[12.4.X] address undef output of dnn - for issue #37970

[mmusich]

resolves #37970

PR description:

This PR addresses issue #37970.

In the cand level DNN run after MkFit the output is undetermined if the propagation to the beamline is invalid. This happens in 0.1% of cases at most.

This behaviour is flagged and the output set to -1 (i.e. failing the selection).

I also added a check on ndof > 0 (used to divide chi2), which didn't create problems so far.

PR validation:

Example of difference in workflow 12434

no PR #37918 ------ | with PR #37918 ----- | fixed

0.951501 the output | 0.951501 the output | 0.951501 the output
0.931231 the output | 0.931231 the output | 0.931231 the output
0.877139 the output | 0.877139 the output | 0.877139 the output
0.557774 the output | 0.557774 the output | 0.557774 the output
0.97176 the output | 0.97176 the output | 0.97176 the output
0.761749 the output | -nan the output | -1 the output
0.485212 the output | 0.485212 the output | 0.485212 the output
-0.888864 the output | -0.888864 the output | -0.888864 the output
0.990027 the output | 0.990027 the output | 0.990027 the output
0.769534 the output | 0.769534 the output | 0.769534 the output
0.962413 the output | 0.962413 the output | 0.962413 the output

The physics changes in MTV validation are almost zero, as expected.

if this PR is a backport please specify the original PR and why you need to backport that PR:

verbatim backport of #38004

[cmsbuild]

A new Pull Request was created by @mmusich (Marco Musich) for CMSSW_12_4_X.

It involves the following packages:

• RecoTracker/MkFit (reconstruction)

@jpata, @cmsbuild, @clacaputo, @slava77 can you please review it and eventually sign? Thanks.
@makortel, @felicepantaleo, @GiacomoSguazzoni, @JanFSchulte, @rovere, @VinInn, @ebrondol, @gpetruc, @mmusich, @mtosi, @dgulhan this is something you requested to watch as well.
@perrotta, @dpiparo, @qliphy you are the release manager for this.

cms-bot commands are listed here

• Backported from Fix for issue 37970 - address undetermined output of cand DNN in mkFit #38004

[mmusich]

backport of #38004

[mmusich]

type bug-fix

[jpata]

@cmsbuild please test

[cmsbuild]

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-ecc7ac/24853/summary.html
COMMIT: 0435d74
CMSSW: CMSSW_12_4_X_2022-05-19-1100/slc7_amd64_gcc10
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmssw/38011/24853/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

• No significant changes to the logs found
• Reco comparison results: 4 differences found in the comparisons
• DQMHistoTests: Total files compared: 50
• DQMHistoTests: Total histograms compared: 3672464
• DQMHistoTests: Total failures: 89
• DQMHistoTests: Total nulls: 1
• DQMHistoTests: Total successes: 3672352
• DQMHistoTests: Total skipped: 22
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: -0.004 KiB( 49 files compared)
• DQMHistoSizes: changed ( 312.0 ): -0.004 KiB MessageLogger/Warnings
• Checked 208 log files, 45 edm output root files, 50 DQM output files
• TriggerResults: no differences found

[jpata]

+reconstruction

• backport of Fix for issue 37970 - address undetermined output of cand DNN in mkFit #38004, fixes DQM instability in 12434.0, pixelLessStep_quickAssociatorByHits #37970
• minimal physics changes are expected, but given this is a bugfix, we expect it to be OK for 12_4_0

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next CMSSW_12_4_X IBs (tests are also fine) and once validation in the development release cycle CMSSW_12_5_X is complete. This pull request will now be reviewed by the release team before it's merged. @perrotta, @dpiparo, @qliphy (and backports should be raised in the release meeting by the corresponding L2)

[qliphy]

+1
bugfix