IL NUOVO CIMENTO VOL. 108 B, N. 8 Agosto 1993
Propagator Relative to the Step Potential.
L. CHETOUANI(1), L. GUECHI (i), T. F. HAMMANN(2) (*) and Y. MERABET(1) (1) Dgpartement de Physique Thgorique, Institut de Physique, Universitg de Constantine
Constantine, Algeria (2) Laboratoire de Mathdmatiques, Physique Mathgmatique et Informatique
Facultg des Sciences et Techniques, Universitg de Haute Alsace 4, rue des Fr~res Lumi~re, F-68093 Mulhouse, France
(ricevuto il 17 Settembre 1992; approvato il 30 Marzo 1993)
Summary. -- The propagator associated to the Heaviside potential VoO(X) is obtained under its compact form by summation on all the states of the particle. It is proved that the base of the states is complete indeed and that for Vo--~ ~, the propagator tends towards the well-known propagator of a particle subjected to move along a semi-axis.
PACS 03.65.Bz - Foundations, theory of measurement, miscellaneous theories. PACS 03.65.Ca - Formalism.
1. - Introduction.
The purpose of this paper is to build very explicitly the propagator relative to the step potential or Heaviside potential
(1) V(x) = VoO(X) (Vo > 0),
by summing up all the ~_ (x) and ~_~ (x) states of the particle; (we use here the standard convention according to which the arrow in the wave function indicates the way in which the particle moves).
The resolution of Schr5dinger's equation and the calculation of the transmission and reflection coefficients are trivial, whereas the calculation of the propagator leads to highly interesting mathematical and physical problems. Summing up, the wave functions often proves to be a method more manageable than the path integral approach: as an example, we can quote the Kepler-Coulomb problem for which the Green's function has been obtained through the traditional method  long before it could be solved in the framework of the path integral approach .
Finding a solution to the Heaviside potential within the path integral for- realism has been tried, but this solution has not proved satisfactory since, for
(*) To whom request for reprints should be addressed.
59 - I l Nuovo Cimento B 879
880 L. CHETOUANI, L. GUECHI, T. F. HAMMANN and Y. MERABET
Vo-~ ~ the well-known propagator associated with a particle subjected to move along a semi-axis, could not be found.
In this paper we aim at obtaining an accurate expression for the propagator.
2. - P ropagator for the f in i te-step potent ia l .
The solutions to SchrSdinger's equation
he d 2 ] + VoO(X) ~_(x ) = E~_+(x)
(2) 2m dx 2 +-
are well known. When the particle moves from right to left, for E > V0 energies, the wave function can be written as
m ]1/2 (3a) ~+ (x) = 2r~h2k 2 [exp[ - ik2x] - c~ exp[ik2x]], x > 0,
m 11/2 (3b) '~(x ) = 2rch2k 1 5rexp[ - ik lX ] , x < 0,
ki=[2 llJ2 k2 0 ]1J2 h 2 J ' h2 '
kl - k2 2(kl k2) 1/2 - and ~ - ,
kl + k2 kl + k2
respectively, represent the reflection and transmission coefficients. When the particle moves from left to right with a positive E energy, the wave function can be written as
m 11/2 5r exp [ - ik2 x ] , x > O, (4a) ~_~ (x) = 27~h2 k 2
m ]1/2 (4b) ~+(x) = 2=h2k1 [exp[iklX] + g2 exp[ - ik lX ] ] , x < 0,
the [m/2rch2kl] 1/2 and [m/27~h2k2] 1/2 constants being obtained by normalization of the wave functions,
-E )~,__ (5) : : -+-+,
PROPAGATOR RELATIVE TO THE STEP POTENTIAL 881
through the formula
I i (6) dxexp[ i (k + i0)x] - k + i0 0
The propagator is linked to these wave functions by the following summation on the states
I i ET i T (7) K(b, a; T) = dE~(b) r - - -~ + dE~___)(b)r - - - .
We shall keep this notation for E < V0, where the k2 constant then becomes imaginary.
In order to put eq. (7) into a compact form, we shall consider all the possible cases according to the initial and final positions of the particle:
First case: b < 0 and a < O.
