Problem with hgg_plugin

Asked by Amin Aboubrahim on 2018-02-21

Dear experts,

I am using hgg_plugin to generate samples of Higgs production through gluon fusion. My proc_card reads:

import model sm-no_b_mass -name
define X = t t~
define V = w+ w- z a

add model hgg_plugin
generate p p > h $ X V h HIG=1 QED=1 HIW=0 @0
add process p p > h j $ X V h HIG=1 QED=1 HIW=0 @1
add process p p > h j j $ X V h HIG=1 QED=1 HIW=0 @2
add process p p > h j j j $ X V h HIG=1 QED=1 HIW=0 @3

When running the gridpack, I get a warning:

P2_gg_hgg SubProcesses doesn't have available phase-space.
            Please check mass spectrum.

then an error:

INFO: Idle: 0, Running: 7, Completed: 79 [ 10m 10s ]
CRITICAL: Fail to run correctly job 11595733.
            with option: {'log': None, 'stdout': None, 'argument': ['0', '4', '6'], 'nb_submit': 1, 'stderr': None, 'prog': '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/survey.sh', 'output_files': ['G4', 'G6'], 'time_check': 1519185112.906674, 'cwd': '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/P3_gg_hgqq', 'required_output': ['G4/results.dat', 'G6/results.dat'], 'input_files': ['madevent', 'input_app.txt', 'symfact.dat', 'iproc.dat', 'dname.mg', '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/randinit', '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/lib/PDFsets']}
            file missing: /home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/P3_gg_hgqq/G6/results.dat
            Fails 1 times
            No resubmition.
INFO: Idle: 0, Running: 6, Completed: 80 [ 10m 40s ]
scancel: error: slurm_kill_job2() failed Invalid job id specified
scancel: error: slurm_kill_job2() failed Job/step already completing or completed
INFO: All jobs finished
INFO: Idle: 0, Running: 0, Completed: 86 [ 11m 10s ]
  === Results Summary for run: run_02 tag: tag_1 ===

     Cross-section : 110.2 +- 0.2423 pb
     Nb of events : 0

INFO: End survey
combine_events
INFO: Combining Events
store_events
INFO: Storing parton level results
INFO: End Parton
decay_events -from_cards
create_gridpack

I thought it was a problem with slurm cluster but I got the same thing when I ran it again and the crash happened at the same step. There is a file "error" in the directory and it contains this:

Error: Failed despite same graph: 28 . Have jets (>0) 0 0 0 1 0 0 , should be 5 6

I don't understand what that means.
Is there something wrong I have done or some bug?

Thanks a lot,
Amin

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2018-02-23
Last query:
2018-02-23
Last reply:
2018-02-23

Hi,

This is clearly a bug.
I have succesfully reproduce it and locate the function responsible for such bad behaviour (I think).
But I need more time to find the correct patch (that part of the code is very complex) and to test it.

Cheers,

Olivier

> On 21 Feb 2018, at 05:27, Amin Aboubrahim <email address hidden> wrote:
>
> New question #664726 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/664726
>
> Dear experts,
>
> I am using hgg_plugin to generate samples of Higgs production through gluon fusion. My proc_card reads:
>
> import model sm-no_b_mass -name
> define X = t t~
> define V = w+ w- z a
>
> add model hgg_plugin
> generate p p > h $ X V h HIG=1 QED=1 HIW=0 @0
> add process p p > h j $ X V h HIG=1 QED=1 HIW=0 @1
> add process p p > h j j $ X V h HIG=1 QED=1 HIW=0 @2
> add process p p > h j j j $ X V h HIG=1 QED=1 HIW=0 @3
>
> When running the gridpack, I get a warning:
>
> P2_gg_hgg SubProcesses doesn't have available phase-space.
> Please check mass spectrum.
>
> then an error:
>
> INFO: Idle: 0, Running: 7, Completed: 79 [ 10m 10s ]
> CRITICAL: Fail to run correctly job 11595733.
> with option: {'log': None, 'stdout': None, 'argument': ['0', '4', '6'], 'nb_submit': 1, 'stderr': None, 'prog': '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/survey.sh', 'output_files': ['G4', 'G6'], 'time_check': 1519185112.906674, 'cwd': '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/P3_gg_hgqq', 'required_output': ['G4/results.dat', 'G6/results.dat'], 'input_files': ['madevent', 'input_app.txt', 'symfact.dat', 'iproc.dat', 'dname.mg', '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/randinit', '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/lib/PDFsets']}
> file missing: /home/abouibrahim.a/MG5_aMC_v2_6_0/bin/H/SubProcesses/P3_gg_hgqq/G6/results.dat
> Fails 1 times
> No resubmition.
> INFO: Idle: 0, Running: 6, Completed: 80 [ 10m 40s ]
> scancel: error: slurm_kill_job2() failed Invalid job id specified
> scancel: error: slurm_kill_job2() failed Job/step already completing or completed
> INFO: All jobs finished
> INFO: Idle: 0, Running: 0, Completed: 86 [ 11m 10s ]
> === Results Summary for run: run_02 tag: tag_1 ===
>
> Cross-section : 110.2 +- 0.2423 pb
> Nb of events : 0
>
> INFO: End survey
> combine_events
> INFO: Combining Events
> store_events
> INFO: Storing parton level results
> INFO: End Parton
> decay_events -from_cards
> create_gridpack
>
> I thought it was a problem with slurm cluster but I got the same thing when I ran it again and the crash happened at the same step. There is a file "error" in the directory and it contains this:
>
> Error: Failed despite same graph: 28 . Have jets (>0) 0 0 0 1 0 0 , should be 5 6
>
> I don't understand what that means.
> Is there something wrong I have done or some bug?
>
> Thanks a lot,
> Amin
>
>
>
>
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Amin Aboubrahim (amin83) said : #2

