Ma5 gives different amount of tau+ tau- for root file and lhco file

Hi,

This is very surprizing. Could you double check the way you produced your two samples? In particular the LHCO one. Maybe you are comparing not comparable things.

Cheers,

Benjamin

On 29 Sep 2015, at 18:17 , degos <email address hidden> wrote:

> New question #271900 on MadAnalysis 5:
> https://answers.launchpad.net/madanalysis5/+question/271900
>
> Hi,
> I use Madgraph5 to generate 1800 g g > b b~. In the delphes_card (delphes3.2.0 is called), for the tau-tagging module, I set
> add EfficiencyFormula {0} {1.00}
> add EfficiencyFormula {15} {0.00}
> Here I assume all the jets are misidentified as tau.
>
> Authors of delphes told me that the charge of fake tau is simply a flip coin in delphes-3.2.0, namely that the amount of tau+ should approximately equal to the tau-.
>
> I use the user-friendly framework of Madanalysis5 (latest version) to analyze the ".root" file. I obtain 1008 tau+ and 1030 tau-, which is consistent with the flip coin charge of fake tau as the author of delphes said. However, I use Madanalysis5 to analyze the correponding ".lhco" file, and obain 1188 tau+ and 850 tau-. Compared to analysis of ".root" file, the amount of tau+ and tau- are both changed, and particularly tau+ is much more than tau-, which is not consistent with as authors of delphes said.
>
> In addition, if the tau-tagging mudule is taken as (assume the jets not to be misidentified as tau)
> add EfficiencyFormula {0} {0.00}
> add EfficiencyFormula {15} {1.00}
> I use the Madanalysis5 to analyze the ".lhco" file, and find that the amount of tau+ equals to tau- approximately. The results are approxmately same as the analysis of ".root" file.
>
> I discussed detailedly with authors of delphes on the problem in the following website
> https://cp3.irmp.ucl.ac.be/projects/delphes/ticket/725
>
> Is it a bug of Madanalysis5? How to understand the problem?
>
> Many thanks.
>
> --
> You received this question notification because you are an answer
> contact for MadAnalysis 5.

Hi,

I use two different approaches of producing the .root file and lhco file:
1. Madgraph5 calls delphes to produce the .root file and lhco file.
2. I copy the .hep.gz file to delphes folder, and directly use delphes to produce .root file and lhco file.
I get the same conclusion. (the detailed operation is introduced in https://cp3.irmp.ucl.ac.be/projects/delphes/ticket/725 )

Since I am not familiar with .root file and lhco file, I do not know whether the thing is correct or a bug.

You can download my "tag_1_pythia_events.hep.gz" and "bb.lhco"​ from the following website.
https://cp3.irmp.ucl.ac.be/projects/delphes/ticket/725

Could you check it? How to understand the problem?

Many thanks.

Hi Degos,

This may be related to delphes and the way the LHCO file is created. I cannot find a bug in ma5 with the LHCO reader. If you call delphes within ma5, you should in any case only use the outputted root file.

Cheers,

Benjamin

On 29 Sep 2015, at 23:47 , degos <email address hidden> wrote:

> Question #271900 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/271900
>
> Status: Answered => Open
>
> degos is still having a problem:
> Hi,
>
> I use two different approaches of producing the .root file and lhco file:
> 1. Madgraph5 calls delphes to produce the .root file and lhco file.
> 2. I copy the .hep.gz file to delphes folder, and directly use delphes to produce .root file and lhco file.
> I get the same conclusion. (the detailed operation is introduced in https://cp3.irmp.ucl.ac.be/projects/delphes/ticket/725 )
>
> Since I am not familiar with .root file and lhco file, I do not know
> whether the thing is correct or a bug.
>
> You can download my "tag_1_pythia_events.hep.gz" and "bb.lhco"​ from the following website.
> https://cp3.irmp.ucl.ac.be/projects/delphes/ticket/725
>
> Could you check it? How to understand the problem?
>
> Many thanks.
>
> --
> You received this question notification because you are an answer
> contact for MadAnalysis 5.

Hi Benjamin，

Thank you for your help. Is it more safe to take the results from analysis of ".root file"?

Best

Hi Degos,

Well, I know how the root file is created. I do not know how Delphes creates the LHCO file. I will continue investigating any possible issue in there, but I cannot find anything for the moment.

Cheers,

Benjamin

On 30 Sep 2015, at 10:47 , degos <email address hidden> wrote:

> Question #271900 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/271900
>
> Status: Answered => Open
>
> degos is still having a problem:
> Hi Benjamin，
>
> Thank you for your help. Is it more safe to take the results from
> analysis of ".root file"?
>
> Best
>
> --
> You received this question notification because you are an answer
> contact for MadAnalysis 5.

