multi run with cluster

Hi,

I just realized that I was using an old version of MG, i.e., MG5_v1_5_13. I updated to MG5_aMC and realized that there seems to be something wrong with the pgs simulator. I'm doing now the exact same process as described below to generate one combined LHE file using multi_run (has 20000 events). I then try to act on these events with pythia and pgs using ./bin/madevent pythia run_01 and get the following errors

INFO: Running PGS
WARNING: mcfio_NextEvent: Error Repositioning stream

WARNING: mcfio_NextEvent: Error Repositioning stream

WARNING: mcfio_NextEvent: Error Repositioning stream

WARNING: mcfio_NextEvent: Error Repositioning stream

WARNING: mcfio_NextEvent: Error Repositioning stream

An LHCO file is actually produced but strangely, only has 99 events. I set the run_card.dat to do matching so I expect a smaller amount of events but this is a drop from 20000 events. Moreover, I've tried this a few different times (even for a couple different processes) and I always get 99 events in the tag_1_pgs_events.lhco.gz file. Have you seen this before or happen to know what the problem is?

Peter

On 2014-04-09, at 1:31 PM, Peter Winslow <email address hidden> wrote:

> New question #246760 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/246760
>
> Hi,
>
> I'm trying to generate larger statistics for background processes using the multi_run cmd. I also would like to run this on my local cluster. Here's what I've tried:
>
> - Generate an output file for a very simple process (p p > t t~) using the interface. Before launching the process, I exit the interface.
>
> - modify the input/mg5_configuration.txt file, making the following changes: 1) automatic_html_opening = False, 2) run_mode = 1, 3) cluster_type = pbs, 4) cluster_queue = None. I also specify all Card files at this point.
>
> - cd into the output file and use the cmd "$./bin/madevent multi_run 2". Two subsequent runs are then performed, they finish fine, and the LHE files are combined in the Events folder to create one LHE file with a single event weight and a larger number of events.
>
> My question is:
>
> - Although multi_run seems to be working quite well, it's not clear to me that it is actually using the cluster. When I start running, i.e., enter "$./bin/madevent multi_run 2", I then open a separate terminal window and run "qstat" to see the current jobs. When I do this, I don't see anything. Is there a way to see whether it has actually made use of the cluster or not?
>
> Thanks very much,
> Peter
>
> --
> You received this question notification because you asked the question.

Hi Peter,

Your configuration seems correct.

The cluster is not used for all the operation, the compilation is done locally.
but when you see the line like
INFO: Idle: 122, Running: 22, Completed: 0 [ current time: 22h32 ]
Then we are using the cluster and you can check it via the qstat.
If you don’t see anything, you can use the top command to check.

Cheers,

Olivier

On Apr 9, 2014, at 6:31 PM, Peter Winslow <email address hidden> wrote:

> New question #246760 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/246760
>
> Hi,
>
> I'm trying to generate larger statistics for background processes using the multi_run cmd. I also would like to run this on my local cluster. Here's what I've tried:
>
> - Generate an output file for a very simple process (p p > t t~) using the interface. Before launching the process, I exit the interface.
>
> - modify the input/mg5_configuration.txt file, making the following changes: 1) automatic_html_opening = False, 2) run_mode = 1, 3) cluster_type = pbs, 4) cluster_queue = None. I also specify all Card files at this point.
>
> - cd into the output file and use the cmd "$./bin/madevent multi_run 2". Two subsequent runs are then performed, they finish fine, and the LHE files are combined in the Events folder to create one LHE file with a single event weight and a larger number of events.
>
> My question is:
>
> - Although multi_run seems to be working quite well, it's not clear to me that it is actually using the cluster. When I start running, i.e., enter "$./bin/madevent multi_run 2", I then open a separate terminal window and run "qstat" to see the current jobs. When I do this, I don't see anything. Is there a way to see whether it has actually made use of the cluster or not?
>
> Thanks very much,
> Peter
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.