MadGraph5_aMC@NLO

Bug #1334014
Comment #12

Comment 12 for bug 1334014

Revision history for this message

Paolo Torrielli (paolo-torrielli) wrote on 2014-07-16: Re: [Bug 1334014] crash at Compiling MCatNLO

#12

Hi Azar,

event generation and showering are independent, so when you get
a run summary, the LH event file has already been produced and
the cross section is reliable.

As for my suggested patch, sorry I was referring to our latest
release 2.1.2, while now I see from your error messages that you
are using 2.1.1.
I don’t think the new version 2.1.2 will naturally cure your StdHEP
compilation problem without patch; anyway I always recommend to use
the latest version. If you want to try that one, you should refer
to the patch I suggested in my latest email (but I would also try
without patch to see if the problem has naturally gone).

If you instead want to patch 2.1.1, you copy pythia6426.f in srcHerwig
and then, in the script MCatNLO_MadFKS_HERWIG6.Script, you add

HWUTI+=' pythia6426.o’

at line 153, just before

read -a HWUTIA <<< "$HWUTI”

Let me know whether the various solutions work.
Cheers.
Paolo

On 16 Jul 2014, at 01:42, Azar Mustafayev <email address hidden> wrote:

> Hi Paolo,
>
> I couldn't implement your patch, because there is no reference to
> jimmy.o in that script file as you described. I attached my script file.
> Should I add the line you described somewhere below line 155?
> Also, I have only pythia6426.f not pythia6428.f
>
> Thanks,
> Azar
>
> On 07/14/2014 08:21 PM, Paolo Torrielli wrote:
>> Hi Azar,
>>
>>> I tried to generate events with PYTHIA6Q as you suggested and it did go
>>> through returning cross esction of 8903 pb. This is close to 9.7e3 pb
>>> you have for Z+j at NLO in Table 1 of your paper. I guess the difference
>>> can be attributed different PDFs and scale used.
>> Ok, this seems consistent with the fact that, in the relevant files of
>> StdHEP, no internal Herwig6 routine is called, while this is the case
>> for Pythia6. Thus, while running Pythia6, StdHEP finds what it wants.
>> I still don’t understand why StdHEP misses the Pythia6 routines while
>> using Herwig6; this is a compiler/setup dependent fact that unfortunately
>> I can’t reproduce.
>> Anyway, as a very dirty patch to get Herwig6 running in your case,
>> (since I don’t know right now how to solve the issue properly), you
>> can do the following:
>> - copy the file MCatNLO/srcPythia/pythia6428.f to MCatNLO/srcHerwig.
>> - edit the file MCatNLO/Scripts/MCatNLO_MadFKS_HERWIG6.Script and
>> replace “ HWUTI+=' jimmy.o’ “ with “ HWUTI+=' jimmy.o pythia6428.o’ “
>> at line 145.
>> This links the Pythia6 source code also in the case of a Herwig6 run,
>> but still you will never use it during the shower, it is just to avoid
>> StdHEP complaining.
>> Let me know if the Herwig6 shower goes through in this case.
>>
>> As for the cross section, the exact setup employed in the paper is
>> specified right before the tables, and the difference you see it may
>> be due to scale and PDFs, as you say, or to cuts.
>> In order to change PDF, you just need to edit run_card.dat. To change
>> the functional form for reference scale, you edit the file SubProcesses/
>> setscales.f and in particular function scale_global_reference(pp), at
>> the bottom of that file. To change cuts, you edit run_card again, or
>> SubProcesses/cuts.f for non-standard cuts.
>>
>>> In running the code I received multiple messages of the following type,
>>> but I guess they are harmless:
>>> Exception AttributeError: AttributeError("'_DummyThread' object has no
>>> attribute '_Thread__block'",) in <module 'threading' from
>>> '/usr/lib64/python2.7/threading.pyc'> ignored
>> Yes, this is harmless.
>>
>> Cheers.
>> Paolo
>>
>>
>>
>>
>>> On 07/14/2014 12:02 AM, Paolo Torrielli wrote:
>>>> Hi Azar,
>>>>
>>>> could you also try to generate and shower a sample with PYTHIA6Q instead of HERWIG6, and see if you get the
>>>> same kind of problem?
>>>>
>>>> Another question: are you sure that while compiling StdHEP you use the same fortran and c++ compilers that you
