Cannot normalize to luminosity

Dear Antony,

I think that you have forgotten to account for the event weight in your normalization, as it could be different from one.

Cheers,

Benjamin

Dear Benjamin,

I changed
"
H0_->IncrementNEvents(__event_weight__);

H0_->Fill(_P_j_I1I_PTorderingfinalstate[ind[0]]->pt(),__event_weight__);
"

into

"
// H0_->IncrementNEvents(__event_weight__);

      H0_->Fill(_P_j_I1I_PTorderingfinalstate[ind[0]]->pt(),event.mc()->weight());
"

but it doesn't work...was it our point?

cheers,

Antony

Hi Antony,

It will indeed not work. Your rescaling assumes that the original event weight is equal to one, which is generally not the case. You need to calculate the new weight with respect to the desired luminosity and cross section.

Cheers,

Benjamin

On 23 Apr 2014, at 10:31 , Antony Martini <email address hidden> wrote:

> Question #247483 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/247483
>
> Antony Martini posted a new comment:
> Dear Benjamin,
>
> I changed
> "
> H0_->IncrementNEvents(__event_weight__);
>
> H0_->Fill(_P_j_I1I_PTorderingfinalstate[ind[0]]->pt(),__event_weight__);
> "
>
> into
>
> "
> // H0_->IncrementNEvents(__event_weight__);
>
> H0_->Fill(_P_j_I1I_PTorderingfinalstate[ind[0]]->pt(),event.mc()->weight());
> "
>
> but it doesn't work...was it our point?
>
> cheers,
>
> Antony
>
> --
> You received this question notification because you are a direct
> subscriber of the question.

Ok but so how do you do that? because I suppose that ma5 does it in the user friendly mode...

thanks,

Antony

because I tried also :
"
double eventweight=1;
    if (event.mc()!=0) eventweight=event.mc()->weight();
    UInt_t ind[1];
    for (ind[0]=0;ind[0]<_P_j_I1I_PTorderingfinalstate.size();ind[0]++)
    {
      H0_->Fill(_P_j_I1I_PTorderingfinalstate[ind[0]]->pt(),eventweight);
    }

"
 and it's still the same...

Hi Antony,

This exactly means that the event weight is the one present in the LHE file. In your previous root file, you could simply divide by the integral of the histogram over the number of entries.

Cheers,

Benjamin

On 23 Apr 2014, at 12:07 , Antony Martini <email address hidden> wrote:

> Question #247483 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/247483
>
> Antony Martini posted a new comment:
> because I tried also :
> "
> double eventweight=1;
> if (event.mc()!=0) eventweight=event.mc()->weight();
> UInt_t ind[1];
> for (ind[0]=0;ind[0]<_P_j_I1I_PTorderingfinalstate.size();ind[0]++)
> {
> H0_->Fill(_P_j_I1I_PTorderingfinalstate[ind[0]]->pt(),eventweight);
> }
>
> "
> and it's still the same...
>
> --
> You received this question notification because you are a direct
> subscriber of the question.

Hi Benjamin,
                            I am a similar question. At parton level my event file contains 50,000 events. Now I am analysing the root file. Suppose, I want to analyze the events at Lumi = 300 fb^-1 and finally normalize all events to 1.

In normal mode, If I use the following commands, it is solving the job.
set sig.xsection = 0.2046
 set main.stacking_method = normalize2one
set main.lumi = 1000
set main.normalize = lumi

I n expert mode, how to the same thing ? Could you please tell me which lines should I add for the above purpose in the .cpp file ?

Another question, Like normal mode, Is there a way to obtain the signal significance in the expert mode automatically ?

Regards,
Shiba

Hi Shiba,

In the expert mode, you can get the selection efficiency from the output SAF file. Then, it is up to you to work out the rest (nothing predefined exists), e.g. by using an external routine.

Cheers

Benjamin

On 13 Mar 2016, at 14:13 , shibasipu <email address hidden> wrote:

> Question #247483 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/247483
>
> shibasipu posted a new comment:
> Hi Benjamin,
> I am a similar question. At parton level my event file contains 50,000 events. Now I am analysing the root file. Suppose, I want to analyze the events at Lumi = 300 fb^-1 and finally normalize all events to 1.
>
> In normal mode, If I use the following commands, it is solving the job.
> set sig.xsection = 0.2046
> set main.stacking_method = normalize2one
> set main.lumi = 1000
> set main.normalize = lumi
>
> I n expert mode, how to the same thing ? Could you please tell me which
> lines should I add for the above purpose in the .cpp file ?
>
>
> Another question, Like normal mode, Is there a way to obtain the signal significance in the expert mode automatically ?
>
>
> Regards,
> Shiba
>
> --
> You received this question notification because you are subscribed to
> the question.

Hi Benjamin,
                           Thanks for the reply. I am trying to do the above stuffs. I am not sure whether I can implement the above stuffs in the expert mode. I may need your help later.

Regards,
Shiba