Tracking Variable Study

Follow up

Ryan KelleyBoris ManganoVivek Sharma

p.2

Purpose

• The purpose of this study is to look at tracking variables in order to determine which tracks are ‘good’ and which are ‘fakes’.

• Basically, what is a good set of cuts to use (for example isolation studies)?

p.3

Define matched vs. unmatched

• Loop through all recoTracks in the event and use the RecoToSimAssociator function (associate by hits) in CMSSW to determine if this track is matched to a simulator level particle.

• If there is no simTrack associated, then it is an unmatched or fake recoTrack.

p.4

Tracking Variables Considered

• Most Powerful Cut-- hits > 7.

• SimTrack PT < 1 GeV cut.

• Don’t understand the two ‘towers’ (Boris believes from increased material).

p.5

Tracking Variables Considered

• Similar spread for both matched and unmatched.

• z0 < 30 cm doesn’t really give you anything.

• Sharp peak between d0 < 1mm.

• Fakes are very spread.

p.6

Tracking Variables Considered

• Factor of 10 difference and increasing from 2 > 10.

p.7

Z (Matched)• ~50k Z

events where each recoTrack is matched to a SimTrack.

p.8

Z (Unmatched)• ~50k Z events

where each recoTrack is NOT matched to a SimTrack.

p.9

QCD (Matched)• ~120k QCD (50 < pt

< 80 GeV) events where each recoTrack is matched to a SimTrack.

p.10

QCD (Unmatched)

• Need to run on a larger sample

p.11

Ratio of all tracking cuts to no cuts

p.12

Things to do

• Try reverse matching: Defined unmatched as a SimTrack that doesn’t have a RecoTrack (done).

• What are the towers (Boris believes it’s the extra material due to cabling/etc. at around 1.5).

• Run on larger samples (done, but should have run on more QCD samples with different pt ranges).

• Try to cut on expected hits instead of total hits--some of the regions have more layers than others. A hard number may be not work well the regions with fewer layers.

p.13

BackupsNormalized valid hits plot.