In the case, the r and ~_~ wave functions of eq. (7) are given respectively by eqs. (3b) and (4b), and the propagator
882 L. CHETOUANI, L. GUECHI, T. F. HAMMANN and Y. MERABET
Thanks to the formula
exp[-iak~] = [-~ ) _J dpexp[i(ape ~ [3kl)], with fl= 2~p,
allowing us to linearize k~, eq. (10) can be written as
- - - - dp exp [i~p2] 9 (11) K~
PROPAGATOR RELAT IVE TO THE STEP POTENTIAL 883
2 ~-VY sin(z0 - ~) (16) J2 = + Z o - - O)
+ - - Z 0 - - 09
I - zo_) -~ (4 + y2)cosh (2y)
dy, for co-Zo > O,
with Zo = V~ (b + a) and oJ = ~/-y/~. Moreover, we have
(17) J iy (oJ Zo) dy = dy =
Jiy (oJ - z o) ] = _+2irc~ Res (4 + y2)cosh((r:/2)y) '
where c stands for the contour of the semi-circle with radius R--~ oo located in the upper half-plane for (Zo - o~) > 0, and in the lower half-plane for (Zo - co) < 0. This distinction comes from the asymptotic behaviour --for large ~--of the modified Bessel functions
(18) Iv(vZ)'~ (l + z2)-l/4 [ Z] (27~)1/e exp v~v/ l+z 2+~ln 1 + ~
the first- and second-kind Bessel functions being related by the standard rela- tions,
(19) J~(x) = exp[i2v]Iv(- ix).
Eventually, we find that the main contribution to the value of the integral (17), comes
from the factor exp [~ In (z/(1 + ~ + z2))]. Thus, it follows that
2 ~ sin (z o - oJ) + (20) J2 = + sign (Zo - co) 2r: V~ J2 (Zo - oJ)
Z 0 - - o) Z 0 - - o)
+ P 16 V~ z_, ~ ( - )l J21 +1 (Zo - o~)
Zo - o~ t=o (2l - 1)(2l + 3)
Let us now replace p by (p - (m/hT) (b + a)) and we shall thus eventually obtain the
884 L. CHETOUANI, L. GUECHI, T. F. HAMMANN and Y. MERABET
compact expression of the propagator, for a and b < 0:
(21) K> (b, a; T) = K0(b, a; T) + K~>(b, a; T),
] i __1 exp - V0 T dk2 exp [ - i[ak~ - k2 (b - a)]] = (23) Ko(b, a; T) = 2~
is the propagator of a free particle bearing a phase factor, and
(24) K~>(b, a; T) =
1 exp - T - 2~ --/- 0
dk l [~C~exp[ ik2(b+a) ]+c .c . ]exp[ - iak~] +
Id kl tT~ exp [ik2 (b + a)] + c.c. ] exp [i~k~] + kl k2 ,/7
is the propagator resulting from the transmission and the reflection by the potential, obtained by the use of
(25) 1~12- [ ~ ] ~ +c.c.
PROPAGATOR RELATIVE TO THE STEP POTENTIAL 885
We obtain this propagator by noticing that kl dkl--k2 dk2, and by performing the changes V0 o 0, i.e.
(26) K>>(b, a; T) = Ko(b, a; T) -
I L- o p -~m p+ -~(b+a) .
This propagator proves to be a superposition of Gaussians as well, and the second part also refers to the image of point b with respect to the origin.
Third case: b < 0 and a > O.
We can show that the propagator is now
(27) K~(b ,a ;T )= l [ f f dkl kl) 1/2 + c.c.] exp [-iak~] [(~-2 J-exp[i(k2a-klb)] +
[ t- ~1/2 | + [dkl|'~l | J-[exp[i(k~a - kib)] +c.c . ]exp[- i~k~] l -
Let us linearize the exponent in kl 2, first:
(28) KX (b, a; T) =
_ " 57. exp [i[k2 a + kl (8 - b)]] + c.c. +
oo kl(kl/1/2 5r[exp[i[k2a + kl(/~ -- b)]] +c.c.] ].
