add vertex trimming to the phase2 hlt tracking iterations #47113

lguzzi · 2025-01-16T10:25:44Z

PR description:

This PR adds vertex trimming to the phase2 hlt menu with a proc. modifier. This modifier allows to select and build only initial step seeds and high-pt triplet step doublets compatible with a subset of pixel vertices.
Preliminary studies of this configuration are available here for the legacy iterations.

PR validation:

This PR has been validated comparing the results of the customisation function used so far with the proc. modifier approach. Comparison plots show no difference.

If this PR is a backport please specify the original PR and why you need to backport that PR. If this PR will be backported please specify to which release cycle the backport is meant for:

not needed

cmsbuild · 2025-01-16T10:26:06Z

cms-bot internal usage

cmsbuild · 2025-01-16T10:27:10Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47113/43320

cmsbuild · 2025-01-16T10:27:36Z

A new Pull Request was created by @lguzzi for master.

It involves the following packages:

Configuration/ProcessModifiers (operations)
HLTrigger/Configuration (hlt)

@Martin-Grunewald, @antoniovilela, @cmsbuild, @davidlange6, @fabiocos, @mandrenguyen, @mmusich, @rappoccio can you please review it and eventually sign? Thanks.
@Martin-Grunewald, @SohamBhattacharya, @VourMa, @fabiocos, @makortel, @missirol, @mmusich, @rovere this is something you requested to watch as well.
@antoniovilela, @mandrenguyen, @rappoccio, @sextonkennedy you are the release manager for this.

cms-bot commands are listed here

mmusich · 2025-01-16T10:27:59Z

Configuration/ProcessModifiers/python/vertexTrimming_cff.py

+
+# This modifier is used to enable vertex trimming in the tracking sequences
+# and in all related reconstruction modules
+vertexTrimming = cms.Modifier()


can it be renamed to signal it's about the phase-2 HLT?

I renamed it to phase2_hlt_vertexTrimming in c05a4f2

mmusich · 2025-01-16T10:28:12Z

...ration/python/HLT_75e33/modules/hltHighPtTripletStepTrackingRegionFromTrimmedVertices_cfi.py

+    scalingStartNPix        = cms.double(0.0),
+    scalingEndNPix          = cms.double(1.0),
+  ),
+  mightGet = cms.optional.untracked.vstring,


Suggested change

mightGet = cms.optional.untracked.vstring,

unneeded?

done d17f0fb

mmusich · 2025-01-16T10:28:27Z

...ration/python/HLT_75e33/modules/hltHighPtTripletStepTrackingRegionFromTrimmedVertices_cfi.py

+    useFakeVertices         = cms.bool(False),
+    maxNVertices            = cms.int32(-1),
+    nSigmaZ                 = cms.double(4.0),
+    pixelClustersForScaling = cms.InputTag('siPixelClusters'),


this is an offline product?

yes, I copied it from the defaults. It is not used in this configuration, however I changed it to hltSiPixelClusters in d17f0fb

mmusich · 2025-01-16T10:30:17Z

test parameters:

relvals_opt = --what upgrade
workflows = 29634.753, 29634.754, 29634.755, 29634.756, 29634.757
enable = hlt_p2_integration, hlt_p2_timing

mmusich · 2025-01-16T10:32:35Z

@lguzzi please add workflows that can exercise the trimming.
You can take inspiration from the other variants:

cmssw/Configuration/PyReleaseValidation/python/upgradeWorkflowComponents.py

Lines 1647 to 1685 in 547258c

    
           upgradeWFs['HLTTiming75e33Alpaka'] = deepcopy(upgradeWFs['HLTTiming75e33']) 
        
           upgradeWFs['HLTTiming75e33Alpaka'].suffix = '_HLT75e33TimingAlpaka' 
        
           upgradeWFs['HLTTiming75e33Alpaka'].offset = 0.751 
        
           upgradeWFs['HLTTiming75e33Alpaka'].step2 = { 
        
               '-s':'DIGI:pdigi_valid,L1TrackTrigger,L1,L1P2GT,DIGI2RAW,HLT:75e33_timing', 
        
               '--procModifiers': 'alpaka' 
        
