MG7 Hall of Shame

[theoheimel]

Collect processes here that are currently not working, or give the wrong results.

Careful: the default scale choice in MG5 is not yet implemented in MG7, so if you want to compare cross section results you have to change the scaled to fixed (current default in MG7) or change it in both to, for instance, half transverse mass.

[theoheimel]

✅ FIXED

In some cases the export_mg7.py script (https://github.com/MadGraphTeam/MadGraph7/blob/main/madgraph/iolibs/export_mg7.py#L89) yields an empty list for the available flavors of some channels. This leads to wrong integration results, especially when using MadNIS with trained channel weights. Found for example for _quark _anti_quark > d d~ (second channel), _quark _quark > t t~ _quark _quark (5th and 6th channel), _quark _quark > _quark _quark g (4th and 5th channel).

-> partially fixed by #18 , however still wrong results for some diagrams. For _quark _quark > g _quark _quark, diagrams 1, 2, 9, 10, flavor configurations of the type xy>gxy are wrongly ruled out.

[theoheimel]

✅ FIXED

Process: p p > Z, Z > e+ e- at fixed scale
MG5: 1422 +- 0.8357 pb
MG7: 630.87(77) pb

EDIT: this is related to cuts. In the decay syntax MG5 sets no cuts on leptons. After removing the cuts, MG7 produces the same result.

[theoheimel]

✅ FIXED

Process _quark _quark > _quark _quark g yields 5.95(88)e+05 pb in MG7, whereas _quark _quark > g _quark _quark yields 4.6054(66)e+05 pb.

Potentially fixed by #7 , check again once merged

EDIT: #7 indeed fixed the discrepancy between the two orders with 6.4623(94)e+05 and 6.4592(96)e+05, respectively. However it disagrees with the MG5 results 1.037e+06 +- 2719

EDIT2: the remaining difference is from two bugs: the active_flavors bug (see above), and madspace not respecting that some flavor configs need a mirror process and others don't. Working on the latter in feat-flavorsym-fix.

[oliviermattelaer]

✅ FIXED

g g > h QCD=0 [QCD]; output madevent; launch

• fixed in 3.7.2 (different error here it is the recent color change that triggers it)
• added to CI (MG5: 15.72 +- 0.01514)
• need merging with latest madspin_density: Merge mg5/madspin_density into feat-madmatrix-aloha-merge #16
• fixed

[stloufra]

✅ FIXED

set gauge FD
u u > h u u g g
MG5: 0.006055 ± 0.0001669
MG7: 8.426(11)e-5

[theoheimel]

✅ FIXED

@stloufra I think this one was (at least mostly) fixed by 3b80ea4. I get 6.685(14)e-03 in MG7. There's still a bit of a difference which might or might not be significant (MG5 has a slight bias towards lower cross sections if you ask for low numbers of unweighted events).

[theoheimel]

✅ FIXED

Processes: p p > t t~ j j, p p > t t~ j j j, ...

get_external_flavors_with_iden returns impossible flavors. It broke after merging #18

For the example of _quark g > t t~ _quark g, before the merge the returned flavors were

dict_values([[(1, 1, 1, 1, 1, 1), (2, 1, 1, 1, 2, 1), (3, 1, 1, 1, 3, 1), (4, 1, 1, 1, 4, 1)]])

and afterwards they were

dict_values([[(1, 1, 1, 1, 1, 1), (2, 1, 1, 1, 2, 1), (3, 1, 1, 1, 3, 1), (4, 1, 1, 1, 4, 1)], [(1, 1, 1, 1, 2, 1), (1, 1, 1, 1, 3, 1), (1, 1, 1, 1, 4, 1), (2, 1, 1, 1, 1, 1), (2, 1, 1, 1, 3, 1), (2, 1, 1, 1, 4, 1), (3, 1, 1, 1, 1, 1), (3, 1, 1, 1, 2, 1), (3, 1, 1, 1, 4, 1), (4, 1, 1, 1, 1, 1), (4, 1, 1, 1, 2, 1), (4, 1, 1, 1, 3, 1)]])

[theoheimel]

✅ FIXED

generate p p > j j j and generate p p > 3j behave differently. The second syntax does not appear to use multiparticles in the final state.

[oliviermattelaer]

✅ FIXED

generate p p > t t~ j j , t > w+ b, t~ > w- b~
output
launch

Traceback (most recent call last):
  File "/Users/omattelaer/Documents/git_workspace/MadGraph7/.claude/worktrees/jolly-dhawan-48ffcc/PROCMG7_sm_35/bin/generate_events", line 8, in <module>
    main()
    ~~~~^^
  File "/Users/omattelaer/Documents/git_workspace/MadGraph7/.claude/worktrees/jolly-dhawan-48ffcc/madgraph/iolibs/template_files/mg7/madevent.py", line 1744, in main
    run_generation()
    ~~~~~~~~~~~~~~^^
  File "/Users/omattelaer/Documents/git_workspace/MadGraph7/.claude/worktrees/jolly-dhawan-48ffcc/madgraph/iolibs/template_files/mg7/madevent.py", line 1730, in run_generation
    run_single()
    ~~~~~~~~~~^^
  File "/Users/omattelaer/Documents/git_workspace/MadGraph7/.claude/worktrees/jolly-dhawan-48ffcc/madgraph/iolibs/template_files/mg7/madevent.py", line 1644, in run_single
    process.survey()
    ~~~~~~~~~~~~~~^^
  File "/Users/omattelaer/Documents/git_workspace/MadGraph7/.claude/worktrees/jolly-dhawan-48ffcc/madgraph/iolibs/template_files/mg7/madevent.py", line 437, in survey
    self.event_generator = self.survey_phasespaces(self.phasespaces)
                           ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^
  File "/Users/omattelaer/Documents/git_workspace/MadGraph7/.claude/worktrees/jolly-dhawan-48ffcc/madgraph/iolibs/template_files/mg7/madevent.py", line 426, in survey_phasespaces
    event_generator = self.build_event_generator(ps_filtered)
  File "/Users/omattelaer/Documents/git_workspace/MadGraph7/.claude/worktrees/jolly-dhawan-48ffcc/madgraph/iolibs/template_files/mg7/madevent.py", line 394, in build_event_generator
    channel.event_generator = ms.ChannelEventGenerator(
                              ~~~~~~~~~~~~~~~~~~~~~~~~^
        contexts=self.contexts,
        ^^^^^^^^^^^^^^^^^^^^^^^
    ...<6 lines>...
        histograms=subproc.histograms
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    )
    ^
RuntimeError: Argument, value 2: Invalid shape, expected (105), got (123)
 [in SubchannelWeights]