vbf pT(Hjj) distribution changes between versions

Asked by Malcles

Good evening,

I am from the CMS experiment working on the SM Higgs decaying to two photons.
We have noticed that the distributions, for the VBF production process, of the generator-level pT(Hjj)
  are very different between our 2016 sample and our 2017 sample generated with madgraph5_AMC@NLO:
https://escott.web.cern.ch/escott/FinalFits/Stage1STXS/VBFcomparison/default/stack_histogram_gen_ptHjj_VBF_inclusive_.png
 Here are the versions we use and the PDF sets:

• MG5_aMC versions:
 - 2016: 2.2.2;
 - 2017: 2.4.2
• PDF sets:
 - 2016: NNPDF 3.0 NLO 5FS (ID 29220);
 - 2017: NNPDF 3.1 NNLO 5FS

This seems not to be due to the PDF sets because the difference between powheg 2016 and 2017 is
 much lower, with the same PDF set change.

Do you know what could be the reason for such a difference between versions 2.2.2 and 2.4.2?
The generator pT(Hjj) is used in our analysis and in some bins, the number of signal events derived
from the 2016 and 2017 MC are quite different because of that.
 Since there is no reason to have different numbers we wonder which sample we should trust
 or if there is an explanation for that.

Thanks a lot for your help!

Best regards,

Julie

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
marco zaro Edit question
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#1

Hi,

Did you test with 2.6.4? and/or take a look at the UpdatesNotes (between those two versions and after)
 to see if we discuss that issue and/or if this is already solved in the latest issue.

For example, one of the comment that might be relevant for you:

2.4.1: OM: Fix some bug in MLM merging, where chcluster was forced to True (introduced in 2.2.0)
Now it seems that such particular feature was again changed in 2.6.1

Cheers,

Olivier

PS: I do not have time this week to look at this, but in order to look, I would need to have a script that I can run with the different version that reproduce your findings.

Revision history for this message
Malcles (malcles) said :
#2

Hi Olivier,

Thanks very much for your reply. We will first try to look at the UpdatesNotes.

The version 2.6.1 is now being tested for CMS MC production, so we will also check
if this effect remains in 2.6.1.

Regards,

Julie

> On Jan 31, 2019, at 10:57, Olivier Mattelaer <email address hidden> wrote:
>
> Your question #678214 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/678214
>
> Olivier Mattelaer posted a new comment:
> Hi,
>
> Did you test with 2.6.4? and/or take a look at the UpdatesNotes (between those two versions and after)
> to see if we discuss that issue and/or if this is already solved in the latest issue.
>
> For example, one of the comment that might be relevant for you:
>
> 2.4.1: OM: Fix some bug in MLM merging, where chcluster was forced to True (introduced in 2.2.0)
> Now it seems that such particular feature was again changed in 2.6.1
>
> Cheers,
>
> Olivier
>
> PS: I do not have time this week to look at this, but in order to look,
> I would need to have a script that I can run with the different version
> that reproduce your findings.
>
> --
> You received this question notification because you asked the question.

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#3

Hi,

Did you have more information on this issue?
If you give me more details on how the simulation was done, I can try to reproduce this locally and give more hint on what the issue is/was.

Cheers,

Olivier

Revision history for this message
Malcles (malcles) said :
#4

Hi Olivier,

I put in cc two colleagues that are willing to follow as well this problem.

I put attached to this email the detailed configuration of the 2 dataset generated (configs.tgz).

I think you have access to cvmfs at cern, do you? Here are also the links on the gridpacks for the 2016 generation:

/cvmfs/cms.cern.ch/phys_generator/gridpacks/slc6_amd64_gcc481/13TeV/madgraph/V5_2.2.2/vbfh_5f_NLO_125/v1/vbfh_5f_NLO_125_tarball.tar.xz<http://cms.cern.ch/phys_generator/gridpacks/slc6_amd64_gcc481/13TeV/madgraph/V5_2.2.2/vbfh_5f_NLO_125/v1/vbfh_5f_NLO_125_tarball.tar.xz>

and for the 2017:

/cvmfs/cms.cern.ch/phys_generator/gridpacks/2017/13TeV/madgraph/V5_2.4.2/vbfh_5f_NLO_125/v1/vbfh_5f_NLO_125_slc6_amd64_gcc481_CMSSW_7_1_30_tarball.tar.x<http://cms.cern.ch/phys_generator/gridpacks/2017/13TeV/madgraph/V5_2.4.2/vbfh_5f_NLO_125/v1/vbfh_5f_NLO_125_slc6_amd64_gcc481_CMSSW_7_1_30_tarball.tar.xz>z

Thanks a lot for your help,

Julie

On Feb 4, 2019, at 20:52, Olivier Mattelaer <<email address hidden><mailto:<email address hidden>>> wrote:

Your question #678214 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/678214

   Status: Open => Needs information

Olivier Mattelaer requested more information:
Hi,

Did you have more information on this issue?
If you give me more details on how the simulation was done, I can try to reproduce this locally and give more hint on what the issue is/was.

Cheers,

Olivier

--
To answer this request for more information, you can either reply to
this email or enter your reply at the following page:
https://answers.launchpad.net/mg5amcnlo/+question/678214

You received this question notification because you asked the question.

Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :
#5

Ok,

So this was not at all the type of generation that I was thinking about.
And I'm likely not the best one to answer your question in the MG5aMC community.
So I will assign this to Marco since he has better knowledge than me about such type NLO run.

So for his reference the process is a NLO one in 5 flavor
p p > h j j $$ w+ w- z [QCD]

The run_card is basically the default one, but they use a very small cut on the jet (0.01 GeV) (Could you explain the logic of such soft cut?)

Cheers,

Olivier

PS: The comparison plot that you shared above seems to be protected so we are not allowed to look at it.
Could you either send it to us or otherwise describe what the issue is?