           } 
        
           upgradeWFs['HLTTiming75e33TiclV5'] = deepcopy(upgradeWFs['HLTTiming75e33']) 
        
           upgradeWFs['HLTTiming75e33TiclV5'].suffix = '_HLT75e33TimingTiclV5' 
        
           upgradeWFs['HLTTiming75e33TiclV5'].offset = 0.752 
        
           upgradeWFs['HLTTiming75e33TiclV5'].step2 = { 
        
               '-s':'DIGI:pdigi_valid,L1TrackTrigger,L1,L1P2GT,DIGI2RAW,HLT:75e33_timing', 
        
               '--procModifiers': 'ticl_v5' 
        
           } 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIter'] = deepcopy(upgradeWFs['HLTTiming75e33']) 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIter'].suffix = '_HLT75e33TimingAlpakaSingleIter' 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIter'].offset = 0.753 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIter'].step2 = { 
        
               '-s':'DIGI:pdigi_valid,L1TrackTrigger,L1,L1P2GT,DIGI2RAW,HLT:75e33_timing', 
        
               '--procModifiers': 'alpaka,singleIterPatatrack' 
        
           } 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIterLST'] = deepcopy(upgradeWFs['HLTTiming75e33']) 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIterLST'].suffix = '_HLT75e33TimingAlpakaSingleIterLST' 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIterLST'].offset = 0.754 
        
           upgradeWFs['HLTTiming75e33AlpakaSingleIterLST'].step2 = { 
        
               '-s':'DIGI:pdigi_valid,L1TrackTrigger,L1,L1P2GT,DIGI2RAW,HLT:75e33_timing', 
        
               '--procModifiers': 'alpaka,singleIterPatatrack,trackingLST' 
        
           } 
        
           upgradeWFs['HLTTiming75e33AlpakaLST'] = deepcopy(upgradeWFs['HLTTiming75e33']) 
        
           upgradeWFs['HLTTiming75e33AlpakaLST'].suffix = '_HLT75e33TimingAlpakaLST' 
        
           upgradeWFs['HLTTiming75e33AlpakaLST'].offset = 0.755 
        
           upgradeWFs['HLTTiming75e33AlpakaLST'].step2 = { 
        
               '-s':'DIGI:pdigi_valid,L1TrackTrigger,L1,L1P2GT,DIGI2RAW,HLT:75e33_timing', 
        
               '--procModifiers': 'alpaka,trackingLST' 
        
           }

cmsbuild · 2025-01-16T10:43:14Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47113/43321

Found files with invalid states:
- Configuration/ProcessModifiers/python/vertexTrimming_cff.py:
  - Added: 8c2f00a
  - Deleted: c05a4f2

cmsbuild · 2025-01-16T10:43:39Z

Pull request #47113 was updated. @Martin-Grunewald, @antoniovilela, @cmsbuild, @davidlange6, @fabiocos, @mandrenguyen, @mmusich, @rappoccio can you please check and sign again.

mmusich · 2025-01-16T10:50:30Z

Configuration/ProcessModifiers/python/phase2_hlt_vertexTrimming_cff.py

+# This modifier is used to enable vertex trimming in the tracking sequences
+# and in all related reconstruction modules
+phase2_hlt_vertexTrimming = cms.Modifier()
+


this line is uneeded @lguzzi. Anyway please rebase to have just a single commit (also add the requested workflows).

cmsbuild · 2025-01-16T10:51:08Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47113/43322

Found files with invalid states:
- Configuration/ProcessModifiers/python/vertexTrimming_cff.py:
  - Added: 8c2f00a
  - Deleted: c05a4f2

cmsbuild · 2025-01-16T10:51:31Z

Pull request #47113 was updated. @Martin-Grunewald, @antoniovilela, @cmsbuild, @davidlange6, @fabiocos, @mandrenguyen, @mmusich, @rappoccio can you please check and sign again.

lguzzi · 2025-01-16T13:28:08Z

I somehow assumed the trimming logic would have been coupled with the recovery of the tracks coming from the excluded vertices by properly merging the trimmed-full tracks and the pixel tracks from the excluded vertices. Last time I have checked, that is what we have documented in our documentation site here, but I fail to see that logic implemented here.

