Reweighting at NLO fails in many different ways

[Alfredo Glioti]

Hello,

I am reporting the wrong and inconsistent behavior that MG displays when trying to reweight events at NLO. For the sake of reproducibility all the following runs were made using MG version 3.1.0 where I have updated the file madgraph/various/lhe_parser.py as in the recent revision here:

   https://bazaar.launchpad.net/~maddevelopers/mg5amcnlo/3.1.1/revision/966

since without this fix, the reweight procedure crashes at the beginning.

All the runs were made with the default MG settings for all the cards, except the reweight card where I am scanning on the mass of the Z. I attached the card am using to this post.

I am simulating p p > mu+ mu- at NLO QCD and then trying to reweight using the NLO mode with the card attached. Since the new version of MG is more "picky" with the specification of the coupling orders compared to the previous versions I have done a few different examples with different specification of the couplings in the process. Different specifications of the coupling order lead to a different output/crash.

Also I have simulated the process both using the default SM model and the model SMEFTatNLO 1.0.2

  http://feynrules.irmp.ucl.ac.be/wiki/SMEFTatNLO

with a restriction card, that I am attaching below, that sets to zero most operators except two. Again, the choice of the model changes the type of output/crash.

Below I am reporting a table of the six tries I have done. To recap, I have found three possible behaviors
   1) reweight is successful but all the weights are zero. In the LHE file all the weights in the list of reweights are exactly zero, same with the total cross section in the new hypotheses.
   2) crash given by a python dict error "KeyError : ''" that happens when the new weights are computed.
   3) crash given by some numerical overflow "OverflowError : (34, 'Numerical result out of range')" that happens when the new weights are computed.

For point 1 I have attached the LHE file, for point 2 and 3 I have attached the debug log of one of the failed runs.

I hope this report is comprehensive enough and that you will be able to reproduce and possibly fix all these kind of behaviors.

Best,
Alfredo

================ model: SM; no couplings specified

generate p p > mu+ mu- [QCD]
reweight is successful, but all weights are zero

================ model: SM; no QCD coupling specified

generate p p > mu+ mu- [QCD] QED=2
reweight is successful, but all weights are zero

================ model: SM; all couplings specified

generate p p > mu+ mu- [QCD] QED=2 QCD=0

reweight crashes:

command "reweight /data3/MadGraph/MG5_aMC_v3_1_0/runs/report/smwithQEDQCD/Events/run_01/events.lhe.gz_0.lhe -from_cards --multicore=create" interrupted with error:
KeyError : ''
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in '/data3/MadGraph/MG5_aMC_v3_1_0/runs/report/smwithQEDQCD/Events/run_01/events.lhe.gz_0.lhe_tag_1_debug.log'.
Please attach this file to your report.

================ model SMEFT; no couplings specified

import model SMEFTatNLO-NLOGWGphi
generate p p > mu+ mu- [QCD]

warning at generation level:

"WARNING: All Born diagrams do not factorize the same sum of power(s) of the the perturbed order(s) QED NP QCD.
This is potentially dangerous as the real-emission diagrams from aMC@NLO will not be consistent with these virtual contributions."

reweight crashes:

Command "reweight /data3/MadGraph/MG5_aMC_v3_1_0/runs/report/SMEFT/Events/run_01/events.lhe.gz_0.lhe -from_cards --multicore=create" interrupted with error:
OverflowError : (34, 'Numerical result out of range')
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in '/data3/MadGraph/MG5_aMC_v3_1_0/runs/report/SMEFT/Events/run_01/events.lhe.gz_0.lhe_tag_1_debug.log'.
Please attach this file to your report.

================ model: SMEFT; no QCD coupling specified

import model SMEFTatNLO-NLOGWGphi
generate p p > mu+ mu- [QCD] NP=2 QED=2

warning at generation level:

"WARNING: Some loop diagrams contributing to this process are discarded because they are not pure (QCD)-perturbation.
Make sure you did not want to include them."

reweight crashes:

Command "reweight /data3/MadGraph/MG5_aMC_v3_1_0/runs/report/SMEFTwithQED/Events/run_01/events.lhe.gz_0.lhe -from_cards --multicore=create" interrupted with error:
OverflowError : (34, 'Numerical result out of range')
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in '/data3/MadGraph/MG5_aMC_v3_1_0/runs/report/SMEFTwithQED/Events/run_01/events.lhe.gz_0.lhe_tag_1_debug.log'.
Please attach this file to your report.

================ model SMEFT; all coupling specified

import model SMEFTatNLO-NLOGWGphi
generate p p > mu+ mu- [QCD] NP=2 QED=2 QCD=0

warning at generation level:

"WARNING: Some loop diagrams contributing to this process are discarded because they are not pure (QCD)-perturbation.
Make sure you did not want to include them."

reweight crashes:

Command "reweight /data3/MadGraph/MG5_aMC_v3_1_0/runs/report/SMEFTwithQEDQCD/Events/run_01/events.lhe.gz_0.lhe -from_cards --multicore=create" interrupted with error:
KeyError : ''
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in '/data3/MadGraph/MG5_aMC_v3_1_0/runs/report/SMEFTwithQEDQCD/Events/run_01/events.lhe.gz_0.lhe_tag_1_debug.log'.
Please attach this file to your report.

================