Hi Olivier,

Thank you for checking.
Do you think I can use the heft model instead?

Best,
Amin

Amin Aboubrahim (amin83) said : #3

Hi again,

I found a similar issue in another process, even though the error is different. I am generating p p > leptons + jets and with gridpack I get:

P2_gq_llgq,P2_gg_vltapqq,P2_qq_vltapqq SubProcesses doesn't have available phase-space.
            Please check mass spectrum.

then the error:

Error when reading /home/abouibrahim.a/MG5_aMC_v2_6_0/bin/LL/SubProcesses/P2_gg_llqq/G1/results.dat
Command "generate_events -f" interrupted with error:
ValueError : need more than 5 values to unpack
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/LL/run_01_tag_1_debug.log'.
Please attach this file to your report.

Is that the same problem as before? I can send you the log file if needed.

Thank you again,
Amin

Hi,

I would actually advise you to use the heft model. But the problem here is not in the model definition.
So both leads to this crash.

For the second problem, if you can give me the details, I will check after having fix this one.

Cheers,

Olivier

> On 22 Feb 2018, at 05:52, Amin Aboubrahim <email address hidden> wrote:
>
> Question #664726 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/664726
>
> Amin Aboubrahim posted a new comment:
> Hi Olivier,
>
> Thank you for checking.
> Do you think I can use the heft model instead?
>
> Best,
> Amin
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi,

The faulty change in the code seems to be this one:
https://bazaar.launchpad.net/~mg5core1/mg5amcnlo/2.6.2/revision/258.1.62

Reverting it seems to solve the issue (I did not make a full run, just run the previously faulty channel)

But reverting such patch is likely to be a bad solution since it was done for a reason. I still have to investigate but if you are in a rush, try it and if it goes through then you should be fine.

Cheers,

Olivier

Amin Aboubrahim (amin83) said : #6

Hi Olivier,

What do you mean by reverting the patch? You mean undoing the change in code? How can I do that?

For the other crash, here are the details:

define l+ = e+ mu+ ta+
define l- = e- mu- ta-
define vl = ve vm vt
define vl~ = ve~ vm~ vt~
define lept = l+ l- vl vl~
generate p p > lept lept $ X V h QED=2 @0
add process p p > lept lept j $ X V h QED=2 @1
add process p p > lept lept j j $ X V h QED=2 @2

Thank you again.

Best,
Amin

you can use patch -R

(or just apply the patch without the option -R and most patch will propose to you to run it with -R)

thanks for the process, is this also in heft? or in sm?

Cheers,

Olivier

PS: I have run so far tt~/Z/w+/h up to three jets and they all work fine.

> On 22 Feb 2018, at 15:26, Amin Aboubrahim <email address hidden> wrote:
>
> Question #664726 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/664726
>
> Amin Aboubrahim posted a new comment:
> Hi Olivier,
>
> What do you mean by reverting the patch? You mean undoing the change in
> code? How can I do that?
>
> For the other crash, here are the details:
>
> define l+ = e+ mu+ ta+
> define l- = e- mu- ta-
> define vl = ve vm vt
> define vl~ = ve~ vm~ vt~
> define lept = l+ l- vl vl~
> generate p p > lept lept $ X V h QED=2 @0
> add process p p > lept lept j $ X V h QED=2 @1
> add process p p > lept lept j j $ X V h QED=2 @2
>
> Thank you again.
>
> Best,
> Amin
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Amin Aboubrahim (amin83) said : #8

I got you.
Yes, this is also sm-no_b_mass.