I left out the part written by JME because I cannot validate it. If you think it is appropriate I can add the merged track collection producer, then JME can add what is left for puppi in a separate PR.

rovere · 2025-01-16T13:34:11Z

I somehow assumed the trimming logic would have been coupled with the recovery of the tracks coming from the excluded vertices by properly merging the trimmed-full tracks and the pixel tracks from the excluded vertices. Last time I have checked, that is what we have documented in our documentation site here, but I fail to see that logic implemented here.

I left out the part written by JME because I cannot validate it. If you think it is appropriate I can add the merged track collection producer, then JME can add what is left for puppi in a separate PR.

It will be a nice topic for discussion at the next HLT Upgrade meeting!

cmsbuild · 2025-01-16T13:53:09Z

+1

Size: This PR adds an extra 68KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-ee3df1/43792/summary.html
COMMIT: 5511128
CMSSW: CMSSW_15_0_X_2025-01-16-1100/el8_amd64_gcc12
Additional Tests: HLT_P2_INTEGRATION,HLT_P2_TIMING
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/47113/43792/install.sh to create a dev area with all the needed externals and cmssw changes.

HLT P2 Timing: chart

Comparison Summary

Summary:

You potentially added 2 lines to the logs
Reco comparison results: 0 differences found in the comparisons
DQMHistoTests: Total files compared: 49
DQMHistoTests: Total histograms compared: 3819085
DQMHistoTests: Total failures: 94
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 3818971
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 48 files compared)
Checked 214 log files, 184 edm output root files, 49 DQM output files
TriggerResults: no differences found

mmusich · 2025-01-16T14:07:16Z

It will be a nice topic for discussion at the next HLT Upgrade meeting!

@rovere @VourMa I gather we should put this on hold until it's clarified what's the way forward at the first upgrade meeting?
@lguzzi in the meanwhile it would be useful to address the LST scheduling issue following the suggestion at #47113 (comment)

mmasciov · 2025-01-16T14:26:47Z

FYI @cms-sw/tracking-pog-l2

mmasciov · 2025-01-16T15:01:13Z

It will be a nice topic for discussion at the next HLT Upgrade meeting!

@rovere @VourMa I gather we should put this on hold until it's clarified what's the way forward at the first upgrade meeting? @lguzzi in the meanwhile it would be useful to address the LST scheduling issue following the suggestion at #47113 (comment)

In general, for @cms-sw/tracking-pog-l2 it would be easier to indeed have the mixedGeneralTracks approach also reviewable (e.g., in a PR, or in a Tracking POG meeting). This could be done by creating a PR to this branch by JME?
For instance, it appears to me from the private cmssw fork that is currently used for the customization that the same cuts used for full tracks are used to set the 'high-purity' for the 'extra' pixel-only tracks, which may be suboptimal. On a related topic, as mentioned at the last HLT upgrade meeting, it would be great to define the needs from other POGs also in terms of vertex reconstruction.
While I agree that a follow-up report is better (instead of attached minutes and my verbal report that was likely not clear enough), I would encourage that tracking matters are reviewed within the Tracking POG too.

lguzzi · 2025-01-16T16:55:48Z

I added the workflows for trimming alone and with alpaka. Trimming with LST does not work at the moment (ScheduleExecutionFailure).

