Fork me on GitHub

Opened 12 years ago

Closed 11 years ago

#95 closed Bug (fixed)

Photons

Reported by: alberghi@… Owned by:
Priority: major Milestone:
Component: Delphes code Version: 1.9
Keywords: Cc:

Description

I am looking at the photon transverse energy for a process ttbar decay process.
I surprisingly find that the GEN and ANALYSIS distributions are very different.
To select photons I am using conditions like
abs(Particle_PID[n]) == 22 && Particle_PT[n] > 20 && abs( Particle_Eta[n]) < 2.5
in the GEN tree
and
Photon_PT[n] > 20 && abs( Photon_Eta[n]) < 2.5
in the Analysis tree.

Could you help me find what point I am missing ?

Thank you

Attachments (2)

True-photon-ET.eps (8.6 KB ) - added by anonymous 12 years ago.
Rec-Photon_ET.eps (9.8 KB ) - added by anonymous 12 years ago.

Download all attachments as: .zip

Change History (6)

by anonymous, 12 years ago

Attachment: True-photon-ET.eps added

by anonymous, 12 years ago

Attachment: Rec-Photon_ET.eps added

comment:1 by Sourav Mandal, 12 years ago

In the analysis tree cut try adding something like "Photon_EHoverEE < .1", and the distributions should look more similar.

In the code it looks like Delphes tracks the calo towers hit by the photons (presuming they are identifiiable in the tracker?), then retrieves the calo information from those towers. Since the photons in a top decay are likely to be collimated with hadronic final states, it's not surprising that there is hadron contamination in these towers, increasing the registered PT and changing the photon PT distribution.

By filtering on EHoverEE you'll get the hits which are predominantly photonic, though obviously you'll lose many real photons.

in reply to:  1 comment:2 by alberghi@…, 12 years ago

Thanks a lot for your prompt answer.
Implementing your suggested cut on EHoverEE the things went much better.
We thought that the photon identification was done more effectively as it is for electrons and muons.
Instead it seems that the user should implement the identification by himself.
In our view, as a first approximation where we have tracking acceptance, photon identification should be the same as the electron one at the calorimeter level, requiring, differently respect to electrons, no pointing track to the calorimeter deposit.
If possible, we would like to better understand what Delphes really does with photons.
Could you please tell us where we can find the relevant code so that we can have a look ?

comment:3 by favereau, 12 years ago

Hi,

in Delphes.cpp, have a look around lines containing

list_of_towers_with_photon

(around line 535 and 600) It should give you a general idea.

Jerome

comment:4 by cp3-support, 11 years ago

Resolution: fixed
Status: newclosed
Note: See TracTickets for help on using tickets.