MadGraph5_aMC@NLO

running NLO process with -o flag crashes, missing file all logs_0.html

Bug #1947631 reported by William Shepherd on 2021-10-18

This bug affects 1 person

Affects		Status	Importance	Assigned to	Milestone
	MadGraph5_aMC@NLO	Fix Released	High	marco zaro

Bug Description

We are trying to test the "generate only" functionality for an NLO process.

The process is simple: generate p p > z [ QCD ].

The initial event generation runs fine:
./bin/generate_events

The follow-up attempt at reusing the grids fails:
./bin/generate_events -o

The reported error is:
FileNotFoundError : [Errno 2] No such file or directory: '/storage/performance/jww004/MG5_aMC_v3_2_0/NLO_Z/Events/run_02/alllogs_0.html'

The log file with error information is attached.

The only content of the second Events folder (run_02) is:
alllogs_2.html run_02_tag_1_banner.txt

Thanks very much --

Revision history for this message

William Shepherd (will-shepherd) wrote on 2021-10-18:

run_02_tag_1_debug.log Edit (21.7 KiB, text/plain)

Revision history for this message

Olivier Mattelaer (olivier-mattelaer) wrote on 2021-10-18:

Hi Rikkert,

I have assigned this to you (partly because the next two weeks will be crazy busy for me).
but if you think that I should handle this do not hesitate to re-assign it to me.

Olivier

Changed in mg5amcnlo:
assignee:	nobody → Rikkert Frederix (frederix)

Revision history for this message

Rikkert Frederix (frederix) wrote on 2021-10-19:

Hi Marco,

I traced it into "def write_orders_tag_info(self):" function of amcatnlo_run_interface.py. This function needs fixing. It seems completely wrong to use a log file to infer information that's needed when running the code. You cannot assume that any of the 'alllogs_*.html' exist. I think this function needs to be rewritten. Easiest would be to directly write it in python when generating the process, no?

Rikkert

Changed in mg5amcnlo:
importance:	Undecided → High
assignee:	Rikkert Frederix (frederix) → marco zaro (marco-zaro)

Rikkert Frederix (frederix) on 2021-10-19

Changed in mg5amcnlo:
status:	New → Confirmed

Revision history for this message

marco zaro (marco-zaro) wrote on 2021-10-20:

Hi everybody.
A quick fix is pushed in 3.2.1 rev 974. The information is written only once, but still sourcing it from the logs. I will have a look to do a proper fix, as Rik suggests writing such a file from the python exporter
Cheers,

Marco

Revision history for this message

marco zaro (marco-zaro) wrote on 2021-12-03:

Hi,
a proper fix should be this one

https://bazaar.launchpad.net/~maddevelopers/mg5amcnlo/3.3.1/revision/981

Do you mind to have a look?

Thanks!

Cheers,

Marco

Revision history for this message

Olivier Mattelaer (olivier-mattelaer) wrote on 2021-12-03:

Thanks Marco,

Sounds good to me.
I'm running all the tests of the suite for the moment, then I will run specific test associated to the instruction of the bug and we should be good to go with this.

Thanks,

Olivier