Use pixel seed extension for phase1 in initialStep

[makortel]

The title says it all. The change was discussed and agreed in May 9 TRK POG meeting
https://indico.cern.ch/event/527844/contributions/2161117/attachments/1269921/1881271/slides_phase1_20160509.pdf

The main point is to exercise regularly the pixel seed extension code from #14356. For 2017 production we may revert this and go back to "quadruplets by triplet propagation" also in initialStep.

Please find below MultiTrackValidator plots with 1000 TTbar+PU35 events showing the effect of this PR on top of 8_1_0_pre5+#14545 (strip alignment as in run2_asymptotic GT)
https://mkortela.web.cern.ch/mkortela/tracking/validation/CMSSW_8_1_0_pre5_pse/

Tested in CMSSW_8_1_X_2016-05-24-1100, non-2017 workflows should have no effect. 2017 workflows should have small effect on the level of generalTracks (mainly decrease of duplicate rate), while initialStep and highPtTripletStep have larger changes (mainly migration from highPtTripletStep to initialStep).

@rovere @VinInn

[cmsbuild]

A new Pull Request was created by @makortel (Matti Kortelainen) for CMSSW_8_1_X.

It involves the following packages:

RecoTracker/IterativeTracking

@cmsbuild, @cvuosalo, @slava77, @davidlange6 can you please review it and eventually sign? Thanks.
@ghellwig, @GiacomoSguazzoni, @rovere, @VinInn, @mschrode, @gpetruc, @dgulhan this is something you requested to watch as well.
@slava77, @Degano, @smuzaffar you are the release manager for this.

cms-bot commands are list here #13028

[VinInn]

@cmsbuild , please test

[cmsbuild]

The tests are being triggered in jenkins.
https://cmssdt.cern.ch/jenkins/job/ib-any-integration/13161/console

[cmsbuild]

+1
Tested at: 48c74d0
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-14621/13161/summary.html

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-14621/13161/summary.html

[slava77]

in 10224 on 100 events:
fewer and shorter electron seeds

there are a few more gedGsfElectrons and they are apparently with smaller fbrem

Charged hadron yield reduces at higher eta (much smaller effect in generalTracks)

as expected in the PR description: there is a migration from highPtTripletStep to initialStep in generalTracks

The effect on electron seeds is even larger in ZEE PU35 sample 10225

It looks like electrons are more effectively made with these shorter seeds, since the effect on ged electrons is somewhat positive.

There is some decrease in fakes+dups in generalTracks
https://mkortela.web.cern.ch/mkortela/tracking/validation/CMSSW_8_1_0_pre5_pse/ttbar_pu_ootb/effandfake4.pdf
.. these are in line with the presentation in the tracking POG

Before signing, I wanted to check if the large effect on electron seeding is intended.

[makortel]

Thanks @slava77 for pointing it out. Electron seeds become shorter because they use initialStepSeeds (among others) and those are changed from quadruplets to triplets in this PR. I produced MTV plots for gsfElectrons (1000 events of ttbar+PU)
https://mkortela.web.cern.ch/mkortela/tracking/validation/CMSSW_8_1_0_pre5_pse_ele/ttbar_pu_electronGsf.html
and I don't see any statistically significant effect. So even though the electron seeds change, the downstream effects seem to be negligible. And anyway the (tracking) tuning is not final, but point taken.

[slava77]

@makortel remind me, please, what are the selections for TP in the gsf MTV.
Is it primary electrons or something else/more?

[makortel]

@slava77 For efficiencies, first only TrackingParticles with abs(pdgId)==11 are selected
https://github.com/cms-sw/cmssw/blob/CMSSW_8_1_X/Validation/RecoTrack/python/TrackValidation_cff.py#L272
and then the usual plot-specific selections are applied
https://github.com/cms-sw/cmssw/blob/CMSSW_8_1_X/Validation/RecoTrack/python/TrackingParticleSelectionsForEfficiency_cff.py

Fake and duplicate rates are calculated wrt. all (also non-electron) TrackingParticles.

[slava77]

+1

for #14621 48c74d0

• configuration code changes are in line with the description; only phase1 is affected
• jenkins tests pass and show differences only in phase 1 workflow
• local tests with higher stats using 10224 and 10225 with PU and 100{24,25,26} without pileup don't show major changes in final (generalTracks) performance: efficiency is about the same and duplicate rates are smaller as expected; electron seeding is using triplets now and the electron efficiency is about the same (MTV GSF plots are good), there is some reduction in fbrem.

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next CMSSW_8_1_X IBs (tests are also fine). This pull request requires discussion in the ORP meeting before it's merged. @slava77, @davidlange6, @Degano, @smuzaffar

[davidlange6]

+1