GNN Track-Tracksters Linking in TICLv5 #49652

Moanwar · 2025-12-17T09:44:48Z

This PR introduces a new GNN-based linking method between tracks and tracksters within TICLv5. The method builds a graph consisting of tracks and their surrounding tracksters, and then uses a Graph Neural Network (GNN) to evaluate whether each track-trackster edge is valid. Tracksters that are predicted to be linked to a given track are then stored accordingly.

In the current implementation, the linking selection is prioritized using the track pT and the DeltaR between the track and tracksters. This strategy may be refined in the future for further optimization.

A new procmodifier has been added to enable this method in workflows *.211. The process modifier is named ticlv5_TrackLinkingGNN.

This PR should be tested with TICL workflows such as 29888.211 and 29688.211, using the GNN models uploaded here:
cms-data/RecoHGCal-TICL#11

Adding : @felicepantaleo @hatakeyamak @waredjeb @AuroraPerego

cmsbuild · 2025-12-17T09:45:14Z

cms-bot internal usage

Moanwar · 2025-12-17T09:45:39Z

test parameters:
pull_request = cms-data/RecoHGCal-TICL#11
workflow_opts = -w upgrade
workflows = 29888.211,29688.211

Moanwar · 2025-12-17T09:46:28Z

please test

cmsbuild · 2025-12-17T09:47:10Z

-code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-49652/47196

There are other open Pull requests which might conflict with changes you have proposed:
- File RecoHGCal/TICL/interface/TICLInterpretationAlgoBase.h modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/plugins/TICLCandidateProducer.cc modified in PR(s): Add directories for TICL-EGM code development #48319

Code check has found code style and quality issues which could be resolved by applying following patch(s)

code-format:
https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-49652/47196/code-format.patch
e.g. curl -k https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-49652/47196/code-format.patch | patch -p1
You can also run scram build code-format to apply code format directly

Moanwar · 2025-12-17T09:55:21Z

@cmsbuild, please test

cmsbuild · 2025-12-17T10:05:11Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-49652/47197

There are other open Pull requests which might conflict with changes you have proposed:
- File RecoHGCal/TICL/interface/TICLInterpretationAlgoBase.h modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/plugins/TICLCandidateProducer.cc modified in PR(s): Add directories for TICL-EGM code development #48319

mmusich · 2025-12-18T20:12:21Z

@cmsbuild please abort

Moanwar · 2025-12-18T20:16:01Z

@Moanwar

I added workflow .7521 for HLT checks in the ph2_hlt matrix

The changes you applied are insufficient for automatic redirection of the workflow number in the short matrix. Please add

diff --git a/Configuration/PyReleaseValidation/python/relval_Run4.py b/Configuration/PyReleaseValidation/python/relval_Run4.py
index 58c8f7e0dac..8559ff5325c 100644
--- a/Configuration/PyReleaseValidation/python/relval_Run4.py
+++ b/Configuration/PyReleaseValidation/python/relval_Run4.py
@@ -74,6 +74,7 @@ numWFIB.extend([prefixDet+34.7501])# HLTTrackingOnly75e33
 numWFIB.extend([prefixDet+34.751]) # HLTTiming75e33, alpaka
 numWFIB.extend([prefixDet+34.7511])# HLTTiming75e33, phase2CAExtension
 numWFIB.extend([prefixDet+34.752]) # HLTTiming75e33, ticl_v5
+numWFIB.extend([prefixDet+34.7521]) # HLTTiming75e33, ticl_v5, ticlv5TrackLinkingGNN
 numWFIB.extend([prefixDet+34.753]) # HLTTiming75e33, alpaka,singleIterPatatrack
 numWFIB.extend([prefixDet+34.754]) # HLTTiming75e33, alpaka,singleIterPatatrack,trackingLST
 numWFIB.extend([prefixDet+34.755]) # HLTTiming75e33, alpaka,trackingLST

Sure, no problem-thanks for pointing this out!

cmsbuild · 2025-12-18T20:24:44Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-49652/47228

There are other open Pull requests which might conflict with changes you have proposed:
- File Configuration/PyReleaseValidation/scripts/runTheMatrix.py modified in PR(s): [Do not merge] For cms-bot PR testing #48637
- File RecoHGCal/TICL/interface/TICLInterpretationAlgoBase.h modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/plugins/TICLCandidateProducer.cc modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/plugins/TICLGraph.h modified in PR(s): Add directories for TICL-EGM code development #48319

cmsbuild · 2025-12-18T20:25:04Z

Pull request #49652 was updated. @AdrianoDee, @DickyChant, @Martin-Grunewald, @Moanwar, @antoniovagnerini, @cmsbuild, @davidlange6, @fabiocos, @ftenchini, @jfernan2, @mandrenguyen, @miquork, @mmusich, @srimanob can you please check and sign again.

mmusich · 2025-12-18T20:24:40Z

RecoHGCal/TICL/python/iterativeTICL_cff.py

+            inputNames = cms.vstring('x', 'edge_index', 'edge_attr'),
+            output = cms.vstring('output'),
+            delta_tk_ts = cms.double(0.1),
+	    thr_gnn = cms.double(0.5),


please fix the indentation.

