wrong number of events

Hi,

Indeed one bug fix in 2.2.2 has a side effect to reduce the efficiency of event generation in that process.
We are testing some phase-space improvement for 2.3.0 and I hope it can help for that efficiency but I can not promise
anything.

Cheers,

Olivier

On Nov 8, 2014, at 1:21 AM, Ramkrishna <email address hidden> wrote:

> New question #257133 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/257133
>
> I generated events with pp -> w+w- qq. And I requested total number of 10000 events at 14 TeV but I got only 621 events.
>
> Also I searched and found a link (https://answers.launchpad.net/mg5amcnlo/+question/255057) and here I got that this is the problem with Madgraph that we got low number of events. But there it is also written that it should improve with latest version. So, I installed latest version 2.2.2 but it seems that the inefficiency increased instead of decreasing.
>
> Any help will be highly appreciated.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Thanks Olivier,

But can I believe on the generated 621 events?

Because since it does not generated my required no of events due to some bug so I think it is possible that there happened something wrong with the generated events.

Hi,

Yes you can trust those events.

Cheers,

Olivier
On Nov 8, 2014, at 10:21 AM, Ramkrishna <email address hidden> wrote:

> Question #257133 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/257133
>
> Status: Answered => Open
>
> Ramkrishna is still having a problem:
> Thanks Olivier,
>
> But can I believe on the generated 621 events?
>
> Because since it does not generated my required no of events due to some
> bug so I think it is possible that there happened something wrong with
> the generated events.
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

One more thing do you have idea why this is happening?

Yes,

Our phase-space integrator is not optimized for generating particles with large eta.
Therefore the events generated in that phase-space area have very different weight, which slows down the code (slower convergence of the integral due to a large variance of the function)
and reduce significantly the efficiency of the un-weighting (due to a very large upper-bound).

Cheers,

Olivier

On Nov 10, 2014, at 10:36 AM, Ramkrishna <email address hidden> wrote:

> Question #257133 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/257133
>
> Status: Answered => Open
>
> Ramkrishna is still having a problem:
> One more thing do you have idea why this is happening?
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Hello experts,

I was running the polarized sample for pp -> w+ w- j j, where one w is longitudinally polarized and another one is transverse polarized. So, When I used all the default cut then I got 100% generation efficiency.

And when I just changed one cut (0.0 = drjj, its default value is 0.4 ) then I got only 22 events out of demanded 50k events. This seems to me strange since my thinking was If I will loose the cut then it should not decrease the efficiency of cut. It should decrease when we tighten the cut.

Please suggest am I thinking wright or its the kind of same problem that you answered earlier related to phase-space optimization?

Thanks Olivier Mattelaer, that solved my question.