Revision history for this message
marco zaro (marco-zaro) said :
#6

Hi,
I cannot see the comparison plot either.
Which parton shower, and which version were you using in this comparison?
There have been some further changes in version 2.5.2 concerning the shower starting scale, but this should not matter in this comparison.
Thanks for letting me know
cheers,

Marco

Revision history for this message
Malcles (malcles) said :
#7

Hi Olivier and Marco,

Thanks for your answer, I attached the requested plot to this message.

Julie

[cid:03DD3016-AAD2-4268-A5B6-73881097B797]

On Feb 7, 2019, at 14:27, Olivier Mattelaer <<email address hidden><mailto:<email address hidden>>> wrote:

Your question #678214 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/678214

   Status: Open => Answered

Olivier Mattelaer proposed the following answer:
Ok,

So this was not at all the type of generation that I was thinking about.
And I'm likely not the best one to answer your question in the MG5aMC community.
So I will assign this to Marco since he has better knowledge than me about such type NLO run.

So for his reference the process is a NLO one in 5 flavor
p p > h j j $$ w+ w- z [QCD]

The run_card is basically the default one, but they use a very small cut
on the jet (0.01 GeV) (Could you explain the logic of such soft cut?)

Cheers,

Olivier

PS: The comparison plot that you shared above seems to be protected so we are not allowed to look at it.
Could you either send it to us or otherwise describe what the issue is?

--
If this answers your question, please go to the following page to let us
know that it is solved:
https://answers.launchpad.net/mg5amcnlo/+question/678214/+confirm?answer_id=4

If you still need help, you can reply to this email or go to the
following page to enter your feedback:
https://answers.launchpad.net/mg5amcnlo/+question/678214

You received this question notification because you asked the question.

Revision history for this message
marco zaro (marco-zaro) said :
#8

It is still not visible…
apparently you cannot attach files to launchpad questions

> On 7 Feb 2019, at 14:37, Malcles <email address hidden> wrote:
>
> Question #678214 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/678214
>
> Status: Answered => Open
>
> Malcles is still having a problem:
>
> Hi Olivier and Marco,
>
> Thanks for your answer, I attached the requested plot to this message.
>
> Julie
>
> [cid:03DD3016-AAD2-4268-A5B6-73881097B797]
>
>
> On Feb 7, 2019, at 14:27, Olivier Mattelaer
> <<email address hidden><mailto:<email address hidden>>>
> wrote:
>
> Your question #678214 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/678214
>
> Status: Open => Answered
>
> Olivier Mattelaer proposed the following answer:
> Ok,
>
> So this was not at all the type of generation that I was thinking about.
> And I'm likely not the best one to answer your question in the MG5aMC community.
> So I will assign this to Marco since he has better knowledge than me about such type NLO run.
>
> So for his reference the process is a NLO one in 5 flavor
> p p > h j j $$ w+ w- z [QCD]
>
> The run_card is basically the default one, but they use a very small cut
> on the jet (0.01 GeV) (Could you explain the logic of such soft cut?)
>
> Cheers,
>
> Olivier
>
> PS: The comparison plot that you shared above seems to be protected so we are not allowed to look at it.
> Could you either send it to us or otherwise describe what the issue is?
>
> --
> If this answers your question, please go to the following page to let us
> know that it is solved:
> https://answers.launchpad.net/mg5amcnlo/+question/678214/+confirm?answer_id=4
>
> If you still need help, you can reply to this email or go to the
> following page to enter your feedback:
> https://answers.launchpad.net/mg5amcnlo/+question/678214
>
> You received this question notification because you asked the question.
>
> --
> You received this question notification because you are assigned to this
> question.

Revision history for this message
Malcles (malcles) said :
#9

This link should work, hopefully:

https://cernbox.cern.ch/index.php/s/f0PmqkRMUVW9PLn

Julie

Revision history for this message
marco zaro (marco-zaro) said :
#10

Thanks, it works, i can see the difference
What about the parton shower employed and their version?
Marco

Revision history for this message
Malcles (malcles) said :
#11

The parton shower is pythia8, so if you do not receive the attached files you did not received the detailed
 configurations I attached in a previews message. I copied them at this link:

https://cernbox.cern.ch/index.php/s/19Mi2RofjrVdvwe

Thanks,

Julie

> On Feb 7, 2019, at 15:02, marco zaro <email address hidden> wrote:
>
> Your question #678214 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/678214
>
> marco zaro posted a new comment:
> Thanks, it works, i can see the difference
> What about the parton shower employed and their version?
> Marco
>
> --
> You received this question notification because you asked the question.

Revision history for this message
marco zaro (marco-zaro) said :
#12

But I cannot find the Pythia8 version you were using

Revision history for this message
Malcles (malcles) said :
#13

Hi Marco,

for 2017 it was: PYTHIA8 2.0 version 8.230, whereas for 2016 it was: PYTHIA 8.205.

Cheers

Revision history for this message
marco zaro (marco-zaro) said :
#14

ok, can you try to shower 2017 LHE events with py8205 and see how this behaves w.r.t the two other samples?
also, can you plot the scalup entry of the LHE events in the two samples (this can be done without shower)?
thanks,
Marco

> On 7 Feb 2019, at 17:02, Malcles <email address hidden> wrote:
>
> Question #678214 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/678214
>
> Malcles gave more information on the question:
> Hi Marco,
>
> for 2017 it was: PYTHIA8 2.0 version 8.230, whereas for 2016 it was:
> PYTHIA 8.205.
>
> Cheers
>
> --
> You received this question notification because you are assigned to this
> question.

Can you help with this problem?

Provide an answer of your own, or ask Malcles for more information if necessary.

To post a message you must log in.