EFT reweighting in top decays

Asked by Kirill Skovpen on 2020-09-07

Hello,

We are trying to run reweight for EFT (dim6top_LO_UFO) in top decays with automatic recalculation of the top quark decay width. We would like to do it in the gridpack mode for MC production at CMS, and therefore would prefer to avoid including fortran source files.

Our understanding is that this line in the reweight card:

set width 6 auto

would allow us to do it, but through a recompilation process for some parts of the MG code. Is there any other way of achieving this without forcing recompilation?

We thought in the direction of possibly extending rwgt.pkl to also include precomputed widths, but this may require some modifications to the source code (if possible at all). Any feedback would be very much appreciated!

Thanks a lot,

Alexander Grohsjean, Kirill Skovpen

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2020-09-10
Last query:
2020-09-10
Last reply:
2020-09-07

Hi,

I have check the model and then do not include the "decays.py" file in their file.
So indeed even for two body decay we will need to write a dedicated code and compile it
(if the model was including that file, we would need to do that only for three body decay --if the code thinks that this is needed).

It is not possible to prevent the auto-width to write code (and therefore compile it) since the code is taylored to the benchmark point so they are no guarantee that you can reuse the code from one benchmark to the next.

The advantage of the "auto" mode is not clear to me since it is kind of a waste of ressource to recompute such width at each run of the gridpack.
Consequently, the solution should be "simple" is to compute those width in advance (potentially with the madwidth tool as well or simply via the test run).

Cheers,

Olivier

> On 7 Sep 2020, at 13:15, Kirill Skovpen <email address hidden> wrote:
>
> New question #692793 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/692793
>
> Hello,
>
> We are trying to run reweight for EFT (dim6top_LO_UFO) in top decays with automatic recalculation of the top quark decay width. We would like to do it in the gridpack mode for MC production at CMS, and therefore would prefer to avoid including fortran source files.
>
> Our understanding is that this line in the reweight card:
>
> set width 6 auto
>
> would allow us to do it, but through a recompilation process for some parts of the MG code. Is there any other way of achieving this without forcing recompilation?
>
> We thought in the direction of possibly extending rwgt.pkl to also include precomputed widths, but this may require some modifications to the source code (if possible at all). Any feedback would be very much appreciated!
>
> Thanks a lot,
>
> Alexander Grohsjean, Kirill Skovpen
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hello Olivier,

Many thanks for the quick reply!

On a technical implementation within MG, would you have any specific recommendation on how to efficiently update the reweight card, so that auto is turned into the value at the given EFT point?

Thanks,

Alexander, Kirill

We do not have anything really smart for that.

my suggestion would be to do the folowing
generate h > b b~
output

and then do in the created directory: be sure that you have the correct param_card as starting point)
./bin/madevent PATH_TO_REWEIGHT_CARD_WITH_AUTO

This will trigger all the required computation and within each banner you will get the associate width.
The boring part is then the parsing of those banner and the re-writting of the reweight_card obviously.

> On 7 Sep 2020, at 15:41, Kirill Skovpen <email address hidden> wrote:
>
> Question #692793 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/692793
>
> Status: Answered => Open
>
> Kirill Skovpen is still having a problem:
> Hello Olivier,
>
> Many thanks for the quick reply!
>
> On a technical implementation within MG, would you have any specific
> recommendation on how to efficiently update the reweight card, so that
> auto is turned into the value at the given EFT point?
>
> Thanks,
>
> Alexander, Kirill
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thanks Olivier Mattelaer, that solved my question.