Name: V. Shtabovenko Date: 03/11/19-09:55:54 AM Z


Hi Vitaly,

if you want to be completely general, then you indeed
will have to handle the internal summations before squaring
the amplitude. Even if there are no IndexDeltas with interal indices ,
you can still have e.g. quark propagators with masses that depend on a
generation index. I think this is also mentioned on p.43 of the FeynArts
manual.

I’m also not sure if IndexDeltas with interal indices exist in
principle, but perhaps you could just write Thomas an email asking about
that.

Another possibility is to study the corresponding code lines in
FormCalc, since it obviously must know how to handle all those
summations in a proper way.

I’m always ready to fix bugs or remove obstacles that prevent people
from using FeynCalc in a particular way, but I’m not particularly keen
to implement the handling of all that SumOver stuff out of the box myself.
Since different generation indices may also appear inside masses or
couplings, one would need to have something more generic than Contract,
especially since those summations must not obey Einstein’s convention.

Cheers,
Vladyslav

Am 11.03.19 um 03:42 schrieb Vitaly Magerya:
> Hi, Vladyslav. Thanks for the fix.
>
> I’m thinking that in a more general setup only the SumOver[…,
> External] could be dropped from the conjugate amplitude; all of the
> internal ones should be taken into account, in a case of quark loops
> in both the left and the right amplitudes.
>
> I’m not sure if there can be IndexDeltas with Generation indices
> involved in an internal loop though; if so, then those need to be
> summed over *before* the multiplication by the conjugate amplitude,
> because FCRenameDummyIndices will not rename them.
>
> I agree that qgraf/FORM/color.h combination is more flexible, of
> course. On the other hand it is more bug prone too, so it makes
> sense to double-check it with FeynArts+FeynCalc.
>