduced the Dirac indices for clarity and
G
ABCD
(x
1
, y
1
, x
2
, y
2
) =
_
(i)
4


A
(x
2
)

B
(x
1
)

A
(y
2
)

A
(y
1
)
Z[]
_
J=0
=
_
e
M[

A
]
_
G
+
[x
1
, y
1
; A

] G
+
[x
2
, y
2
; A

] e
L[A]
__
A =
_
DJ,J=0
(126)
where we have introduced the short form for the operator
M
_

A
_
=
i
2
_
d
4
x
_
d
4
x

(x)
D

+
(x x

)

A

(x)
. (127)
T will be interpreted as the probability amplitude for the electron-electron scattering process. Every process in
quantum electrodynamics can be generated out of the expansion in eqn. (124). It is due to this reason that Z[] is
also called the generating function.
For the electron-electron process we shall be interested in the case when the sources of the process (x
1
), (x
2
), (y
1
)
and (y
2
) are interaction free states with well dened initial or nal momentums. Thus we choose
(x
1
) = e
ip1x1
u(p
1
)[

p
1
+m]
(y
1
) = e
ip

1
y1
[

1
+m]u(p

1
)
(x
2
) = e
ip2x2
u(p
2
)[

p
2
+m]
(y
2
) = e
ip

2
y2
[

2
+m]u(p

2
). (128)
Accordingly T will be denoted as T(p
1
, p

1
, p
2
, p

2
). In the next section we shall concentrate on approximate solutions
to T(p
1
, p

1
, p
2
, p

2
).
18
VI. APPROXIMATION TO ELECTRON-ELECTRON SCATTERING
A. Quenched or ladder approximations
To start with we shall assume that e
L[A]
in eqn. (126) does not contribute. Further we assume that the operator M
in eqn. (127) act in such a way that they connect the two fermion propagators via photon lines. In the diagrammatic
language this amounts to omitting all those graphs that involve fermion loops and keeping only those graphs that
have photons exchanged between the fermion lines. This is called the quenched or ladder approximation. In this
approximation eqn. (126) reads
G
ABCD
(x
1
, y
1
, x
2
, y
2
) =
_
e
M12[

A
]
[G
+
[x
1
, y
1
; A

] G
+
[x
2
, y
2
; A

]]
_
A=
_
DJ, J=0
(129)
where we introduced M
12
in place of M to signify that it connects the two fermion lines. Using the identities
e
M12
= 1 +
_
1
0
d e
M12
M
12
(130)
and

(z)
G
+
[x
1
, y
1
; A

] = G
+
[x
1
, z; A

] e

G
+
[z, y
1
; A

] (131)
we thus get
T(p
1
, p

1
, p
2
, p

2
) = i
_
1
0
d e
M12
_
d
4
z
1
_
d
4
z
2
D

+
(z
1
z
2
)

_
d
4
x
1
e
ip1x1
u(p
1
)[

p
1

+m]G
+
[x
1
, z
1
; A

1
] e

_
d
4
y
1
e
ip

1
y1
G
+
[z
1
, y
1
; A

1
][

1
+m]u(p

1
)

_
d
4
x
2
e
ip2x2
u(p
2
)[

p
2
+m]G
+
[x
2
, z
2
; A

2
] e

_
d
4
y
2
e
ip

2
y2
G
+
[z
2
, y
2
; A

2
][

2
+m]u(p

2
). (132)
B. Eikonal approximation to G+[x, y; A

]
Next step is to evaluate G
+
[x, y; A

] in the eikonal approximation. We dened G

+
[x, y; A

] in eqn. (119) to satisfy

the dierential equation given by
[(i

m) +e

(x)] G
+
[x, y; A

] =
(4)
(x y). (133)
In the eikonal approximation we have [1922]
G
+
[x, y; A

] = i
_

0
ds e
ims
(x y s
p
m
) e
ie
_
s
0
d
1
m
p

A(x
p
m
)
. (134)
Using the above eikonal approximation to G
+
[x, y; A

] in eqn. (132) it becomes possible to evaluate T(p

1
, p

1
, p
2
, p

2
).
Here we shall simply state the result without the detailed steps in the derivation. We dene
T(p
1
, p

