remove outlier rejection from mergedDuplicateTracks

[VinInn]

Outlier rejection was essentially defeating the intentions of the mergedDuplicate algorithm.
The effect of this PR can be seen (for TTBAR 13TeV 25s 35PU aka 25202.0) in
http://innocent.home.cern.ch/innocent/RelVal/dup/plots_summary_highPurity/summary.pdf
more details
http://innocent.home.cern.ch/innocent/RelVal/dup/plots_highPurity/effandfake1.pdf
http://innocent.home.cern.ch/innocent/RelVal/dup/plots_highPurity/dupandfake1.pdf
etc
my interpretation is that MergeDuplicate merges PixelLess and TobTec with PixelSeeded (that’s the intention).
So we have less good tracks in particular in PixelLess and TobTec while fakes are not merged and stays the same.
At the end efficiency decreases (good duplicates are not accounted twice anymore), fake-rate increases (denominator is smaller)

The new tracks are of a "lower quality" (longer but with a higher chi2), but again this is in the intention of the algorithm

[cmsbuild]

A new Pull Request was created by @VinInn (Vincenzo Innocente) for CMSSW_7_6_X.

remove outlier rejection from mergedDuplicateTracks

It involves the following packages:

RecoTracker/FinalTrackSelectors

@cmsbuild, @cvuosalo, @slava77 can you please review it and eventually sign? Thanks.
@ghellwig, @makortel, @GiacomoSguazzoni, @rovere, @VinInn, @mschrode, @istaslis, @gpetruc, @cerati, @dgulhan this is something you requested to watch as well.
You can sign-off by replying to this message having '+1' in the first line of your reply.
You can reject by replying to this message having '-1' in the first line of your reply.
If you are a L2 or a release manager you can ask for tests by saying 'please test' in the first line of a comment.
@Degano you are the release manager for this.
You can merge this pull request by typing 'merge' in the first line of your comment.

[slava77]

@cmsbuild please test

[cmsbuild]

The tests are being triggered in jenkins.

[cmsbuild]

+1
Tested at: 8c837d4
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-11625/8597/summary.html

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-11625/8597/summary.html

[slava77]

Here are some observations based on tests in CMSSW_7_6_X_2015-09-29-0900+ #11492 as a baseline (by now this is the IB baseline):

• the effect is the most significant in high-pt jet sample
• ttbar with PU and other "non-extreme" samples show behavior similar to or even less significant than in 25202

Duplicate rate indeed goes down as expected:
wf202 (run1 ttbar with PU; more similar to current run2 PU)

1338 (dijet flat spectrum up to 3 TeV): for some reason the reduction in dup rate is not symmetric, but this may be stat fluctuation

1313 (3 TeV dijet)

The efficiency, however gets worse:

• wf 202 (and others) show decrease in efficiency at low number of hits (nothing visible at high pt).
  
• wf 1313 shows loss in efficiency at high pt or/and closer to the core of the jet; as well as with increasing displacement
  
  
  

It looks like some tuning may be needed to reduce the effect on efficiency.
In the past a similar loss of efficiency (maybe a bit larger than the one observed here) was a reason to revert changes in tracking. So, I can't sign without some more arguments.

A regular set of tracking pre-validation plots is necessary to be more conclusive about the proposed change.

[slava77]

@VinInn @makortel
please comment.
If the loss of efficiency seems significant, please close this PR.