New trainings for mvaTauID + one backport from CMSSW_10_0_X #21977

roger-wolf · 2018-01-26T16:55:06Z

Dear Slava, Andrea,

this PR consists of a modification of the miniAOD sequences, adding new mvaTauID discriminants that have been trained with 2017 MCv1 (tauIdMVAIsoDBoldDMwLT2017). Modified files are:

RecoTauTag/Configuration/python/loadRecoTauTagMVAsFromPrepDB_cfi.py
RecoTauTag/Configuration/python/updateHPSPFTaus.py
PhysicsTools/PatAlgos/python/slimming/miniAOD_tools.py

This modification impacts both the standard and legacy80X workflows via an updated makePatTausTask.

In addition the protection to prevent the PATTauProducer from asserting when encountering requests for tauID WP that are not available on existing AOD, has been backported from 100X (*). Changed files are:

PhysicsTools/PatAlgos/plugins/PATTauProducer.cc
PhysicsTools/PatAlgos/plugins/PATTauProducer.h
PhysicsTools/PatAlgos/python/producersLayer1/tauProducer_cfi.py

The development has been made by @mbluj.

The changes wrt to status quo are in the mvaTauID output only. Performance comparisons and more detailed comments can be checked from here (**)

Please consider this PR for reminiAODv2.

Cheers,
Roger

(*)
#21186

(**)
cms-tau-pog#71

(cherry picked from commit 75e3b6a)

(cherry picked from commit f409fc9)

(cherry picked from commit 48c0759)

(cherry picked from commit b2d6c67)

(cherry picked from commit 4feb9c9)

…the event (cherry picked from commit 31cdc5e)

(cherry picked from commit cc795b6)

…but only at first occurence per instance (cherry picked from commit 01b008a)

… for reMiniAOD with 94X (X>3)

Modified MVAIso tau-id 2017v1 for miniAODv2 with 94X (X>3)

cmsbuild · 2018-01-26T16:55:27Z

A new Pull Request was created by @roger-wolf (Roger Wolf) for CMSSW_9_4_X.

It involves the following packages:

PhysicsTools/PatAlgos
RecoTauTag/Configuration

@perrotta, @cmsbuild, @monttj, @slava77 can you please review it and eventually sign? Thanks.
@TaiSakuma, @gouskos, @imarches, @ahinzmann, @acaudron, @mmarionncern, @rappoccio, @jdamgov, @jdolen, @nhanvtran, @gpetruc, @gkasieczka, @schoef, @ferencek, @mverzett, @mariadalfonso, @pvmulder, @seemasharmafnal, @JyothsnaKomaragiri this is something you requested to watch as well.
@davidlange6, @slava77, @fabiocos you are the release manager for this.

cms-bot commands are listed here

slava77 · 2018-01-26T17:15:30Z

@roger-wolf
is everything introduced in this PR already available in the master branch?

slava77 · 2018-01-26T17:15:43Z

@cmsbuild please test

cmsbuild · 2018-01-26T17:15:46Z

The tests are being triggered in jenkins.
https://cmssdt.cern.ch/jenkins/job/ib-any-integration/25653/console Started: 2018/01/26 20:24

slava77 · 2018-01-26T19:34:34Z

RecoTauTag/Configuration/python/updateHPSPFTaus.py

@@ -0,0 +1,53 @@
+import FWCore.ParameterSet.Config as cms


the name of this file requires some disambiguation.
There is also updateHPSPFTaus_cff with completely different content.

perhaps customiseHPSPFTausForPAT ?