>>>> use to compile HERWIG6 in the shower phase? Normally this is the case, unless you specify some particular compiler
>>>> in the input/mg5_configuration.txt file.
>>>> If you use different compilers, there may be some compilation conflict or something similar.
>>>>
>>>> Cheers.
>>>> Paolo
>>>>
>>>
>>> ** Attachment added: "mg5_configuration.txt"
>>> https://bugs.launchpad.net/bugs/1334014/+attachment/4152648/+files/mg5_configuration.txt
>>>
>>> --
>>> You received this bug notification because you are subscribed to the bug
>>> report.
>>> https://bugs.launchpad.net/bugs/1334014
>>>
>>> Title:
>>> crash at Compiling MCatNLO
>>>
>>> Status in MadGraph5_aMC@NLO Generator:
>>> New
>>>
>>> Bug description:
>>> Hello,
>>>
>>> I am trying to learn how to generate NLO events and decided to
>>> reproduce Z+j results from your paper, but MG crashed compiling
>>> MCatNLO. Here is the relevant screen output
>>>
>>> INFO: Events generated
>>> decay_events -from_cards
>>> INFO: Prepairing MCatNLO run
>>> INFO: Compiling MCatNLO for HERWIG6...
>>> ../lib/libstdhep.a(luntmp.o): In function `luntmp_':
>>> luntmp.F:(.text+0x82): undefined reference to `pyerrm_'
>>> luntmp.F:(.text+0x5bc): undefined reference to `pyerrm_'
>>> luntmp.F:(.text+0x61c): undefined reference to `pycomp_'
>>> luntmp.F:(.text+0xaa6): undefined reference to `pyerrm_'
>>> luntmp.F:(.text+0xee6): undefined reference to `pycomp_'
>>> luntmp.F:(.text+0xfee): undefined reference to `pyerrm_'
>>> ../lib/libstdhep.a(lutran.o): In function `lutran_':
>>> lutran.F:(.text+0x2a8): undefined reference to `pyerrm_'
>>> lutran.F:(.text+0x2e2): undefined reference to `pyerrm_'
>>> lutran.F:(.text+0x30d): undefined reference to `pyerrm_'
>>> lutran.F:(.text+0xaf4): undefined reference to `pyerrm_'
>>> collect2: error: ld returned 1 exit status
>>> make: *** [HW_EXE_DEFAULT] Error 1
>>> mv: cannot stat ‘HW_EXE_DEFAULT’: No such file or directory
>>>
>>> Then there are some lengthy fortran compiler lines and the program crashes suggesting to report the error.
>>> Could you tell what went wrong and how to fix this?
>>>
>>> Thanks,
>>> Azar
>>>
>>> To manage notifications about this bug go to:
>>> https://bugs.launchpad.net/mg5amcnlo/+bug/1334014/+subscriptions
>
>
> ** Attachment added: "MCatNLO_MadFKS_HERWIG6.Script"
> https://bugs.launchpad.net/bugs/1334014/+attachment/4153502/+files/MCatNLO_MadFKS_HERWIG6.Script
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1334014
>
> Title:
> crash at Compiling MCatNLO
>
> Status in MadGraph5_aMC@NLO Generator:
> New
>
> Bug description:
> Hello,
>
> I am trying to learn how to generate NLO events and decided to
> reproduce Z+j results from your paper, but MG crashed compiling
> MCatNLO. Here is the relevant screen output
>
> INFO: Events generated
> decay_events -from_cards
> INFO: Prepairing MCatNLO run
> INFO: Compiling MCatNLO for HERWIG6...
> ../lib/libstdhep.a(luntmp.o): In function `luntmp_':
> luntmp.F:(.text+0x82): undefined reference to `pyerrm_'
> luntmp.F:(.text+0x5bc): undefined reference to `pyerrm_'
> luntmp.F:(.text+0x61c): undefined reference to `pycomp_'
> luntmp.F:(.text+0xaa6): undefined reference to `pyerrm_'
> luntmp.F:(.text+0xee6): undefined reference to `pycomp_'
> luntmp.F:(.text+0xfee): undefined reference to `pyerrm_'
> ../lib/libstdhep.a(lutran.o): In function `lutran_':
> lutran.F:(.text+0x2a8): undefined reference to `pyerrm_'
> lutran.F:(.text+0x2e2): undefined reference to `pyerrm_'
> lutran.F:(.text+0x30d): undefined reference to `pyerrm_'
> lutran.F:(.text+0xaf4): undefined reference to `pyerrm_'
> collect2: error: ld returned 1 exit status
> make: *** [HW_EXE_DEFAULT] Error 1
> mv: cannot stat ‘HW_EXE_DEFAULT’: No such file or directory
>
> Then there are some lengthy fortran compiler lines and the program crashes suggesting to report the error.
> Could you tell what went wrong and how to fix this?
>
> Thanks,
> Azar
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/mg5amcnlo/+bug/1334014/+subscriptions

Hi Azar,

event generation and showering are independent, so when you get
a run summary, the LH event file has already been produced and
the cross section is reliable.