1
, p
2
, p

2
) = (2)
4

(4)
(p
1
+p

1
p
2
p

2
)M(p
1
, p

1
, p
2
, p

2
) (135)
and in the eikonal approximation we have the result [19]
[ M(p
1
, p

1
, p
2
, p

2
) [=
e
2
2m
2
(p
1
+p
2
)
2
(p
1
p

1
)
2
(1 i
e
2
4
)
(1 +i
e
2
4
)
. (136)
We notice the close similarity of the above expression to the non-relativistic scattering amplitude due to the Coulomb
potential. This is not surprising because eikonal approximation is a stringent semi-classical limit.
19
ACKNOWLEDGMENTS
Most of the work presented here evolved during my regular meetings with Kimball Milton and Ines Cavero. The
meetings played a very signicant role in my learning. I would like to thank Kimball Milton for agreeing to be
my PhD thesis advisor. I have learnt from him the approach to attend to every problem by starting from scratch.
Most of all, I would like to thank him for being the devils advocate
2
during all the stimulating discussions we have
had. Almost always the discussions led to such remote regions of thought which would have been left completely
unexplored otherwise. I am also grateful to him for scrutinizing this report for errors and defects. I have tried my
best to incorporate his suggestions.
I am grateful to Bahman Roostaei for allowing me to use the condensed matter groups computer (IBM
RS/6000/7028-6E4 which uses powerpc multiprocessors and has a 4GB memory). The plots reproduced here would
have been impossible to draw without such a giant memory. Thanks are also due to Pravin Chaubey for helping me
during the search on the etymology of the word eikon.
I would like to thank David Branch, Leonid Dickey, Ronald Kantowski, and Chung Kao for agreeing to be in my
PhD advisory committee. I would also like to thank Shankar for all the nice discussions and April Hendley for being
a patient listener.
[1] Translated by Margarita Kamilieri who is originally from Greece. She is majoring in social work at the University of
Oklahoma.
[2] Merriam-Websters Collegiate dictionary, 10th edition.
[3] Roger G. Newton, Scattering Theory of Waves and Particles, 2nd ed., Springer-Verlag New York, Inc., (1982).
[4] Page 610, ref. [3].
[5] R. J. Glauber, Lectures in Theoretical Physics, ed. by W. E. Brittin and L. G. Dunham, Interscience Publishers, Inc., New
York, Volume I, page 315, (1959).
[6] Charles J. Joachain, Quantum Collision Theory I, II, North Holland Publishing Company, (1975).
[7] Leonard I. Schi, Quantum Mechanics, 3rd ed., McGraw-Hill Book Company, New York, (1968).
[8] F. Smithies, Integral equations, Cambridge Tracts in Mathematics and Mathematical Physics, ed. by P. Hall and F.
Smithies, Cambridge University Press, No. 49, (1958).
[9] B. A. Lippmann and Julian Schwinger, Phys. Rev. 79 469 (1950).
[10] Jagdish Mehra and Kimball A. Milton, Climbing the Mountain: The Scientic Biography of Julian Schwinger, Oxford
University Press, Inc., New York, (2000), page 171.
[11] R. J. Glauber and V. Schomaker, Phys. Rev. 89 667 (1953).
[12] E. Feenberg, Phys. Rev. 40 40 (1932).
[13] Julian Schwinger, On the Greens Functions of Quantized Fields. I, II, Proc. Natl. Acad. Sci. U.S.A. 37 452, 455 (1951).
[14] Julian Schwinger, The Theory of Quantized Fields. I-V, Phys. Rev. 82 914 (1951), 91 713, 721 (1953), 92 1283 (1953), 93
615 (1954).
[15] Julian Schwinger, Particles, Sources and Fields, Addison-Wesley, Reading, MA, (1970).
[16] Kimball A. Milton, Lecture Notes, University of Oklahoma, (2002).
[17] H. M. Fried, Greens Functions and Ordered exponentials, Cambridge University Press, (2002).
[18] H. M. Fried, Functional Methods and Models in Quantum Field Theory, The MIT Press, Cambridge, MA, (1972).
[19] Leonard Gamberg and Kimball A. Milton, Phys. Rev. D 61 075013 (2000).
[20] F. Guerin and H. M. Fried, Phys. Rev. D 33 3039 (1986).
[21] Henry D. I. Abarbanel, Lectures in Theoretical Physics XIV A, University of Colorado, page 3 (1971).
[22] Robert L. Sugar, Lectures in Theoretical Physics XIV A, University of Colorado, page 47 (1971).
2
Websters dictionary meaning: a person who champions the less accepted cause for the sake of argument.
20