Problem with u u > w+ w+ d d, wrong number of unweighted events !

Hi,

This is actually an old problem linked to an efficiency to generate events in the case of VBF process for extremely back to back jet.
Surprisingly this problem seems to be strongly reduces with the latest version of the code.
Since I was not expecting any improvement on this topic, I’m currently investigating why such improvement appears.

Cheers,

Olivier

On Sep 29, 2014, at 4:04 PM, D Li <email address hidden> wrote:

> Question #255057 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/255057
>
> Description changed to:
> Hello!
>
> I am running the aforementioned process within the standard model, at
> tree level, with default settings of all parameters. I narrowed the
> problem down and find that it arises from different settings for the
> order of couplings.
>
> This process have two different types of leading order, the first is
> QED=4, QCD=0, corresponding to e.g. quarks exchanging a photon and then
> each emit a W+. The second is QED=2, QCD=2, corresponding to e.g. quarks
> exchanging a gluon and then each emit a W boson.
>
> To get the full leading order result, I run: u u > w+ w+ d d QED=4, asking for 5000 unweighted events.
> I get 102 diagrams, and the output is:
> Cross-section : 0.6239 +- 0.001531 pb
> Nb of events : 807
>
> The number of events is 807, which is problematic. It seems like people
> had similiar problem due to IR divergences. But this should not be the
> case here, since all final state particles are massive.
>
> Then I run: u u > w+ w+ d d QED=4 QCD=0, asking also for 5000 unweighted events.
> I get 86 diagrams, and:
> Cross-section : 0.3286 +- 0.0009676 pb
> Nb of events : 5000
>
> Which seems to be correct.
>
> Then I run: u u > w+ w+ d d QED=2 QCD=2, asking also for 5000 unweighted events.
> I get 16 diagrams, and:
> Cross-section : 0.2348 +- 0.0006297 pb
> Nb of events : 5000
>
> Which seems to be correct as well.
>
>
> The only difference between these three are settings for max order of coupling constants. The first process should be the combination of the last 2 processes. There may be interference effects that are not captured by the seperate calculation. But I am having a hard time seeing why there is a problem with the combined calculation.
>
> But in any case, would these unweighted events I generated be OK to use
> even when the number is smaller?
>
>
> -------------------------------------------------------Update 09/29
> ---------------------------------------------------------
>
> Today I find that whether this problem show up depends on the requested number of events.
> For example, for the QED=4 QCD=0 case, if I ask for 100K events, it will give:
>
> Cross-section : 0.7304 +- 0.0005239 pb
> Nb of events : 41237
>
> , where the number of events generated is smaller than the requested
> number. As noted above, this did not happen when I asked for 5000
> events.
>
>
>
> Thank you!!
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hi Olivier:

Thank you!

Do you think the underweight events that do get generated are correct?

Best wishes,
Li

Sorry I mean unweighted.

If they are correct and usable, is there a way to set the parameters such that this problem is alleviated?

Many Thanks!

Best wishes,
Daliang

Hi,

Yes those event should be usable.

I’m still trying to understand why the version 2.1.2 actually improves the situation compare to the previous version.

Cheers,

Olivier

On Sep 29, 2014, at 4:34 PM, D Li <email address hidden> wrote:

> Question #255057 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/255057
>
> Status: Answered => Open
>
> D Li is still having a problem:
> Hi Olivier:
>
> Thank you!
>
> Do you think the underweight events that do get generated are correct?
>
> Best wishes,
> Li
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thank you! That helps!