I've followed your proposals.
Cheers,
Roger

slava77 · 2018-01-26T19:39:22Z

RecoTauTag/Configuration/python/updateHPSPFTaus.py

+    process.hpsPFTauDiscriminationByVVTightIsolationMVArun2v1DBoldDMwLTMVAIsoFor94XMiniv2.mapping[0].cut = cms.string("RecoTauTag_tauIdMVAIsoDBoldDMwLT2017v1_WPEff40")    
+    #create a new task and put the moduled therin
+    process.hpsPFTauMVAIsoFor94XMiniv2Task = cms.Task()
+    process.hpsPFTauMVAIsoFor94XMiniv2Task.add(process.hpsPFTauDiscriminationByIsolationMVArun2v1DBoldDMwLTrawMVAIsoFor94XMiniv2)


it would be less verbose to simply pass all taggers in the Task constructor arguments.

roger-wolf · 2018-01-26T22:18:29Z

Hi Slava,

yes, everything that is commited here is also part of the master branch, with the exception of the 94X legacy processing specific parts of course. For more info I refer to @mbluj . I'll up date according to your comments tomorrow, as soon as I get to it.

Cheers,
Roger

slava77 · 2018-01-26T22:35:16Z

After some thought I think that we should make a commitment to support miniAOD processing using inputs from 94X Fall17 RECO/AOD production campaign.
So, this means an introduction of a modifier similar to run2_miniAOD_80XLegacy.
I propose to call it run2_miniAOD_94XFall17.
Since tau is currently the only customer, Roger, may I ask that you create this modifier.
Given this modifier we can then follow this:

in master:
- create run2_miniAOD_94XFall17
- define hpsPFTauIsolationMVArun2v1DBoldDMwLTTask in HPSPFTaus_cff.py with the 8 needed modules
  - use this task in HPSPFTaus_cff.py hpsPFTauMVAIsolation2Task
- append hpsPFTauMVAIsoTask to makePatTausTask in miniAOD_tools via run2_miniAOD_94XFall17
- .. process.patTaus.tauIDSources remains unchanged
- .. modules entering hpsPFTauMVAIsoTask remain unchanged
in 94X (almost the same, but with minor modifications)
- create run2_miniAOD_94XFall17
- define hpsPFTauIsolationMVArun2v1DBoldDMwLTTask in in HPSPFTaus_cff.py with 7 needed modules
  - add hpsPFTauDiscriminationByVVLooseIsolationMVArun2v1DBoldDMwLT only for this era
  - use this task in HPSPFTaus_cff.py hpsPFTauMVAIsolation2Task
- append hpsPFTauMVAIsoTask to makePatTausTask in miniAOD_tools via run2_miniAOD_94XFall17
- process.patTaus.tauIDSources remains unchanged
- modules entering hpsPFTauMVAIsoTask are modified via run2_miniAOD_94XFall17 to pick up the right payload values

let me know if this is clear enough or if we should discuss more.

cmsbuild · 2018-02-21T14:13:23Z

Pull request #21977 was updated. @perrotta, @monttj, @cmsbuild, @franzoni, @slava77, @fabiocos, @davidlange6 can you please check and sign again.

slava77 · 2018-02-21T14:18:57Z

@cmsbuild please test

cmsbuild · 2018-02-21T14:19:17Z

The tests are being triggered in jenkins.
https://cmssdt.cern.ch/jenkins/job/ib-any-integration/26211/console Started: 2018/02/21 16:30

cmsbuild · 2018-02-21T17:27:09Z

+1
Tested at: e30e6ef
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-21977/26211/summary.html

The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic:
4554941
b5f1c48
3f87478
You can see more details here:
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-21977/26211/git-log-recent-commits
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-21977/26211/git-merge-result

cmsbuild · 2018-02-21T17:27:12Z

Comparison job queued.

cmsbuild · 2018-02-21T19:21:06Z

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-21977/26211/summary.html

Comparison Summary:

No significant changes to the logs found
Reco comparison results: 117 differences found in the comparisons
DQMHistoTests: Total files compared: 26
DQMHistoTests: Total histograms compared: 2718773
DQMHistoTests: Total failures: 107
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 2718504
DQMHistoTests: Total skipped: 162
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0 KiB( 0 files compared)
Checked 107 log files, 9 edm output root files, 26 DQM output files

roger-wolf · 2018-02-26T09:43:15Z

Dear colleagues,

let me know whethere there is anything else pending to get this PR signed.

