The Klein-Gordon Propagator

Dan Styer, Oberlin College Physics Department, Oberlin, Ohio 44074 dstyer@physics.oberlin.edu 5 October 1999

This document lls in some of the details behind the discussion of amplitudes to go from place to place found in Richard Feynmans QED: The Strange Theory of Light and Matter (Princeton University Press, Princeton, New Jersey, 1985). Unlike Feynmans book, this document is technical. To understand it, you need to understand terms like contour integral and residue. The discussion I wish to elucidate is found on pages 8791 of QED, and presents amplitudes for the rst two of the three basic actions, namely a photon goes from place to place and an electron goes from place to place. Feynmans so-called polarization-free photon and spin-zero electron are technically called KleinGordon particles of zero and nite mass, respectively. The amplitude to go from place to place that Feynman mentions is called the Klein-Gordon propagator. An integral expression for this propagator is given in, for example, Claude Itzykson and Jean-Bernard Zuber, Quantum Field Theory (McGraw-Hill, New York, 1980), page 35, equation (1-178), or in Kurt Gottfried and Victor Weisskopf, Concepts of Particle Physics (Oxford University Press, New York, 1986), volume II, page 230, equation (48) [note misprint: d4 x should read d4 Q]. Using the phase conventions of QED, the propagator to change space-time position by x = (ct, r) = (x0 , x) is 1 d4 p ipx GF (x) = e , (1) 4 2 m2 + i (2) p where x y x0 y0 x y. The aim of this document is to show how the qualitative amplitude descriptions of QED follow from this expression.

The energy integral

Write the propagator as GF (x) = d3 p ipx e (2)4 dp0 eip0 x0 p2 0 p2 1 . m2 + i (2)

Dene E = + p2 + m2 , and then evaluate the energy integral I dp0 eip0 x0 p2 0 1 , E2 + i (3)

using contour integration in the complex p0 plane. To locate the poles, write p2 E 2 + i = p2 E 0 0 i 2E
2

= p0 + E

i 2E

p0 E

i 2E

(4)

Thus there are two poles: one just above the real axis and one just below. The rst pole has location: while the second has location: + E i 2E residue: + exp {+i (E i /2E) x0 } . 2 (E i /2E) 0) (7) (6) E i 2E residue: exp {i (E i /2E) x0 } , 2 (E i /2E) (5)

If x0 > 0, we close the contour on the top half plane enclosing the rst pole to nd (in the limit I = +2i eiEx0 2E ,

while if x0 < 0, we close the contour on the bottom half plane enclosing the second pole to nd I = 2i + These two expressions can be written as one, I = 2i whence we conclude GF (x) = i 2 d3 p ipx eiE|x0 | e , (2)3 E where E = + p2 + m2 . (10) eiE|x0 | , 2E (9) e+iEx0 2E . (8)

Propagator for massless particles

If m = 0, then E = |p| and the above expression becomes GF (x) = i 2 d3 p ipx ei|p||x0 | e . (2)3 |p| (11)

For the case x0 = 0, this integral is evaluated in the appendix and is GF (x) = Thus 2 i 1 i 1 = . (2)2 x2 x2 (2)2 (r)2 (ct)2 0 (12)

if if

(r)2 > (ct)2 (r)2 < (ct)2

(i.e. v > c) (i.e. v < c)

then GF i then GF +i

These amplitudes correspond to the two little arrows pointing to the right and to the left in gure 56 on page 90 of QED. The remaining case is (r)2 = (ct)2 , that is v = c. In this case x2 = x2 and 0 GF (x) = i 2 d3 p ipx ei|p||x| e . (2)3 |p| (13)

Convert this integral into spherical coordinates (using = cos()) to nd GF (x) = The integral over is i 2 2(2)3
+1 0 +1

p2 dp
1

d eip|x|

eip|x| . p

(14)

d eip|x| =
1

2 sin(p|x|) , p|x|

(15)

so GF (x) i 1 (2)2 |x| i 1 = 2 |x|2 (2) =

0

dp sin(p|x|)eip|x| du sin(u)(cos(u) i sin(u)).

(16)

Now, the integral

0

du sin(u) cos(u) whereas

0

is bounded,

(17)

du sin2 (u)

approaches innity.

(18)

Thus for the remaining case v = c, we have GF (x) = (real positive innity) . |x|2 (19)

This amplitude corresponds to the big arrow pointing straight up in gure 56 on page 90 of QED.

Appendix: Fourier transform of the Yukawa Potential

Theorem: If f (r) = and if f (k) f (r) = = d3 r eikr f (r) d3 k +ikr e f (k), (2)3 3 (21) (22) ek0 r r

with

k0 > 0,

(20)

then f (k) = Proof: f (k) = = = = = = = 2

0

4 2. k 2 + k0

(23)

d3 r eikr

ek0 r r
+1

(24) d eikr ek0 r r (where = cos ) (25) (26) (27) (28) (29)
r=0

r2 dr
1

2
0

r2 dr

2 sin(kr) kr

ek0 r r

4 dr sin(kr)ek0 r k 0 4 m dr e(ikk0 )r k 0 4 e(ikk0 )r m k ik k0 4 2 k 2 + k0

(30)

Propagator for massive particles

Returning to the very beginning of this discussion, GF (x y) = But, recognizing the geometric series, 1 p2 m2 + i = = Therefore GF (x y) = d4 p eip(xy) + m2 (2)4 p2 + i d4 p eip(xy) + (m2 )2 (2)4 (p2 + i )2 d4 p eip(xy) + . (2)4 (p2 + i )3 (32) 1 (p2 + i )[1 m2 /(p2 + i )] 1 m2 (m2 )2 (m2 )3 1+ 2 + 2 + 2 + . (p2 + i ) p +i (p + i )2 (p + i )3 d4 p ip(xy) 1 e . (2)4 p2 m2 + i (31)

Note that the rst term in this series is nothing but the zero-mass propagator, which we will call GF (x y) =
(0)

d4 p eip(xy) . (2)4 p2 + i

(33)

Im going to write the second integral in a funny way, using the four-dimensional Dirac delta function (4) (p): d4 p eip(xy) d4 p eipx eip y = d4 p (4) (p p). (34) 4 (p2 + i )2 4 (p2 + i )(p 2 + i ) (2) (2) 4

Use the integral expression (4) (p p) = for that delta function to write d4 p eip(xy) = (2)4 (p2 + i )2 d4 x

d4 x i(p p)x e (2)4 d4 p eip (x y) (2)4 (p 2 + i ) d4 p eip(xx ) . (2)4 (p2 + i )

(35)

(36)

Recognizing the two zero-mass propagators on the right, we write d4 p eip(xy) = (2)4 (p2 + i )2 d4 x GF (x x )GF (x y).
(0) (0)

(37)

In a similar way, the third integral in the series (the one multiplying (m2 )2 ) can be written as a double integral of a product of three zero-mass propagators, and so forth. We conclude that GF (x y) = GF (x y) +m2 +(m2 )2 +. This is precisely the boxes within boxes form described in footnote 3 on page 91 of QED. Note that the analysis of this section links the nite-mass propagator to the zero-mass propagator without ever using the previously-obtained explicit form of the zero-mass propagator. d4 x GF (x x )GF (x y) d4 x d4 x GF (x x )GF (x x )GF (x y)
(0) (0) (0) (0) (0) (0)

(38)