t __ b -a b+a a fo rp- -
tghp = fl - b ' 2a 2:~
~-a for b-a b+a - - , - - < p < , a 2~ 2a
886 L. CHETOUANI, L. GUECHI, T. F. HAMMANN and Y. MERABET
and let us decompose the integration interval on the p variable as follows:
I -~ ' ~ a[U]b-a2a ' b+a[u]b+a2a-- 2a ' +~[
and let us calculate the integral (28) in the last two intervals first. In the ](b + a)/2~, ~ [ interval, the calculation is similar to that of the first case;
the result obtained is nil. In the ](b - a)/2a, (b + a)/2a[ interval, we have to evaluate the expression
(b + a) /2~ ["
I dpexp[i~p ] [ 2 (b - a)/2:~
;2 ] 2i a ~ dOexp[-iO]exp cosh,z Sb ~a ~
[ ; [ ]]] 2 ~ ~ d0exp[ -0 ]exp iV~as inh(0+~) +c.c. . + ~ 3b ~a cosh~
Let us use the Fourier-series expansion formula
exp[ixsinr = Jo(x) + 2 ~ J2n(x)cos(2nr + 2i ~ J2n+l(x)sin[(2n + 1)r n=l n=0
to calculate the first integral, as well as the formulae 
() I = 7: cos (z sinh 0), cos(Oy) cosh ~ y ,~iy (z) dy = 2 0
() sin(0y) sinh ~ y ,~/dy (z) dy = --2 sin (z sinh 0), 0
together with the following relations between the ,~ (z) imaginary-argument Bessel functions, the H~l)(iz) first-kind H~inkel functions, and the Jr(z) first-kind Bessel functions :
i7~ [ 7V 1 H (1) iCy(z) = - - exp - (iz), 2 2 y] Y
(1 ) " Hiy (~z)= exp [~y] Jiy (iz) - J-iy (iz)
J~ (exp [m=i] z) = exp [mvrd] J,, (z),
in order to calculate the second integral and eventually show that the expression (29) is nil.
Let us notice that there is another way of showing that the integral (28), within the interval ](b - a)/2~, + ~ [ of variation of the variable p, is nil. Indeed, the inte-
PROPAGATOR RELATIVE TO THE STEP POTENTIAL 887
f dkx(kl lj2
J -exp[i[kea + k l ( fl - b)]] +c.c. = J3 + J~ ,
,~exp[i[koa + kl(~ - b)]] +c.c. = J4 + J4*,
can be brought down to an integral on the contour f~ thanks to the following adequate variable changes:
i) kl = X/7 cos 0 in J3 + J3* followed by the replacing of 0 by 7: - 0 in J.i*,
ii) k 1 = ~g/~cosh r with r = i0, in J4 + J4* followed by the replacing of 0 by 0 - =, in Jd*, which leads to
(31) 2 ~ ] dOsinOcosOexp[-iO]exp[V~[i(~ - b)cos0 - asin0]]. f7
The path of integration fB is shown in the following drawing:
0 ) ) X
- i~ r : - i~
f(z) = V~ sin (2z) exp [ - iz] exp [ V~[i(/? - b) cos z - a sin z]],
is analytic inside and on the ABOr:A contour. Therefore, we can write
f dzf(z) =- f dzf(z), V1 xe
z=x- iy , O
888 L. CHETOUANI, L. GUECHI, T. F. HAMMANN and Y. MERABET
(32) Izf(z)l y~ee y.
"exp[y]exp[ -V~s inx[ ( f l -b2 +a)exp[y]+(a+ b- f l )exp[ -y] ] ]
i f f l>b-a , or a l sop>(b-a) /2a . Consequently, the propagator (28) gets transformed into
-~0, y.--~ zc
(33) KX (b, a; T) =
_ 1 a dpexp[iap e] dkl kl J_exp[i[ke a _ kl(/? + b)]] +c.c. 2r.
(a - b) /2~ 0
+ kl J [exp [i[k2a - kl (~ + b)]] + c.c.] .
There is even a more compact way to write this equation, by prolonging 5 r analytically into the complex plane (Re ~ is symmetric, whereas Im v]- is anti- symmetric). Thus
(34) KX (b, a; T) =
2r: ~ dpexp[iap 2] dkl (a - b ) /2a - o~
exp [i[k2 a - kx (fl + b)]].
As we have
k~ ~Sr = 1+ co~,
the expression (31) can be decomposed in the following way:
(35) KX (b, a; T) =
1 - dpexp[izcp 2] dkl(1
27: (a - b ) /2~ - ee
By introducing the identity
+ cd~ )exp[i[k2a - kl( fl + b)]].
+oo +zo -bee
I I f =1,
PROPAGATOR RELATIVE TO THE STEP POTENTIAL 889
into the second integral of eq. (35), we obtain, since ks = Vk~- Y,
f dkl -o r
c~ exp[i[k2a - k1(/3 + b)]] =
----i f dz f duexp[i[Vu -ya-z ]] f dkl 27:
~2 exp [ikl (y. - /3 - b)].
By using the results of the first case,
1 ] dkl [2 exp[ikl(Z-/3 b)] -2 2r~
[ 2mVo ( _ fl _ b) I ]
X- /3 -b 0( X -/3 - b),
we eventually come to the following result, for V0 > 0:
(38) KX(b ,a ;T )= ~ (a- b)/2a
dp exp [iap 2] ] dkl exp [i[k2 a - kl (/3 + b)]]. -o r
9 1 -2 dy J~ Y 9 o Y h
Fourth case: b > 0, a < 0.
