Reading weights

Asked by Amin Aboubrahim on 2017-12-05

Dear MA5 team,

I am using MA5 in the non-expert mode. I have some SM background samples from the Snowmass group and a signal sample. I'm trying to implement some cuts and plot histograms. The Snowmass samples come with weighted events. MA5 PDF output warns me of insufficient statistics so I'm suspecting that it is not taking the weights into consideration. It is a trick that Snowmass group has made to generate fewer number of events while still respecting the statistics.

1) Is there a way to plot those histograms while reading the weights? The weights are stored in the Weight leaf of the ROOT file.
2) Is the main.SBratio still not functional for ROOT 6?

Thank you very much.
Amin

Question information

Language:
English Edit question
Status:
Solved
For:
MadAnalysis 5 Edit question
Assignee:
No assignee Edit question
Solved by:
Benjamin Fuks
Solved:
2017-12-05
Last query:
2017-12-05
Last reply:
2017-12-05
Benjamin Fuks (fuks) said : #1

Hi Amin,

> 1) Is there a way to plot those histograms while reading the weights? The weights are stored in the Weight leaf of the ROOT file.
The weights are in principle read and used by ma5. You can check it by yourself in editing the generated user.cpp file and printing the weights.

> 2) Is the main.SBratio still not functional for ROOT 6?
This method is not used anymore. You must used the fom one (factor of merit).

Cheers,

Benjamin

Amin Aboubrahim (amin83) said : #2

Hi Benjamin,

Thank you for the information. Indeed the histograms are plotted taking the weights into consideration. I think this does not prevent MA5 from complaining about insufficient statistics. I guess I can ignore the warning!
So one can set main.fom.formula=4 to get S/sqrt(S+B)?

Few more questions:
1) My ROOT file has fat jets. Does plotting PT(j[1]) take fat jets and jets both into consideration?
2) Can we plot the N-subjettiness in the non-expert mode?
3) In generating SM backgrounds for a 100 TeV collider, getting the good statistics for 1000 fb^-1 requires a huge number of events to be generated. Do you know of a method to get weighted events that would allow one to generate fewer events? I'd rather not use the Snowmass samples because they have a lot of problems! I know this is off topic but I thought I'd ask an expert for an advice!

Thanks again,
Amin

Best Benjamin Fuks (fuks) said : #3

Hi Amin,

> So one can set main.fom.formula=4 to get S/sqrt(S+B)?
Yes.

>
> Few more questions:
> 1) My ROOT file has fat jets. Does plotting PT(j[1]) take fat jets and jets both into consideration?
Only one type of jets (that could be either fat or non fat jets) is supported. If you have several types of jets in your root file, only the 'normal' ones will be considered.

> 2) Can we plot the N-subjettiness in the non-expert mode?
No.

> 3) In generating SM backgrounds for a 100 TeV collider, getting the good statistics for 1000 fb^-1 requires a huge number of events to be generated. Do you know of a method to get weighted events that would allow one to generate fewer events? I'd rather not use the Snowmass samples because they have a lot of problems! I know this is off topic but I thought I'd ask an expert for an advice!
You can generate weighted events with most event generators on the market, I guess.

Regards,

Benjamin

Amin Aboubrahim (amin83) said : #4

Thanks Benjamin Fuks, that solved my question.