Number of events are being generated less than required

Hi,

Thanks for the information, it is always interesting to learn process where our phase-space integrator struggles.
The bad news for you is that updating the phase-space integrator is not something trivial (taking years and without any guarantee of success). So I do not have high hope to be able to really solve our phase-space integrator to this case.

I have investigate a bit the first of your process:
And the first point is that the issue is at production level, not within the decay.
since "generate mu- mu+ > wx j j" is failing to generate enough events (at 20 TeV collider).

Then I tried to have some stronger cut on all particle to avoid any soft divergency
(set ptheavy 200; set ptj 200)
and then it worked:
  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.0008681 +- 2.264e-06 pb
     Nb of events : 10000
     (run time is around 1m20 min)

Then I tried to remove the cut on the W (since the soft is protected by his mass):
(set ptheavy 0; set ptj 200)
  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.001336 +- 3.406e-06 pb
     Nb of events : 10000
    (run time is around 1m40s)

Then I reduce the remaining cut to 50:
(set ptheavy 0; set ptj 50)
Which fails.

Then I investigate the case where I was keeping those "soft" cut on the jet.
But putting back the cut on ptheavy:
(set ptheavy 200; set ptj 50)
fails in ~10 min...

Adding a mmjj cut is helping a lot:
(set ptheavy 200; set ptj 50; set mmjj 200)
  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.0009496 +- 2.592e-06 pb
     Nb of events : 10000
     (run time ~20s)

reducing that cut
(set ptheavy 200; set ptj 50; set mmjj 100)
  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.00102 +- 3.428e-06 pb
     Nb of events : 10000
     (still around 20s)

reducing that cut to be below the WW threshold for the process:
(set ptheavy 200; set ptj 50; set mmjj 80)
which also struggle to generate events:

I then tried the following definition:
generate mu- mu+ > wx j j $ wx z
(so forbidding the w/z propagator to be onshell when decaying into two jet)
(with cut: set ptheavy 200; set ptj 50; set mmjj 80)
then
  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.0009964 +- 2.233e-06 pb
     Nb of events : 10000
    (run time around 2 min)

and then the same with only default cut (so quite soft one on the jet and none on the W)
  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.001664 +- 6.828e-06 pb
     Nb of events : 7585
(so seems really at the border of efficiency here again --likely too soft cut-- for protecting from issue with the photon)

So the problem seems to be the double resonances part of the phase-space which is problematic, however the process:
generate p p > wx wx
does not struggle (with ptheavy set 200)
as well as (without any cut)
generate p p > w+ w-, w+ > mu+ vm, w- > j j

So the reason can be related to the assymetric handling of the phase-space when you have a double resonance (and where gauge invariance can be an issue if you do not forbid such area with cut).
To check that, I have tested the following syntax:
generate mu+ mu- > mu+ vm j j
(with ptl=50, ptj=50)

  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.003928 +- 1.189e-05 pb
     Nb of events : 10000
    (around 25s)

I also checked that syntax with the complex mass scheme to avoid any gauge issue related to the finite value of the width (which is only working if you do not select a subset of diagram so only for this syntax). This work even faster --not the previous syntax was slow--
  === Results Summary for run: run_01 tag: tag_1 ===

     Cross-section : 0.003933 +- 1.372e-05 pb
     Nb of events : 10000
     (run time around 13s)

So in conclusion the issue for the first process (and likely the same for the two others) is the fact that you single out one W to decay while you do have diagram with two of them. You either need to forbid such double onshell contribution via stronger cut or do not use such type of syntax to be inclusive with all type of decay (which then allow to use complex mass scheme to be fully gauge invariant --and even faster--).

Given those, I tend to believe that the real issue is more on the physics side than on the phase-space integrator itself.
In any case, I hope that one of those two solution can fix your issue.

Cheers,

Olivier