segfault with syscalc

[marco zaro]

Hi,
I am doing the following:

import model heft
generate a a > h
output

then launch and edit the runcard changing:
     lhapdf = pdlabel ! PDF set
     244600 = lhaid ! if pdlabel=lhapdf, this is the lhapdf number
and
NNPDF23_nlo_as_0118_qed.LHgrid = sys_pdf

Events are correctly generated, but syscalc fails.

> SysCalc Failed. Please read the associate log to see the reason. Did you install the associate PDF set?

Looking into the log (Events/run_01/tag_1_parton_syscalc.log) I have

Init PDF set NNPDF23_nlo_as_0118_qed.LHgrid
Using 101 members for this set
read up to orgpdf
Set beam info: 1 1
Set original PDF = 244600 with member 0
Init original PDF 244600

If I run syscalc by hand
../SysCalc/sys_calc Events/run_01/events.lhe bin/internal/syscalc_card.dat outfile.txt

I get a segmentation fault error (see below)

Cheers,

Marco

Init PDF set NNPDF23_nlo_as_0118_qed.LHgrid
 *************************************
 * LHAPDF Version 5.9.1 *
 * Configured for the following: *
 * All PDFs *
 * FULL MEMORY option *
 * Maximum 3 concurrent set(s) *
 *************************************

 >>>>>> PDF description: <<<<<<
 NNPDF23_nlo_as_0118_qed.LHgrid
 NNPDF2.3QED NLO, arxiv:1308.0598
 NLO QCD + LO QED PDF set, alphas(MZ)=0.118
 NNPDF Collaboration: R.D. Ball, V. Bertone,
 S. Carrazza, L. Del Debbio, S. Forte,
 A. Guffanti, N.P. Hartland, J. Rojo
 This set has 100 member PDF
 mem=0 --> average on replicas.
 Evolution: pol. interpolation on the LH grid.
 >>>>>> <<<<<<

Using 101 members for this set
read up to orgpdf
Set beam info: 1 1
Set original PDF = 244600 with member 0
Init original PDF 244600
 *************************************
 * LHAPDF Version 5.9.1 *
 * Configured for the following: *
 * All PDFs *
 * FULL MEMORY option *
 * Maximum 3 concurrent set(s) *
 *************************************

 >>>>>> PDF description: <<<<<<
 NNPDF23_nlo_as_0118_qed.LHgrid
 NNPDF2.3QED NLO, arxiv:1308.0598
 NLO QCD + LO QED PDF set, alphas(MZ)=0.118
 NNPDF Collaboration: R.D. Ball, V. Bertone,
 S. Carrazza, L. Del Debbio, S. Forte,
 A. Guffanti, N.P. Hartland, J. Rojo
 This set has 100 member PDF
 mem=0 --> average on replicas.
 Evolution: pol. interpolation on the LH grid.
 >>>>>> <<<<<<

Segmentation fault: 11

[Alexis Kalogeropoulos (alkaloge)]

Hello

I cannot reproduce your problem - running in MG5_aMC_v2_3_0 and LHAPDF_6.1.5 it works ok for me- Can you maybe try switching to a more recent LHAPDF version ?

[marco zaro (marco-zaro)]

Hi Alexis,
Thanks,
so I have tried with LHAPDF6 (btw, the SysCalc makefile uses the lhapdf-config which is in the PATH, not the one which is specified via the configuration options)
I get NaNs…
Finished parsing 10000 events.
scale fact cross-section :0 nan
scale fact cross-section :1 nan
scale fact cross-section :2 nan
scale fact cross-section :3 nan
scale fact cross-section :4 nan
scale fact cross-section :5 nan
scale fact cross-section :6 nan
scale fact cross-section :7 nan
scale fact cross-section :8 nan
pdf reweighted cross-section :0 nan
pdf reweighted cross-section :1 nan
…
Cheers,

Marco
On 21 Jan 2016, at 11:52, Alexis Kalogeropoulos <email address hidden> wrote:

> Hello
>
> I cannot reproduce your problem - running in MG5_aMC_v2_3_0 and
> LHAPDF_6.1.5 it works ok for me- Can you maybe try switching to a more
> recent LHAPDF version ?
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1536557
>
> Title:
> segfault with syscalc
>
> Status in MadGraph5_aMC@NLO:
> New
>
> Bug description:
> Hi,
> I am doing the following:
>
> import model heft
> generate a a > h
> output
>
> then launch and edit the runcard changing:
> lhapdf = pdlabel ! PDF set
> 244600 = lhaid ! if pdlabel=lhapdf, this is the lhapdf number
> and
> NNPDF23_nlo_as_0118_qed.LHgrid = sys_pdf
>
> Events are correctly generated, but syscalc fails.
>
>> SysCalc Failed. Please read the associate log to see the reason. Did
> you install the associate PDF set?
>
> Looking into the log (Events/run_01/tag_1_parton_syscalc.log) I have
>
> Init PDF set NNPDF23_nlo_as_0118_qed.LHgrid
> Using 101 members for this set
> read up to orgpdf
> Set beam info: 1 1
> Set original PDF = 244600 with member 0
> Init original PDF 244600
>
> If I run syscalc by hand
> ../SysCalc/sys_calc Events/run_01/events.lhe bin/internal/syscalc_card.dat outfile.txt
>
> I get a segmentation fault error (see below)
>
> Cheers,
>
> Marco
>
> Init PDF set NNPDF23_nlo_as_0118_qed.LHgrid
> *************************************
> * LHAPDF Version 5.9.1 *
> * Configured for the following: *
> * All PDFs *
> * FULL MEMORY option *
> * Maximum 3 concurrent set(s) *
> *************************************
>
>>>>>>> PDF description: <<<<<<
> NNPDF23_nlo_as_0118_qed.LHgrid
> NNPDF2.3QED NLO, arxiv:1308.0598
> NLO QCD + LO QED PDF set, alphas(MZ)=0.118
> NNPDF Collaboration: R.D. Ball, V. Bertone,
> S. Carrazza, L. Del Debbio, S. Forte,
> A. Guffanti, N.P. Hartland, J. Rojo
> This set has 100 member PDF
> mem=0 --> average on replicas.
> Evolution: pol. interpolation on the LH grid.
>>>>>>> <<<<<<
>
> Using 101 members for this set
> read up to orgpdf
> Set beam info: 1 1
> Set original PDF = 244600 with me...

[Alexis Kalogeropoulos (alkaloge)]

Hi Marco

Meanwhile, I also saw this - trying on a pp > t t~ with the same lhapdf and sys_pdf however works...Need to investigate further...
@Olivier - can you also have a look ? I am suspecting this is not related to SysCalc (beam_pdfs are initialized properly anyway)

Thanks for reporting it.

Regards

Alexis

[Olivier Mattelaer (olivier-mattelaer)]

Hi,

This is just that SysCalc was not supporting photon PDF.

In attachement, you will find the associate patch to add this feature.

Cheers,

Olivier