Error while compiling subprocesses

According to the debug file, this is due to a crash of the compiler itself.

internal compiler error: Segmentation fault
    Please submit a full bug report,
    with preprocessed source if appropriate.
    See <file:///usr/share/doc/gcc-9/README.Bugs> for instructions.
    make: *** [makefile:60: auto_dsig.o] Error 1
    make: *** Waiting for unfinished jobs....

So my suggestion would be to first use another version of gcc.

Now note that MG5aMC 2.6.7 is outdated and likely does not support GCC10 (or GCC11).
So you might also need to update MG5aMC to a more recent version (I strongly suggest to update to at least 2.9.x which correspond to our long term stable version).

If you want to stick to 2.6.7, I would then advise to try with an update of gcc-9.

Cheers,

Olivier

Thanks a lot Olivier for the prompt response.
Yes, I actually tried using v2.9.7 even v3.3.1. But setting width to 'AUTO' I was getting error (in 2.9.7) as was already reported in https://bugs.launchpad.net/mg5amcnlo/+bug/1946657

But in this version it worked. Can you tell me if that bug-issue is resolved?

Although, here I later changed a parameter to get higher widths, and in that case it actually ran successfully complete a run and in subsequent runs with lower widths it successfully ran.

Thanks a lot again, Olivier.

Saumyen

Will try to move forward on that bug which is indeed more than overdue.

Independently of that, I will likely not trust the result of 2.6.7 (for the width computation).
I suspect that it give the wrong result (and therefore I do prefer the actual version which at least crash)

Cheers,

Olivier

PS: I do not think that ANY version of the auto-width can be trusted here.

> On 10 Feb 2022, at 16:01, Saumyen Kundu <email address hidden> wrote:
>
> Question #700572 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/700572
>
> Status: Answered => Solved
>
> Saumyen Kundu confirmed that the question is solved:
> Thanks a lot Olivier for the prompt response.
> Yes, I actually tried using v2.9.7 even v3.3.1. But setting width to 'AUTO' I was getting error (in 2.9.7) as was already reported in https://bugs.launchpad.net/mg5amcnlo/+bug/1946657
>
> But in this version it worked. Can you tell me if that bug-issue is
> resolved?
>
> Although, here I later changed a parameter to get higher widths, and in
> that case it actually ran successfully complete a run and in subsequent
> runs with lower widths it successfully ran.
>
> Thanks a lot again, Olivier.
>
> Saumyen
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

As you might have seen, I have resolved the issue yesterday (not yet released but will try to do a new release of the Long term support version today or next week).

As said before, I do not think that the auto-width worked as expected for that model in ANY previous version of the code.
The version that introduced the bug mentioned above was actually fixing an issue that not all Feynman Diagram was included in some situation (which was the case for your model)/

Cheers,

Olivier

Thank you so much Olivier for letting me know of the fix.

I modified the two files by replacing the red lines with the green one. So far, I haven't faced with any errors.
Then I relaunch an earlier process which ran well. That process was only at the production level, i.e., I didn't decay the heavy particles. But, when I generated the full process in the first run I got a 'Segmentation fault':
,,
INFO: P1_ssx_chi2chi2x_chi2_chi1z_z_ddx_chi2x_chi1xz_z_uux
INFO: P1_ssx_chi2chi2x_chi2_chi1z_z_ddx_chi2x_chi1xz_z_ccx
Segmentation fault (core dumped)
"
when I again ran it, "Error detected in "generate_events yc7"". The debug.log file is here,
https://www.dropbox.com/s/o1vqn2uluwhgbmj/yc7_tag_1_debug.log?dl=0

Regards,
Saumyen

Which version of gcc9 did you use? Is this a pre-release or something like that?
The error also claims that the bug is within GCC9 and not within our code.

Now given the compiled file where this does occurs,
one solution might be to
replace the output command by
output MYDIR -- hel_recycling = False
(replace MYDIR obviously)

You can also keep that command intact and set within the run_card the parameter hel_recycling on False (the first option is slightly more optimal but the difference is very small).

In both case, setting such parameter will likely lead to a slower code (see 2102.00773) but this should avoid your GCC9 specific issue.

Cheers,

Olivier

Hi Olivier,
I am using gcc-9.3.0.
"gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0"

So, since you iterated that MG5 code is fine and this is related to gcc9, I updated the the compilers (gcc, g++, gfortran) and later the whole system too. Only thing that I noticed is that for g++, it said '1 not fully installed or removed' as in:
,,
0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.
1 not fully installed or removed.
After this operation, 0 B of additional disk space will be used.
''
After these updates, I re-ran the same script and it ran successfully, no error or warning (except for small width of course). So I guess this is working fine. Although don't know what was wrong actually. :P

Just that, after the completion of the run I got this,
,,
INFO: P1_ssx_chi2chi2x_chi2_chi1z_z_ddx_chi2x_chi1xz_z_uux
Command "INFO:" not recognized, please try again
INFO: P1_ssx_chi2chi2x_chi2_chi1z_z_ddx_chi2x_chi1xz_z_ccx
Command "INFO:" not recognized, please try again
Segmentation fault (core dumped)
Command "Segmentation" not recognized, please try again
quit
"

I think this is nothing (to be concerned with), right?

I'll let you know if anything peculiar happens.

Thanks a lot Olivier.

Regards,
Saumyen