If you instead want to patch 2.1.1, you copy pythia6426.f in srcHerwig
and then, in the script MCatNLO_MadFKS_HERWIG6.Script, you add

HWUTI+=' pythia6426.o’

at line 153, just before

read -a HWUTIA <<< "$HWUTI”

Let me know whether the various solutions work.
Cheers.
Paolo

On 16 Jul 2014, at 01:42, Azar Mustafayev <azar@phys.hawaii.edu> wrote:

> Hi Paolo,
> 
> I couldn't implement your patch, because there is no reference to 
> jimmy.o in that script file as you described. I attached my script file. 
> Should I add the line you described somewhere below line 155?
> Also, I have only pythia6426.f not pythia6428.f
> 
> Thanks,
>   Azar
> 
> On 07/14/2014 08:21 PM, Paolo Torrielli wrote:
>> Hi Azar,
>> 
>>> I tried to generate events with PYTHIA6Q as you suggested and it did go
>>> through returning cross esction of 8903 pb. This is close to 9.7e3 pb
>>> you have for Z+j at NLO in Table 1 of your paper. I guess the difference
>>> can be attributed different PDFs and scale used.
>> Ok, this seems consistent with the fact that, in the relevant files of
>> StdHEP, no internal Herwig6 routine is called, while this is the case
>> for Pythia6. Thus, while running Pythia6, StdHEP finds what it wants.
>> I still don’t understand why StdHEP misses the Pythia6 routines while
>> using Herwig6; this is a compiler/setup dependent fact that unfortunately
>> I can’t reproduce.
>> Anyway, as a very dirty patch to get Herwig6 running in your case,
>> (since I don’t know right now how to solve the issue properly), you
>> can do the following:
>> - copy the file MCatNLO/srcPythia/pythia6428.f to MCatNLO/srcHerwig.
>> - edit the file MCatNLO/Scripts/MCatNLO_MadFKS_HERWIG6.Script and
>> replace “ HWUTI+=' jimmy.o’ “ with “ HWUTI+=' jimmy.o pythia6428.o’ “
>> at line 145.
>> This links the Pythia6 source code also in the case of a Herwig6 run,
>> but still you will never use it during the shower, it is just to avoid
>> StdHEP complaining.
>> Let me know if the Herwig6 shower goes through in this case.
>> 
>> As for the cross section, the exact setup employed in the paper is
>> specified right before the tables, and the difference you see it may
>> be due to scale and PDFs, as you say, or to cuts.
>> In order to change PDF, you just need to edit run_card.dat. To change
>> the functional form for reference scale, you edit the file SubProcesses/
>> setscales.f and in particular function scale_global_reference(pp), at
>> the bottom of that file. To change cuts, you edit run_card again, or
>> SubProcesses/cuts.f for non-standard cuts.
>> 
>>> In running the code I received multiple messages of the following type,
>>> but I guess they are harmless:
>>> Exception AttributeError: AttributeError("'_DummyThread' object has no
>>> attribute '_Thread__block'",) in <module 'threading' from
>>> '/usr/lib64/python2.7/threading.pyc'> ignored
>> Yes, this is harmless.
>> 
>> Cheers.
>> Paolo
>> 
>> 
>> 
>> 
>>> On 07/14/2014 12:02 AM, Paolo Torrielli wrote:
>>>> Hi Azar,
>>>> 
>>>> could you also try to generate and shower a sample with PYTHIA6Q instead of HERWIG6, and see if you get the
>>>> same kind of problem?
>>>> 
>>>> Another question: are you sure that while compiling StdHEP you use the same fortran and c++ compilers that you
>>>> use to compile HERWIG6 in the shower phase? Normally this is the case, unless you specify some particular compiler
>>>> in the input/mg5_configuration.txt file.
>>>> If you use different compilers, there may be some compilation conflict or something similar.
>>>> 
>>>> Cheers.
>>>> Paolo
>>>> 
>>> 
>>> ** Attachment added: "mg5_configuration.txt"
>>> https://bugs.launchpad.net/bugs/1334014/+attachment/4152648/+files/mg5_configuration.txt
