EWdim6 reweighting crashes in 5f scheme

[Matthias M]

Dear authors,

I'm trying to study pp->WW production with anomalous couplings using the EWdim6 EFT. To avoid a large number of separate samples, I use the reweight feature. This has worked in the past, but since I upgraded to version 5.2.3.2.2 I see the following crash. The initial event generation works as expected, but after some successfully reweighted events the error below appears:

INFO: starts to compute weight for events with the following modification to the param_card:

INFO: set param_card dim6 1 0.0 # orig: 1.0
set param_card dim6 2 0.0 # orig: 1.0
set param_card dim6 3 0.0 # orig: 10.0

INFO: Event nb 0 current time: 20h40
INFO: Event nb 10 1m 37s
INFO: Event nb 20 3m 32s
INFO: Event nb 30 4m 51s
INFO: Event nb 40 5m 6s
INFO: Event nb 50 5m 26s
INFO: Event nb 60 5m 26s
INFO: Event nb 70 5m 26s
INFO: Event nb 80 5m 32s
Command "reweight run_01" interrupted with error:
KeyError : ((-5, 5), (-16, -3, 4, 15))
Please report this bug on https://bugs.launchpad.net/madgraph5
More information is found in 'MG5_aMC_v2_3_2_2/WW_aTGC_reweight/run_01_tag_3_debug.log'.
Please attach this file to your report.

The key makes me suspect, that there is an issue with the five-flavor scheme. Indeed when I run a comparable configuration in the 4f-scheme, 10k events are successfully reweighted.
Could there be a regression in the handling of the 5f scheme in the revamp of the reweigthing code?

Yours
   Matthias

[Olivier Mattelaer (olivier-mattelaer)]

Hi,

The main reason, is that you do not really do the computation in 5F since I see that
you have the mass of the b which is different of zero. I guess if you force the b mass to
be zero (at the model level) then you should not have any problem.
(Note the you can define such type of model via the “customise_model —save=NAME” options)

If you want to stick with this small inconsistency.
Here is a fix:

=== modified file 'madgraph/various/banner.py'
--- madgraph/various/banner.py 2015-10-01 16:00:08 +0000
+++ madgraph/various/banner.py 2015-10-09 10:34:09 +0000
@@ -536,7 +536,7 @@
      if len(arg) == 1:
          if tag == 'mg5proccard':
              try:
- return card.info[arg[0]]
+ return card.get(arg[0])
              except KeyError, error:
                  if 'default' in opt:
                      return opt['default']
@@ -850,11 +850,18 @@
          # update the counter to pass to the next element
          nline -= 1

- def __getattr__(self, tag, default=None):
+ def get(self, tag, default=None):
      if isinstance(tag, int):
          list.__getattr__(self, tag)
      elif tag == 'info' or tag == "__setstate__":
          return default #for pickle
+ elif tag == "multiparticles":
+ out = []
+ for line in self:
+ if line.startswith('define'):
+ name, content = line[7:].split('=',1)
+ out.append((name, content))
+ return out
      else:
          return self.info[tag]
=== modified file 'madgraph/interface/reweight_interface.py'
--- madgraph/interface/reweight_interface.py 2015-10-01 16:00:08 +0000
+++ madgraph/interface/reweight_interface.py 2015-10-09 10:32:30 +0000
@@ -921,6 +921,11 @@
      commandline="import model %s " % modelpath
      mgcmd.exec_cmd(commandline)

+ #multiparticles
+ print self.banner.get('proc_card', 'multiparticles')
+ for name, content in self.banner.get('proc_card', 'multiparticles'):
+ mgcmd.exec_cmd("define %s = %s" % (name, content))
+
      # 2. compute the production matrix element -----------------------------
      processes = [line[9:].strip() for line in self.banner.proc_card
                   if line.startswith('generate')]

[Matthias M (mmgdw)]

Dear Olivier,

thank you for the quick reply. I can confirm that the patch resolves the problem. For my studies I've updated the model, though, which seems to be the more correct choice.