segfault while compiling g g > g g g g g

Asked by Wojciech Kotlarski

While trying to compile a code for process g g > g g g g g gfortran quits with segmentation fault. This is always reproducible and was checked on two different computers with different versions of Linux (Fedora and Arch) and on a cluster. This concerns only 7-point gluon amplitude. Processes of a type g g > g g g q q compile without a problem. This problem persist from version 1.3.13 but I haven't checked earlier versions.

Question information

English Edit question
MadGraph5_aMC@NLO Edit question
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said :

I don't think that they are any `real` problem whith the output, simply the 7 gluon amplitude output file
are just too large that it's not possible to compile it.

One of the main problem is the size of the color matrix which is much smaller for g g > g g g q q.
Concerning your process, I would advice you to look at the following talk:
The author modifies the output of MG5 in order to be able to compile the output and to compute it.
I suggest you to contact him directly.

They are also work in order to be able to treat efficiently this type of process inside MG5 but that's work under progress.



PS: I consider this more as a question than a real bug. So I'll convert it to a question.

Revision history for this message
Valentin Hirschi (valentin-hirschi) said :

I managed to compile an equivalent of 7-8 gluon amplitude from MG5 simply by splitting the HELAS and JAMP calls and putting them in dedicated subroutines specified in separate files.
The original code was 25 MB all together and I have split it in pieces of less than 500 Ko each. The remaining matrix.f should only contain the color matrix and weight no more than 1MB or so.
Finally make sure that you don't have more than 2-3 line breaks for each JAMP definition line, as this drastically affects compiling time and memory.



Can you help with this problem?

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

To post a message you must log in.