The problem is that, for LST, HLTHighPtTripletStepSeedingSequence needs to run early, so that the HighPtTripletSeeds are available as input to LST building

cmssw/HLTrigger/Configuration/python/HLT_75e33/sequences/HLTInitialStepSequence_cfi.py

Lines 25 to 42 in 547258c

_HLTInitialStepSequenceLST = cms.Sequence(

hltInitialStepSeeds

+hltInitialStepSeedTracksLST

+HLTHighPtTripletStepSeedingSequence

+hltHighPtTripletStepSeedTracksLST

+hltPixelSeedInputLST

+hltSiPhase2RecHits # Probably need to move elsewhere in the final setup

+hltPhase2OTHitsInputLST # Probably need to move elsewhere in the final setup

+hltLST

+hltInitialStepTrackCandidates

+hltInitialStepTrackspTTCLST

+hltInitialStepTrackspLSTCLST

+hltInitialStepTracksT5TCLST

+hltInitialStepTrackCutClassifierpTTCLST

+hltInitialStepTrackCutClassifierpLSTCLST

+hltInitialStepTrackSelectionHighPuritypTTCLST

+hltInitialStepTrackSelectionHighPuritypLSTCLST

)

For trimming, you modify hltHighPtTripletStepHitDoublets, which are part of HLTHighPtTripletStepSeedingSequence

cmssw/HLTrigger/Configuration/python/HLT_75e33/sequences/HLTHighPtTripletStepSeedingSequence_cfi.py

Line 9 in 547258c

HLTHighPtTripletStepSeedingSequence = cms.Sequence(hltHighPtTripletStepClusters+hltHighPtTripletStepSeedLayers+hltHighPtTripletStepHitDoublets+hltHighPtTripletStepHitTriplets+hltHighPtTripletStepSeeds)

and you schedule this modification at the beginning of HLTHighPtTripletStepSequence

cmssw/HLTrigger/Configuration/python/HLT_75e33/sequences/HLTHighPtTripletStepSequence_cfi.py

Line 37 in 5511128

_HLTHighPtTripletStepSequenceTrimming.insert(0, hltTrackingRegionFromTrimmedVertices)

However, in the LST configuration, HLTHighPtTripletStepSeedingSequence has already run when you ask for its modification, leading to the error. I believe that you could fix this issue by adding the modification for LST before this line:

cmssw/HLTrigger/Configuration/python/HLT_75e33/sequences/HLTInitialStepSequence_cfi.py

Line 28 in 547258c

+HLTHighPtTripletStepSeedingSequence

thank you, I moved the tracking region definition to HLTHighPtTripletStepSeedingSequence and added some LST workflows.

cmsbuild · 2025-01-16T16:56:09Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47113/43331

There are other open Pull requests which might conflict with changes you have proposed:
- File Configuration/PyReleaseValidation/python/upgradeWorkflowComponents.py modified in PR(s): [NGT] Implement NanoAOD ntuples production for HLT Muon objects #47069

cmsbuild · 2025-01-16T16:56:14Z

Pull request #47113 was updated. @AdrianoDee, @Martin-Grunewald, @Moanwar, @antoniovilela, @cmsbuild, @davidlange6, @DickyChant, @fabiocos, @mandrenguyen, @miquork, @mmusich, @rappoccio, @srimanob, @subirsarkar can you please check and sign again.

cmsbuild · 2025-01-16T17:04:14Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47113/43332

There are other open Pull requests which might conflict with changes you have proposed:
- File Configuration/PyReleaseValidation/README.md modified in PR(s): [NGT] Implement NanoAOD ntuples production for HLT Muon objects #47069
- File Configuration/PyReleaseValidation/python/upgradeWorkflowComponents.py modified in PR(s): [NGT] Implement NanoAOD ntuples production for HLT Muon objects #47069

cmsbuild · 2025-01-16T17:04:36Z

Pull request #47113 was updated. @AdrianoDee, @Martin-Grunewald, @Moanwar, @antoniovilela, @cmsbuild, @davidlange6, @DickyChant, @fabiocos, @mandrenguyen, @miquork, @mmusich, @rappoccio, @srimanob, @subirsarkar can you please check and sign again.

cmsbuild added this to the CMSSW_15_0_X milestone Jan 16, 2025

cmsbuild added hlt-pending operations-pending pending-signatures tests-pending orp-pending code-checks-pending labels Jan 16, 2025