With calculations similar to those of the third case, we come to the following propagator:
Kg (b, a; T) = ~ dpexp[ iep 2] dkl (b - a ) /2~ - ~
exp [i[k2 b - kl (/3 + a)]].
[ 9 1 -2 dy J2 Y 9 o Y h
3. - Par t i cu la r cases .
3"1. Limit ing case where Vo--~ ~. - We have to check whether for Vo-+ ~, we obtain the well-known propagator relative to a particle subjected to move along a semi-axis.
J2 (2x) J2 (2x) (40) ~ cr for x=O, and + O, for x~O,
X ~--*~ )~ ~--*~
890 L. CHETOUANI~ L. GUECHI~ T. F. HAMMANN and Y. MERABET
it becomes obvious that J2 (~x)/x behaves like the Dirac distribution ~(x). Moreover, according to formula 
f J'~ dx if - l
PROPAGATOR RELATIVE TO THE STEP POTENTIAL 891
4. - L ink wi th the Feynman propagator .
The S action evaluated along the path linking points (a, 0) and (b, T), is given by
(47) S= f [ m22-2 - V(x)]dt,
and, according to the Feynman approach , the propagator is
(48) K (b 'a ;T )= lim (~( m 11/2N-j~ 1 [~ [~'r ]] N~ J j=l \ 2--~h~ ] : dxjexp j=l~ (Sxj) e - ~V(xj) ,
which is a product of the infinitesimal propagators K(xj, xj_ 1; s) in which the infini- tesimal action
S(x~, x~_ 1) = ~ (Sxj) 2 - W(x~),
is calculated along a segment of a line linking the points (xj_ 1, tj _ 2) and (xj, tj), with tj - tj_ 1 = ~ = TIN.
We then obtain the following infinitesimal propagators:
892 L. CHETOUANI, L. GUECHI, T. F. HAMMANN and Y. MERABET
framework of the path integral approach. For V0 ~ cr we effectively obtain the propagator of a particle subjected to move along a semi-axis and for T--* 0, we have demonstrated that the base of states is complete indeed.
 L. H. HOSTLER: J. Math. Phys., 5, 185 (1979).  I. H. DURU and H. KLEINERT: Phys. Lett. B, 84, 185 (1979); Forts. Phys., 30, 401 (1982);
L. CHETOUANI and T. F. HAM~NN: Nuovo Cimento B, 98, i (1987); Phys. Rev. A, 34, 4737 (1986); J. Math. Phys., 28, 598 (1987); L. CHETOUANI, L. GUECHI and T. F. HAMMANN: Nuovo Cimento B, 101, 547 (1988).
 A. O. BARUT and I. H. DURU: Phys. Rev. A, 38, 5906 (1988).  I. S. GRADSHTEYN and I. M. RYZHIK: Table of Integrals, Series and Products (Academic
Press, New York, N.Y., 1965), p. 973, eq. (8.511.4).  I. S. GRADSHTEYN and I. M. RYZHIK: Table of Integrals, Series and Products (Academic
Press, New York, N.Y., 1965), p. 774, eq. (6.796.1).  M. ABRAMOWITZ and I. A. STEGUN: Pocketbook of Mathematical Functions (Harry
Deutsch, Frankfurt am Main, 1984), p. 122.  I. S. GRADSHTEYN and I. M. RYZHIK: Table of Integrals, Series and Products (Academic
Press, New York, N.Y., 1965), p. 973, eq. (8.511.3).  I. S. GRADSHTEYN and I. M. RYZHIK: Table of Integrals, Series and Products (Academic
Press, New York, N.Y., 1965), p. 774, eqs. (6.796.2), (6.796.3).  I. S. GRADSHTEYN and I. M. RYZHIK: Table of Integrals, Series and Products (Academic
Press, New York, N.Y., 1965), p. 954, eq. (8.407.1); p. 970, eq. (8.483.1); p. 968, eq. (8.476.1).
 I. S. GRADSHTEYN and I. M. RYZHIK: Table of Integrals, Series and Products (Academic Press, New York, N.Y., 1965), p. 684, eq. (6.561.17).
 R. P. FEYNMAN and A. R. HIBBS: Quantum Mechanics and Path Integrals (Mc Graw-Hill, New York, N.Y., 1965).