Cheers,
Roger

slava77 · 2018-02-27T14:35:52Z

+1

for #21977 e30e6ef

backport of Rerunning MVA based tauIDs for "run2_miniAOD_94XFall17" and "run2_miniAOD_80XLegacy" eras + one additional fix in boostedTau reconstruction for miniAOD #22105 restricted by modifiers, Adaptation to enable the use of a new MVAIsolation training #21022 restricted by modifiers, and partly Adding PFClusterIsolation to reco Egamma objects #21811 (the run2_miniAOD_94XFall17 modifier definition)
jenkins tests pass and comparisons with baseline show differences only in the run2_miniAOD_80XLegacy-driven workflow 136.7611. The changes are in slimmedTausBoosted count/kinematics and IDs and in slimmedTaus IDs. The changes in the IDs are in the expected range of ID variables.

slava77 · 2018-03-05T15:15:44Z

@fabiocos
I wanted to check if this can be merged in 94MAODX.
Please clarify.
Thank you.

fabiocos · 2018-03-05T16:24:57Z

+operations

fabiocos · 2018-03-05T16:27:00Z

+1

fabiocos · 2018-03-05T16:27:39Z

merge

gpetruc · 2018-03-19T13:14:18Z

Do I understand correctly that this PR changes the MVArun2v1DBoldDMwLT set of discriminators, rather than adding a new set?

slava77 · 2018-03-19T13:23:03Z

Do I understand correctly that this PR changes the MVArun2v1DBoldDMwLT set of discriminators, rather than adding a new set?

One VVL discriminator is added; the rest are indeed modified.

mbluj · 2018-03-19T16:13:13Z

On Mon, Mar 19, 2018 at 2:23 PM, Slava Krutelyov ***@***.***> wrote: Do I understand correctly that this PR changes the MVArun2v1DBoldDMwLT set of discriminators, rather than adding a new set? One VVL discriminator is added; the rest are indeed modified.

Yes, I confirm. It is what Tau POG wanted. Michał

…

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#21977 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AE-odTQs-snZMxJ2RFyPw18tqamiS8hVks5tf7E6gaJpZM4RulK1> .

-- Michał Bluj National Centre for Nuclear Research / Narodowe Centrum Badań Jądrowych Hoża 69 00-681 Warszawa, Poland tel. (+48) 22 55 31 833 e-mail: [email protected]

mbluj and others added 12 commits January 25, 2018 15:36

Add new payloads for MVAIso 2017v1

2acbc55

(cherry picked from commit 75e3b6a)

bugfix

7b92b26

(cherry picked from commit f409fc9)

Add protection against a missing tauId source(s)

0ed49f8

(cherry picked from commit 48c0759)

use nullptr rather than 0 for a null pointer

1c6de2e

(cherry picked from commit b2d6c67)

Add protection againist missing (calo) tauId sources

d226a9e

(cherry picked from commit 4feb9c9)

add a switch allowing/disallowing skipping missing tauId products in …

143a4df

…the event (cherry picked from commit 31cdc5e)

code-checks

b0948e5

(cherry picked from commit cc795b6)

changing LogWarning printout such that it is not raised for each tau …

72ca8c5

…but only at first occurence per instance (cherry picked from commit 01b008a)

Add function modyfing makePatTausTask to include MVAIso tau-Id 2017v1…

6b2d8cc

… for reMiniAOD with 94X (X>3)

fix input parameters

a997d05

Add MVAIso tau-Id with 2017v1 training to miniAOD sequences

e8f62d5

Merge pull request #71 from mbluj/CMSSW_9_4_X_tauIdForMiniAODv2

3f87478

Modified MVAIso tau-id 2017v1 for miniAODv2 with 94X (X>3)

cmsbuild added this to the CMSSW_9_4_X milestone Jan 26, 2018

cmsbuild added analysis-pending comparison-pending orp-pending pending-signatures reconstruction-pending tests-pending labels Jan 26, 2018