Thank you,
Amin

Amin Aboubrahim (amin83) said : #9

I forgot to mention that I am using version 2.6.0. Does reverting this patch work on 2.6.0?

Best,
Amin

should be yes.

Olivier

> On 22 Feb 2018, at 16:16, Amin Aboubrahim <email address hidden> wrote:
>
> Question #664726 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/664726
>
> Amin Aboubrahim posted a new comment:
> I forgot to mention that I am using version 2.6.0. Does reverting this
> patch work on 2.6.0?
>
> Best,
> Amin
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Amin Aboubrahim (amin83) said : #11

Hi Olivier,

After reversing the patch, the first issue was solved, i.e. gluon fusion using hgg_plugin has completed with out errors.
Now the second problem (lepton production) still persists. I am getting this warning and error:

P2_gq_tamtapgq,P2_gq_vltapgq,P0_qq_tamtap SubProcesses doesn't have available phase-space.
            Please check mass spectrum.

Error when reading /home/abouibrahim.a/MG5_aMC_v2_6_0/bin/LL/SubProcesses/P2_gg_llqq/G1/results.dat
Command "generate_events -f" interrupted with error:
ValueError : need more than 5 values to unpack
Please report this bug on https://bugs.launchpad.net/mg5amcnlo
More information is found in '/home/abouibrahim.a/MG5_aMC_v2_6_0/bin/LL/run_01_tag_1_debug.log'.
Please attach this file to your report.

What could be the problem here?

Thank you,
Amin

Hi Amin,

I still have to convince myself that such change is the correct one to do. When I will be convinced (or find a change that will convince me) then I will move on your second problem.

Cheers,

Olivier

Hi,

Just to point that indeed this is not good enough since it fails
for qq~>aagqq~ in the heft
(actually both .and. and .or. fails).

Cheers,

Olivier

Amin Aboubrahim (amin83) said : #14

Thanks Olivier.
I will wait for a resolution to this issue.

Best,
Amin

Hi,

I think that I have found a correct fix for both problem

1) for your lepton issue:
https://bazaar.launchpad.net/~mg5core1/mg5amcnlo/2.6.2/revision/310

2) and for the original issue,you can keep what you did.
I have a more general fix pushed as well:
https://bazaar.launchpad.net/~mg5core1/mg5amcnlo/2.6.2/revision/311

Thanks a lot for those reports,

Olivier

On 23 Feb 2018, at 16:07, Amin Aboubrahim <<email address hidden><mailto:<email address hidden>>> wrote:

Question #664726 on MadGraph5_aMC@NLO changed:
https://answers.launchpad.net/mg5amcnlo/+question/664726

Amin Aboubrahim posted a new comment:
Thanks Olivier.
I will wait for a resolution to this issue.

Best,
Amin

--
You received this question notification because you are an answer
contact for MadGraph5_aMC@NLO.

Amin Aboubrahim (amin83) said : #16

Hi Olivier,

I applied the patches and everything is working now.
Thank you very much.

Best,
Amin

Amin Aboubrahim (amin83) said : #17

Thanks Olivier Mattelaer, that solved my question.

Dear Olivier, I have a similar issue, using my private model files for hjj. After installing the patch the production seems to work but I still get some error messages a posteriori when I run MadAnalysis.

Before I go into more details I wanted to make sure I am applying the patch correctly. What I did is download the tarball for the revision 315 that you mention here and do a new madgraph installation with that (pdfs, pythia, madanalysis, model files....). Is this correct? Or is it a simpler way to just apply the patch on my older madgraph installation?

Thanks

Hi,

I was not aware that tarball were available for each version.
That's actually good to know.
That's should be a perfect way to have a correct version actually.
No for the MA5 problem please open a separate issue to not mix the problem.
(especially since this will be assigned to MA5 author and not to me)

Cheers,

Olivier

> On 6 Mar 2018, at 19:17, Raquel Gomez Ambrosio <email address hidden> wrote:
>
> Question #664726 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/664726
>
> Raquel Gomez Ambrosio posted a new comment:
> Dear Olivier, I have a similar issue, using my private model files for
> hjj. After installing the patch the production seems to work but I still
> get some error messages a posteriori when I run MadAnalysis.
>
> Before I go into more details I wanted to make sure I am applying the
> patch correctly. What I did is download the tarball for the revision 315
> that you mention here and do a new madgraph installation with that
> (pdfs, pythia, madanalysis, model files....). Is this correct? Or is it
> a simpler way to just apply the patch on my older madgraph installation?
>
> Thanks
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.