You are on page 1of 5
BR as UT 6.1 The Bethe-Salpeter Equation For many purposes itis necessary to have a better description for the interaction of two particles than the perturbation theory we have discussed so far. This is particularly so if we deal with bound states: the particles stay together infinitely long and they can therefore interact arbitrarily often as depicted in Fig. 6.1. It is clear that this situation cannot be described by the summation of a few Feyn- man diagrams. If one of the particles is much heavier than the other the problem can bbe considerably simplified: one solves the Dirac equation for the light particle with ‘an additional external potential, which is produced by the heavier particle, Because of the large mass ratio the influence of the lighter particle on the heavier one can be neglected ‘The interaction of two particles with equal or comparable masses is much more difficult to describe since the “recoil effects” cannot be neglected, The most famous ‘example of this is positronium, i.e. the bound system of an electron and a positron. Here one cannot distinguish between the source of the field and the test particle, One ‘must rather treat both particles on an equal footing. There isa further complication due to the fact that the interaction propagates with finite velocity, leading to “retardation, effects”. In a correct relativistic theory there is no preferred common time coordinate. ‘The wave function of the two-particle system depends on two time coordinates and. therefore its interpretation is difficult. In the following we want to write down an exact equation for the two-particle sys- tem, at least in principle, even if it turns out that it is too difficult to find an exact, solution for this equation. Then we shall study an approximation that again leads to ‘akind of Dirac equation with an interaction potential (the Breit interaction, see Fxam- ple 6.4). In onder to study the behaviour of two equal particles a and b, we generalize the de- finition of the propagator, Looking only at one particle, the wave function of that parti- cle is described by a spinor w(x) with four components. As shown in Exercise 6.1 the corresponding initial value problem is solved by the Feynman propagator Sp (x2, x1): va) fans. Ga xy!mu wen « 1) ‘where we have to integrate over a closed three-dimensional hypersurface that includes the space-time point x2; n,.(%1) is the exterior normal vector of this hypersurface. This relation can be extended to the two-particle case. Obviously such a system, must be described by a wave function Yan(x1, x2) that depends on eight coordi- nates: six space coordinates (21, #2) and two time coordinates (t1,f2). Furthermore W. Greiner, J. Reinhardt, Quantum Electrodynamics, 329 (© Springer-Verlag Berlin Heidelberg 2009 Fig. 6.1. The particles in ‘abound state can interact infi- nitely often 330 6._Two-Particle Systems Yan 1, x2) has two spinor indices, ic. 4 x 4= 16 components! The indices a and b refer to the spinor spaces ofthe two particles, The appearance of two time coordinates, in contrast to the nonrelativistic many-particle theory, makes the interpretation of this wave function more difficult. Alas, this cannot be avoided if one wants to treat space and time coordinates in the same way. In analogy to (6.1) we can define a propagator for the two-particle wave funetion estan) = f oor) do C35, 5x1, 9) DHCD) Yanl + (6.2) where one now has to integrate over two three-dimensional hypersurfaces which in- clude the points x3 and xs. As in Exercise 6.1, (6.2) can be specialized to the case of two space-like hypersurfaces, i. to an integration over d°x, at a fixed time coordi- nate fy If the two particles do not interact, the wave function simply is given by a product Wart, x2) = VaCen¥o(e2) (63) so that (6.2) factorizes into two integrals of the form (6.1) if one sets 84% (a3, x45 x1, x2) = ASF (03, x1) SP (4, 22) eo Going beyond this trivial case we have to consider what happens if the two particles interact electromagnetically. In perturbation theory (expanding in powers of the cou pling constant e) the first correction to (6.4) can be found easily,' it is given by the one photon exchange diagram Qf 34 34 (65) i Y 1 The bubble denoted by S represents the full two-particle propagator whereas the un- connected fermion lines represent the free two-particle propagator $° given in (6.4). Using the Feynman rules we can translate (6.5) into the formula aby "(3.45 X1, X2) = 186 (x3, x1) ISL 0%4, x2) + [ ossatreispton. as iStosn0) x [Hieaygivfies.xa cieny? }isetas.xvishr6.xd + (66) Note that in this and the following equations there are two independent spinor in dices. For example the 4 x 4 matrix yf gets multiplied by the matrices Si{(x3, xs) and S€(xs,x1), whereas y? is multiplied by SB(x4, x6) and SP (x6, x2). The two-particle propagator therefore has the character of a \6 x. 16 matrix. In (6.5) and (6.6) we have For simplicity we assume that particles a and b are distinguishable, since otherwise exchange graphs ‘would appear. The case ofa particle interacting with ts antipatile willbe considered in Exercise 6.2 6.1_The Bethe-Salpeter Equation 331 written down only the first term of an infinite series. The result of the infin ccan be represented in terms of a “black box” by a function K (x3, 43.%1,.2) which is called the interaction kernel. The exact form of (6.5) then reads 5 4 a4 YY - tt. 6 my tt or, written out explicitly, 8 (x3, x45 41, 42) = 1982.41) iSH(44, 42) +f otssareatay dap isBlag. 45) 8B 0) x K? (x5, x65 27, ¥8) ISH (x7, x1) 1Sf (x8, 22) (6.8) ‘We have not gained much by doing this since K is an extremely complicated function. Only in first-order perturbation theory does the kernel become very simple, that is, according to (6.6) K6P (xs,x6347.48) = Cieadyf IDK (xs,x6) Cien)y? 84s —17) 56-48) « (6.9) Bl wpe thor on Ro © © Pod ES] - Bee oo © o @) th) ‘The complete function K is a sum over infinitely many graphs of arbitrarily high or- der, Some examples are shown in Fig. 6.2. ‘To go beyond perturbation theory one can apply some cunning and combine at least a certain subset of the terms contributing to the infinite sum K. To that end we define the notion of a reducible interaction kernel. Itis characterized by the fact that it can be split into two unconnected parts by cutting ‘two fermion lines. Correspondingly a kernel is called irreducible if it is so densely interwoven that such a dissection is not possible. Examples of reducible and irre- ducible graphs are shown in Fig. 6.3. In Fig. 6.2 only graphs d, f, and h are reducible. ‘We shall denote the sum of all irreducible contributions to K as the irreducible inter- action kernel K. It is obvious that each reducible graph can be described by joining together several irreducible graphs from the set which contributes to the kernel K. But this can be The cut has to be applied in the “horizontal” direction. The graph (a) to (e) in Fig, 6.2 thus are not reducible Fig. 6.2. A few typical ‘graphs contributing 10 the perturbation series for the interaction kernel K @ () Fig. 6.3. Examples for a re-

You might also like