Hi Benjamin,
Many thanks for your help. This problem seems not appear for the fake electron and muon from jet.

http://madgraph.phys.ucl.ac.be/Manual/lhco.html
The above website mention some differentces between electron and tau in the LHCO file:

"The seventh column gives the number of tracks associated with the object; in the case of a lepton, this number is multiplied by the charge of the lepton. (Thus a muon will appear as -1, a positron as +1, a tau- as -1 or -3, and a jet a positive number or zero.)"

Best

Hi Degos,

Tau identification is made from the reading of the second column. Then the tau charge identification is from the reading of the sign of the content of the seventh column, so that the number of tracks is not relevant. Coujld you please share the would-be faulty LHCO file (but with maybe only 10-20 events included)?

Cheers,

Benjamin

On 01 Oct 2015, at 09:58 , degos <email address hidden> wrote:

> Question #271900 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/271900
>
> Status: Answered => Open
>
> degos is still having a problem:
> Hi Benjamin,
> Many thanks for your help. This problem seems not appear for the fake electron and muon from jet.
>
> http://madgraph.phys.ucl.ac.be/Manual/lhco.html
> The above website mention some differentces between electron and tau in the LHCO file:
>
> "The seventh column gives the number of tracks associated with the object; in the case of a lepton, this number is multiplied by the charge of the lepton. (Thus a muon will appear as -1, a positron as +1, a tau- as -1 or -3, and a jet a positive number or zero.)"
>
> Best
>
> --
> You received this question notification because you are an answer
> contact for MadAnalysis 5.

Hi Benjamin,

I tried to produce several LHCO file with 20 events. Each LHCO file contains several tau with ntrk=0. If tau with ntrk=0 are taken as the plus charge of tau, the results of Madanalysis5 are correct.

Maybe there is either a problem with tau with ntrk=0, either with creation of LHCO file from root file.

Best

Hi,

You should not get taus with 0 tracks. That's why ma5 cannot handle them, as you want.

Cheers,

Benjamin

On 01 Oct 2015, at 13:52 , degos <email address hidden> wrote:

> Question #271900 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/271900
>
> Status: Answered => Open
>
> degos is still having a problem:
> Hi Benjamin,
>
> I tried to produce several LHCO file with 20 events. Each LHCO file
> contains several tau with ntrk=0. If tau with ntrk=0 are taken as the
> plus charge of tau, the results of Madanalysis5 are correct.
>
> Maybe there is either a problem with tau with ntrk=0, either with
> creation of LHCO file from root file.
>
> Best
>
> --
> You received this question notification because you are an answer
> contact for MadAnalysis 5.

Hi Benjamin,

So the problem is from the delphes which does not produce the correct LHCO file?

Best

That's indeed my impression.

Cheers,

Benjamin

On 01 Oct 2015, at 15:08 , degos <email address hidden> wrote:

> Question #271900 on MadAnalysis 5 changed:
> https://answers.launchpad.net/madanalysis5/+question/271900
>
> Status: Answered => Open
>
> degos is still having a problem:
> Hi Benjamin,
>
> So the problem is from the delphes which does not produce the correct
> LHCO file?
>
> Best
>
> --
> You received this question notification because you are an answer
> contact for MadAnalysis 5.

Hi Benjamin,

This problem is a bug of delphes. The author fixed it.

In addition, I use the user-friendly framework of Madanalysis5 to analyze the ".root" file directly. Namely,

import bb.root

Madanalysis5 always reports the folowing warning:
WARNING: GenParticle corresponding to a photon is not found in the gen table

Does this warning imply that the results are maybe wrong?
If not, how to make Madanalysis5 not to report this warning?

Best

Hi,
You can ignore the warning.
Cheers,
Benjamin

-------- Original message --------
From: degos <email address hidden>
Date: 01/10/2015 17:57 (GMT+01:00)
To: <email address hidden>
Subject: Re: [Question #271900]: Ma5 gives different amount of tau+ tau- for
  root file and lhco file

Question #271900 on MadAnalysis 5 changed:
https://answers.launchpad.net/madanalysis5/+question/271900

    Status: Answered => Open

degos is still having a problem:
Hi Benjamin,

This problem is a bug of delphes. The author fixed it.

In addition, I use the user-friendly framework of Madanalysis5 to
analyze the ".root" file directly. Namely,

import  bb.root

Madanalysis5 always reports the folowing warning:
WARNING: GenParticle corresponding to a photon is not found in the gen table

Does this warning imply that the results are maybe wrong?
If not, how to make Madanalysis5 not to report this warning?

Best

--
You received this question notification because you are an answer
contact for MadAnalysis 5.