Comment 2 for bug 921487

more in-depth, thanks to Florian Staub:

The result for a "diff -rq" between both created directory (for a
positive and negative PDG of the chargino) looks like:

============
....
Files dir_run_check_SA_posPDG//Source/DHELAS/aloha_file.inc and
dir_run_check_SA_negPDG//Source/DHELAS/aloha_file.inc differ
Only in dir_run_check_SA_negPDG//Source/DHELAS: FFS1C1_0.f
Only in dir_run_check_SA_negPDG//Source/DHELAS: FFS1C1_2_0.f
Only in dir_run_check_SA_negPDG//Source/DHELAS: FFS1C1_2_3.f
Only in dir_run_check_SA_negPDG//Source/DHELAS: FFS1C1_3.f
Only in dir_run_check_SA_negPDG//Source/DHELAS: FFS2C1_0.f
Only in dir_run_check_SA_negPDG//Source/DHELAS: FFS2C1_3.f
Files dir_run_check_SA_posPDG//SubProcesses/P0_ddx_x2mx2p/configs.inc
and dir_run_check_SA_negPDG//SubProcesses/P0_ddx_x2mx2p/configs.inc
differ
....
============

So, it seems at first glance that Madgraph generates helicity
amplitudes for the conjugated vertex. But, a comparison between
FFS1C1_0.f and FFS1_0.f leads to

=====
< SUBROUTINE FFS1C1_0(F1,F2,S3,COUP,VERTEX)
---
> SUBROUTINE FFS1_0(F1,F2,S3,COUP,VERTEX)
====

So, it's just a renaming of the routines. In that way everything would
be fine, if Madgraph would use always FFS1C1_0 and FFS1_0 and do no
other changes. However, comparing the two files for the matrix
elements (matrix.f), we see

==================
273,274c276,278
< CALL FFS1_2_0(W(1,3),W(1,2),W(1,10),GC_3554,GC_3555,AMP(6))
< CALL FFS1_2_3(W(1,1),W(1,4),GC_2344,GC_2345,MSU2, WSU2, W(1,11))
---
> CALL FFS1C1_2_0(W(1,12),W(1,2),W(1,11),GC_3554,GC_3555,AMP(6))
> CALL FFS1C1_2_3(W(1,3),W(1,10),GC_2344,GC_2345,MSU2, WSU2, W(1
> $ ,13))
276,277c280,282
===================

So, it uses the same expressions for the vertices (GC_3554, ...) in
the amplitudes, but combines them with different W functions (momenta?). It might (?)
by possible to recover the same result, if also the hermitian
conjugated of the vertex is used (i.e. interchanging e.g. GC_3554 <=>
GC_2345 and adding a complex conjugation, or something like that).