mbluj mentioned this pull request Jan 26, 2018

features needed for 94X re-miniAOD #21864

Closed

15 tasks

cmsbuild added tests-started and removed tests-pending labels Jan 26, 2018

slava77 reviewed Jan 26, 2018

View reviewed changes

cmsbuild added tests-started and removed tests-pending labels Feb 21, 2018

cmsbuild added tests-approved and removed tests-started labels Feb 21, 2018

cmsbuild added comparison-available and removed comparison-pending labels Feb 21, 2018

slava77 mentioned this pull request Feb 26, 2018

re-miniAOD workflows on 94X input datasets (94MAOD) #22343

Merged

cmsbuild added reconstruction-approved and removed reconstruction-pending labels Feb 27, 2018

cmsbuild added operations-approved and removed operations-pending labels Mar 5, 2018

cmsbuild added orp-approved and removed orp-pending labels Mar 5, 2018

cmsbuild merged commit 7c9aed8 into cms-sw:CMSSW_9_4_MAOD_X Mar 5, 2018

mbluj deleted the CMSSW_9_4_X_tau-pog_newTauID-MCv1 branch October 10, 2023 10:05

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

New trainings for mvaTauID + one backport from CMSSW_10_0_X #21977

New trainings for mvaTauID + one backport from CMSSW_10_0_X #21977

roger-wolf commented Jan 26, 2018

cmsbuild commented Jan 26, 2018

slava77 commented Jan 26, 2018

slava77 commented Jan 26, 2018

cmsbuild commented Jan 26, 2018 •

edited

Loading

slava77 Jan 26, 2018

roger-wolf Jan 29, 2018

slava77 Jan 26, 2018

roger-wolf commented Jan 26, 2018

slava77 commented Jan 26, 2018

cmsbuild commented Feb 21, 2018

slava77 commented Feb 21, 2018

cmsbuild commented Feb 21, 2018 •

edited

Loading

cmsbuild commented Feb 21, 2018

cmsbuild commented Feb 21, 2018

cmsbuild commented Feb 21, 2018

roger-wolf commented Feb 26, 2018

slava77 commented Feb 27, 2018

slava77 commented Mar 5, 2018

fabiocos commented Mar 5, 2018

fabiocos commented Mar 5, 2018

fabiocos commented Mar 5, 2018

gpetruc commented Mar 19, 2018

slava77 commented Mar 19, 2018

mbluj commented Mar 19, 2018 via email

New trainings for mvaTauID + one backport from CMSSW_10_0_X #21977

New trainings for mvaTauID + one backport from CMSSW_10_0_X #21977

Conversation

roger-wolf commented Jan 26, 2018

cmsbuild commented Jan 26, 2018

slava77 commented Jan 26, 2018

slava77 commented Jan 26, 2018

cmsbuild commented Jan 26, 2018 • edited Loading

slava77 Jan 26, 2018

Choose a reason for hiding this comment

roger-wolf Jan 29, 2018

Choose a reason for hiding this comment

slava77 Jan 26, 2018

Choose a reason for hiding this comment

roger-wolf commented Jan 26, 2018

slava77 commented Jan 26, 2018

cmsbuild commented Feb 21, 2018

slava77 commented Feb 21, 2018

cmsbuild commented Feb 21, 2018 • edited Loading

cmsbuild commented Feb 21, 2018

cmsbuild commented Feb 21, 2018

cmsbuild commented Feb 21, 2018

roger-wolf commented Feb 26, 2018

slava77 commented Feb 27, 2018

slava77 commented Mar 5, 2018

fabiocos commented Mar 5, 2018

fabiocos commented Mar 5, 2018

fabiocos commented Mar 5, 2018

gpetruc commented Mar 19, 2018

slava77 commented Mar 19, 2018

mbluj commented Mar 19, 2018 via email

cmsbuild commented Jan 26, 2018 •

edited

Loading

cmsbuild commented Feb 21, 2018 •

edited

Loading