Failing to generate number of requested events (fails in 2.4.3 works in 2.6.0)

Asked by Lukas Heinrich on 2017-11-01

Hello,

we're trying to generate a BSM model for which we have received a UFO from theorists. It's a Dark Higgs model and the process we are trying to generate is the following

import model DarkHiggs2MDM
generate p p > zp > n1 n1 hs QED<=2, (hs > b b~)
add process p p > zp > n1 n1 hs j QED<=2, (hs > b b~)

The model is attached [2]

I tried in standalone installations (with run/param/proc cards in [1]) of both Madgraph 2.4.3 and 2.6.0 and interestingly the job succeeds in generating all events in 2.6.0 while failing in 2.4.3. The calculated cross-section is the same in both versions.

Fiddling with the cuts in the run card and toggling cut_decays seems to have some impact and has helped a little (increasing the generated events from a few hundred to a few thousand) but I haven't been able to find a setting that can reliably generate all requested events.

Cheers,
Lukas

[1] https://drive.google.com/open?id=0B0SXa0qgQrygYmRaNXUzNXY5d25sOUl5bmFMcHg0OWE2dTFZ
[2] https://drive.google.com/open?id=0B0SXa0qgQrygaWpOZWozS2FjNWlUeFBndmhpZUxkTUlieVVN

PS: My standalone installations are available on Docker Hub:

lukasheinrich/genericmg5pydelphes-base:mg243
lukasheinrich/genericmg5pydelphes-base:mg260

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Last query:
2017-11-01
Last reply:
2017-11-01

Hi Lukas,

I'm glad to hear that we have fixed your problem already.

Thanks for your feedback,

Olivier

Lukas Heinrich (lukasheinrich) said : #2

Hi Olivier,

thanks. Since this is for productio in ATLAS where we have some constraints on what Madgraph versions we can use, do you have a workaround for 2.4.3 ? (I'm checking in parallel whether we can use 2.6 in our ATLAS configuration)

Thanks,
Lukas

Hi Lukas,

The only work around that I can propose to you is to run multiple smaller run and combien the events (in a weighted way).
I do not know think that this can be fix with a work around without changing anything to the code.

Cheers,

Olivier

> On Nov 1, 2017, at 15:34, Lukas Heinrich <email address hidden> wrote:
>
> Question #660213 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/660213
>
> Status: Answered => Open
>
> Lukas Heinrich is still having a problem:
> Hi Olivier,
>
> thanks. Since this is for productio in ATLAS where we have some
> constraints on what Madgraph versions we can use, do you have a
> workaround for 2.4.3 ? (I'm checking in parallel whether we can use 2.6
> in our ATLAS configuration)
>
> Thanks,
> Lukas
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Lukas Heinrich (lukasheinrich) said : #4

Thanks. Just for my understanding what's the problem in 2.4.3 and what change fixed it in 2.6.0?

Hi,

I do not know for sure.

My guess is that this is related to some optimization that I have in order to improve the minimum energy parameter inside the code, but this can also be something else.
The only way to be sure is to go trough each major version and run your code to see at which major version this improvement was done,
Then you have to go inside the development branch of that major version and see which dev update improve such situation (and potentially go to a parallel branch if needed to track this). So this is likely to be a 48h job for tracking this.

I certainly do not consider this worth my time.

Cheers,

Olivier

> On Nov 1, 2017, at 16:03, Lukas Heinrich <email address hidden> wrote:
>
> Question #660213 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/660213
>
> Lukas Heinrich posted a new comment:
> Thanks. Just for my understanding what's the problem in 2.4.3 and what
> change fixed it in 2.6.0?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Can you help with this problem?

Provide an answer of your own, or ask Lukas Heinrich for more information if necessary.

To post a message you must log in.