>>> 
>>> -- 
>>> You received this bug notification because you are subscribed to the bug
>>> report.
>>> https://bugs.launchpad.net/bugs/1334014
>>> 
>>> Title:
>>> crash at Compiling MCatNLO
>>> 
>>> Status in MadGraph5_aMC@NLO Generator:
>>> New
>>> 
>>> Bug description:
>>> Hello,
>>> 
>>> I am trying to learn how to generate NLO events and decided to
>>> reproduce Z+j results from your paper, but MG crashed compiling
>>> MCatNLO. Here is the relevant screen output
>>> 
>>> INFO: Events generated
>>> decay_events -from_cards
>>> INFO: Prepairing MCatNLO run
>>> INFO: Compiling MCatNLO for HERWIG6...
>>> ../lib/libstdhep.a(luntmp.o): In function `luntmp_':
>>> luntmp.F:(.text+0x82): undefined reference to `pyerrm_'
>>> luntmp.F:(.text+0x5bc): undefined reference to `pyerrm_'
>>> luntmp.F:(.text+0x61c): undefined reference to `pycomp_'
>>> luntmp.F:(.text+0xaa6): undefined reference to `pyerrm_'
>>> luntmp.F:(.text+0xee6): undefined reference to `pycomp_'
>>> luntmp.F:(.text+0xfee): undefined reference to `pyerrm_'
>>> ../lib/libstdhep.a(lutran.o): In function `lutran_':
>>> lutran.F:(.text+0x2a8): undefined reference to `pyerrm_'
>>> lutran.F:(.text+0x2e2): undefined reference to `pyerrm_'
>>> lutran.F:(.text+0x30d): undefined reference to `pyerrm_'
>>> lutran.F:(.text+0xaf4): undefined reference to `pyerrm_'
>>> collect2: error: ld returned 1 exit status
>>> make: *** [HW_EXE_DEFAULT] Error 1
>>> mv: cannot stat ‘HW_EXE_DEFAULT’: No such file or directory
>>> 
>>> Then there are some lengthy fortran compiler lines and the program crashes  suggesting to report the error.
>>> Could you tell what went wrong and how to fix this?
>>> 
>>> Thanks,
>>> Azar
>>> 
>>> To manage notifications about this bug go to:
>>> https://bugs.launchpad.net/mg5amcnlo/+bug/1334014/+subscriptions
> 
> 
> ** Attachment added: "MCatNLO_MadFKS_HERWIG6.Script"
>  https://bugs.launchpad.net/bugs/1334014/+attachment/4153502/+files/MCatNLO_MadFKS_HERWIG6.Script
> 
> -- 
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1334014
> 
> Title:
>  crash at Compiling MCatNLO
> 
> Status in MadGraph5_aMC@NLO Generator:
>  New
> 
> Bug description:
>  Hello,
> 
>  I am trying to learn how to generate NLO events and decided to
>  reproduce Z+j results from your paper, but MG crashed compiling
>  MCatNLO. Here is the relevant screen output
> 
>  INFO: Events generated 
>  decay_events -from_cards
>  INFO: Prepairing MCatNLO run 
>  INFO: Compiling MCatNLO for HERWIG6... 
>  ../lib/libstdhep.a(luntmp.o): In function `luntmp_':
>  luntmp.F:(.text+0x82): undefined reference to `pyerrm_'
>  luntmp.F:(.text+0x5bc): undefined reference to `pyerrm_'
>  luntmp.F:(.text+0x61c): undefined reference to `pycomp_'
>  luntmp.F:(.text+0xaa6): undefined reference to `pyerrm_'
>  luntmp.F:(.text+0xee6): undefined reference to `pycomp_'
>  luntmp.F:(.text+0xfee): undefined reference to `pyerrm_'
>  ../lib/libstdhep.a(lutran.o): In function `lutran_':
>  lutran.F:(.text+0x2a8): undefined reference to `pyerrm_'
>  lutran.F:(.text+0x2e2): undefined reference to `pyerrm_'
>  lutran.F:(.text+0x30d): undefined reference to `pyerrm_'
>  lutran.F:(.text+0xaf4): undefined reference to `pyerrm_'
>  collect2: error: ld returned 1 exit status
>  make: *** [HW_EXE_DEFAULT] Error 1
>  mv: cannot stat ‘HW_EXE_DEFAULT’: No such file or directory
> 
>  Then there are some lengthy fortran compiler lines and the program crashes  suggesting to report the error.
>  Could you tell what went wrong and how to fix this?
> 
>  Thanks,
>    Azar
> 
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/mg5amcnlo/+bug/1334014/+subscriptions