change matrix.f to include the choice of helicity/color within that file.

[oliviermattelaer]

Idea

update the fortran code to have the same interface as for GPU and therefore avoid to put the JAMP2 in a common block

• avoid to have to transfer it from GPU to CPU
• good time to test the low memory selection of color to avoid to have all jamp2 in memory at the same time (see Note on color structure #17)

Todo:

• change signature of smatrix
  • without helicity recycling
    - [ ] pass to the low memory mode (see Note on color structure #17)
  • wih helicity recycling
    - [ ] check if relevant to use the low memory mode (see Note on color structure #17)
• change the signature of matrix
  • without helicity recycling
• change addmothers.f
• change the events writing
  • temporary format (avoid the use of buffer)
  • final format

[oliviermattelaer]

note

    CHARACTER*101         HEL_BUFF
    COMMON/TO_HELICITY/  HEL_BUFF
 WRITE(HEL_BUFF,'(20i5)')(NHEL(II,I),II=1,NEXTERNAL)

(need to be replaced via the call to ihel)
-> need to have a function get_helicity(iproc, ihel)
call get_helicity(iproc, ihel, nhel)
that calls get_nhelX(...)

[oliviermattelaer]

case without plugin/helicity recycling/mlm
implemented here:

mg5amcnlo/mg5amcnlo@0587cbb
following commit is make the change consistent with helicity recycling
mg5amcnlo/mg5amcnlo@18a90f5

[oliviermattelaer]

Validation:

• check that the code compile
• check that the cross-section is correct
• Helicity
  • are filled
    • returned non zero from smatrix
    • are passed non zero (and same as smatrix) to write_leshouches
    • are returning non zero from the new function for getting the helicity
      • function return 0 0 0 0 in last commit
  • have correct rate
    • wrong with helicity recycling since the index is remapped to not have "hole" (fixed in mg5amcnlo/mg5amcnlo@3622a74)
• Color
  • are filled
  • have correct rate
• JAMP2
  • check that all jamp2 have been removed from memory (especially the vectorised version)