Warning in reading delphe root files

[liuda]

Hi,

When I try to read the root files output from Delphes 3.0.9, there is a warning:

Warning in <TStreamerInfo::BuildCheck>:

The StreamerInfo of class Jet read from file tpj_1000_2/Events/run_02_0/tag_tpj_delphes_events.root
has the same version (=2) as the active class but a different checksum.
You should update the version to ClassDef(Jet,3).
Do not try to write objects with the current class definition,
the files will not be readable.

Warning in <TStreamerInfo::CompareContent>: The following data member of
the on-file layout version 2 of class 'Jet' differs from
the in-memory layout version 2:

UInt_t BTag;

vs

Int_t BTag;

Warning in <TStreamerInfo::CompareContent>: The following data member of
the on-file layout version 2 of class 'Jet' differs from
the in-memory layout version 2:

UInt_t TauTag;

vs

Int_t TauTag;

What can I do to avoid this? How to update the version to ClassDef(Jet,3)?

Another question: I have checked that the jets in an event are ordered by pT, do all the physical objects (leptons) stored in the same way (ordered by pT)?

Best,

Da Liu

[Michele Selvaggi]

Dear Da Liu,

thanks for pointing this out, we will fix in the next release.

To answer your question: every output collection is pT ordered.

Cheers,

Michele

[liuda]

Dear Michele Selvaggi,

Thank you very much. Is the root file okay for analysis? (I don't use the jet branch in my analysis.)

In addition, I always encounter that the number of events in delphes.root files is smaller than the number of unweighted.root file.
For instance, the parton level event number is 50,000, while the final number in delphes.root is 49,999. Why doest it happen?

Best,

Da Liu

Replying to mselvaggi:

Dear Da Liu,

thanks for pointing this out, we will fix in the next release.

To answer your question: every output collection is pT ordered.

Cheers,

Michele

[Michele Selvaggi]

Hello,

yes you should use the root file for the analysis.

If you are generating events with MadGraph you should use the *.hep format, which contain events after showering and hadronization.
You shoudn't use the unweighted_events.lhe as input for Delphes.
Is this what you are doing?

Could you please describe the procedure you use to generate the Delphes output?

Michele

[liuda]

Hi,

Actually, I run the delphes mode in MaGraph5 (Automatically), I think the input for delphes is *.hep format. I set the event number in the run_card.dat 50,000, but the number in the delphes.root is 49,999.

Best,

Da Liu

Replying to mselvaggi:

Hello,

yes you should use the root file for the analysis.

If you are generating events with MadGraph you should use the *.hep format, which contain events after showering and hadronization.
You shoudn't use the unweighted_events.lhe as input for Delphes.
Is this what you are doing?

Could you please describe the procedure you use to generate the Delphes output?

Michele

[Michele Selvaggi]

I have just checked on a DY sample generated with MG and I recover all the events n the Delphes output.

Do you apply any matching procedure in MadGraph? MG might veto some events.

Could you check if you lose that 1 event also by requiring less events?

In case you don't, could you provide the command lines for generating your sample in MG?

Michele

[liuda]

Hi,

Actually, I used the multi_run command to generate large events (each 50,000) and only few runs (1 or 2) encounter the problem. I have found a run in this case, it seems that it is a problem of pythia, because the number of events in pythia_lhe_events.root file is also 49,999.

Best,

Da Liu