Fork me on GitHub

Opened 11 years ago

Last modified 11 years ago

#252 new Bug

ATLAS simulation gives high pt jet

Reported by: Markus Ebert Owned by:
Priority: minor Milestone:
Component: Delphes code Version: Delphes 3
Keywords: Cc:

Description

Dear Delphes team,

I am currently running simulations using Delphes where I employ both CMS and ATLAS cards for comparison.
I observed that the ATLAS simulation gives some jets with unnaturally high pt values.

As an example, I attached a comparison of ATLAS and CMS created for p p > v v~ j.
My toolchain is MadGraph5 -> pythia -> Delphes.
I imposed a cut of 100 < pt < 200 GeV on parton level.

You can see that there are some events from ATLAS simulation at high pt, up to 1 TeV.
Unless this is some expected behaviour for ATLAS (in which case I hope you could give me an explanation), I think this is a bug in the Delphes ATLAS card.

Kind regards,
Markus

Attachments (1)

selection_0.png (10.8 KB ) - added by Markus Ebert 11 years ago.

Download all attachments as: .zip

Change History (3)

by Markus Ebert, 11 years ago

Attachment: selection_0.png added

comment:1 by Markus Ebert, 11 years ago

Type: How toBug

comment:2 by Michele Selvaggi, 11 years ago

Hi Markus,

thanks for the interesting comparison. What you observe might be related to the fact that in ATLAS, jets are clustered starting from pure calorimeter towers, whereas in CMS the constituents are particle-flow objects (which is some combination of towers and tracks).
The energy response of jets obtained with these two different collections might be different.

As a first sanity check to confirm the previous interpretation, I would suggest you change in the delphes_card_ATLAS.tcl the input collection for the jet clustering from:

Calorimeter/towers

to:

EFlowMerger/eflow

(this should be at line 428 in the ATLAS card.)

Also, please be aware that after the clustering sequence, there is a constant energy scale factor which is applied as a correction to the reconstructed jet energy (line 448). If needed this module supports also a (pt,eta) dependent correction.

Michele

Note: See TracTickets for help on using tickets.