[10_2_X] MXNet-based implementation of the ParticleNet tagger

[hqucms]

Backport of #28902.
For analysis of the full Run2 data and NanoAOD production.

Needs:

• DNN model: DNN models for the ParticleNet tagger cms-data/RecoBTag-Combined#26
• MXNet 1.5: [10.2.X] Update MXNet to 1.5.0 cmsdist#5591

We should also integrate cms-data/RecoBTag-Combined#24 together with the update to MXNet 1.5.

[cmsbuild]

A new Pull Request was created by @hqucms (Huilin Qu) for CMSSW_10_2_X.

It involves the following packages:

PhysicsTools/PatAlgos
RecoBTag/Configuration
RecoBTag/FeatureTools
RecoBTag/MXNet

@perrotta, @cmsbuild, @santocch, @slava77 can you please review it and eventually sign? Thanks.
@rappoccio, @gouskos, @hatakeyamak, @emilbols, @peruzzim, @seemasharmafnal, @mmarionncern, @ahinzmann, @smoortga, @jdolen, @ferencek, @jdamgov, @nhanvtran, @gkasieczka, @schoef, @andrzejnovak, @clelange, @riga, @JyothsnaKomaragiri, @mverzett, @gpetruc, @mariadalfonso this is something you requested to watch as well.
@davidlange6, @silviodonato, @fabiocos you are the release manager for this.

cms-bot commands are listed here

[silviodonato]

@hqucms could you explain in the PR description why you need this backport?

[hqucms]

@silviodonato We would like to have this for analysis of the full Run2 data and NanoAOD production. I updated the PR description with that.

[cmsbuild]

Pull request #29063 was updated. @perrotta, @cmsbuild, @santocch, @slava77 can you please check and sign again.

[silviodonato]

@hqucms could you confirm the last two commits make this PR identical to #28902 ?

[silviodonato]

please test with cms-sw/cmsdist#5591

[cmsbuild]

The tests are being triggered in jenkins.
Tested with other pull request(s) cms-sw/cmsdist#5591
https://cmssdt.cern.ch/jenkins/job/ib-run-pr-tests/5106/console Started: 2020/03/10 16:13

[hqucms]

@hqucms could you confirm the last two commits make this PR identical to #28902 ?

@silviodonato Yes -- this backport is identical as #28902 except that the tagger is not enabled in the MiniAOD step to preserve the event content.

[cmsbuild]

+1
Tested at: 5fe1910
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-6c57c4/5106/summary.html
CMSSW: CMSSW_10_2_X_2020-03-10-1100
SCRAM_ARCH: slc6_amd64_gcc700

[cmsbuild]

Comparison job queued.

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-6c57c4/5106/summary.html

Comparison Summary:

• No significant changes to the logs found
• Reco comparison results: 0 differences found in the comparisons
• DQMHistoTests: Total files compared: 31
• DQMHistoTests: Total histograms compared: 3007491
• DQMHistoTests: Total failures: 1
• DQMHistoTests: Total nulls: 0
• DQMHistoTests: Total successes: 3007300
• DQMHistoTests: Total skipped: 190
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.0 KiB( 30 files compared)
• Checked 129 log files, 14 edm output root files, 31 DQM output files

[slava77]

@cmsbuild please test

to refresh the comparisons

[cmsbuild]

Pull request #29063 was updated. @perrotta, @cmsbuild, @santocch, @slava77 can you please check and sign again.

[cmsbuild]

Pull request #29063 was updated. @perrotta, @cmsbuild, @santocch, @slava77 can you please check and sign again.

[hqucms]

@slava77 Done in fcc6974 and 0beb8ab.

[slava77]

@cmsbuild please test

this will just check that the "production" workflows are not broken; I'll re-check locally for the test with _devel enabled

[cmsbuild]

The tests are being triggered in jenkins.
Tested with other pull request(s) cms-sw/cmsdist#5719
https://cmssdt.cern.ch/jenkins/job/ib-run-pr-tests/5607/console Started: 2020/04/08 19:38

[cmsbuild]

+1
Tested at: 0beb8ab
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-beec7a/5607/summary.html
CMSSW: CMSSW_10_2_X_2020-04-08-1100
SCRAM_ARCH: slc6_amd64_gcc700

[cmsbuild]

Comparison job queued.

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-beec7a/5607/summary.html

Comparison Summary:

• No significant changes to the logs found
• Reco comparison results: 0 differences found in the comparisons
• DQMHistoTests: Total files compared: 31
• DQMHistoTests: Total histograms compared: 3007544
• DQMHistoTests: Total failures: 2
• DQMHistoTests: Total nulls: 0
• DQMHistoTests: Total successes: 3007352
• DQMHistoTests: Total skipped: 190
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.0 KiB( 30 files compared)
• Checked 129 log files, 14 edm output root files, 31 DQM output files

[santocch]

+1

[slava77]

+1

for #29063 0beb8ab

• code changes are in line with the PR description and the follow up review. The backport is consistent with the master version, ignoring some parts that do not apply for the backport (e.g. BoostedJetMXNetJetTagsProducer.cc fillDescriptions and the coupled pfParticleNet_cff.py)
• jenkins tests pass and comparisons with the baseline show no differences (by default the new code is not running)
• the PR was tested locally using 1325.7 and 136.8311 with run2_miniAOD_devel modifier added to enable the non-production extras: changes appear only in slimmedJetsAK8 in the jet tag discriminants with only particleNet taggers added while the available ones are unmodified; a "remote" comparison with the same workflow 136.8311 in 11_1_0_pre5 shows fairly similar results in the particleNet tagger values.

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next CMSSW_10_2_X IBs (tests are also fine) and once validation in the development release cycle CMSSW_11_1_X is complete. This pull request will now be reviewed by the release team before it's merged. @silviodonato, @dpiparo (and backports should be raised in the release meeting by the corresponding L2)

[silviodonato]

+1