Moanwar · 2025-12-18T21:09:03Z

HLTrigger/Configuration/python/HLT_75e33/modules/hltTiclCandidate_cfi.py

+        output = cms.vstring('output'),
+        delta_tk_ts = cms.double(0.1),
+        thr_gnn = cms.double(0.5),
+        type = cms.string('GNNLink')


The choice of the name "type" is intentional to be the same, as it determines whether the old plugin or the new GNN-based plugin is used. However, I’m wondering if this should compile. I tried quickly and got the following error:

TypeError: type does not already exist, so it can only be set to a CMS Python configuration type

Did it compile successfully on your side?

It did for me. @Moanwar you have to replace cms.PSet with dict first -- of course.

Hmm, that’s strange. I have tried compiling and running a few times to make sure it works:
runTheMatrix.py -w upgrade -l 29688.211 -j 0
But each time, I get the same error I mentioned above. I’m not sure I understand why this happens, could you please give that a try ? Thanks!

@Moanwar

cmsrel CMSSW_16_0_X_2025-12-17-2300 cd CMSSW_16_0_X_2025-12-17-2300/src/ cmsenv git cms-merge-topic 49652 # apply patch below [*] scram b -j 20 runTheMatrix.py -w upgrade -l 29688.211 -j 0

runs fine for me.
hope this helps!

[*]

diff --git a/HLTrigger/Configuration/python/HLT_75e33/modules/hltTiclCandidate_cfi.py b/HLTrigger/Configuration/python/HLT_75e33/modules/hltTiclCandidate_cfi.py index 307af675b8e..03efda78c33 100644 --- a/HLTrigger/Configuration/python/HLT_75e33/modules/hltTiclCandidate_cfi.py +++ b/HLTrigger/Configuration/python/HLT_75e33/modules/hltTiclCandidate_cfi.py @@ -49,13 +49,13 @@ hltTiclCandidate = cms.EDProducer("TICLCandidateProducer", from Configuration.ProcessModifiers.ticlv5_TrackLinkingGNN_cff import ticlv5TrackLinkingGNN ticlv5TrackLinkingGNN.toModify(hltTiclCandidate, - interpretationDescPSet = cms.PSet( + interpretationDescPSet = dict( onnxTrkLinkingModelFirstDisk = cms.FileInPath('RecoHGCal/TICL/data/ticlv5/onnx_models/TrackLinking_GNN/FirstDiskPropGNN_v0.onnx'), onnxTrkLinkingModelInterfaceDisk = cms.FileInPath('RecoHGCal/TICL/data/ticlv5/onnx_models/TrackLinking_GNN/InterfaceDiskPropGNN_v0.onnx'), inputNames = cms.vstring('x', 'edge_index', 'edge_attr'), output = cms.vstring('output'), delta_tk_ts = cms.double(0.1), thr_gnn = cms.double(0.5), - type = cms.string('GNNLink') + type = 'GNNLink' ) )

Ah, thanks a lot @mmusich, That makes sense. I was testing without changing to a dict, which does compile successfully, but then it raises the other error mentioned in cmsRun here: - unless it’s working for you as well?

RecoHGCal/TICL/python/iterativeTICL_cff.py

Moanwar · 2025-12-18T21:09:03Z

HLTrigger/Configuration/python/HLT_75e33/modules/hltTiclCandidate_cfi.py

+        output = cms.vstring('output'),
+        delta_tk_ts = cms.double(0.1),
+        thr_gnn = cms.double(0.5),
+        type = cms.string('GNNLink')


The choice of the name "type" is intentional to be the same, as it determines whether the old plugin or the new GNN-based plugin is used. However, I’m wondering if this should compile. I tried quickly and got the following error:

TypeError: type does not already exist, so it can only be set to a CMS Python configuration type

Did it compile successfully on your side?

cmsbuild · 2025-12-18T21:14:37Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-49652/47233

There are other open Pull requests which might conflict with changes you have proposed:
- File Configuration/PyReleaseValidation/scripts/runTheMatrix.py modified in PR(s): [Do not merge] For cms-bot PR testing #48637
- File RecoHGCal/TICL/interface/TICLInterpretationAlgoBase.h modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/plugins/TICLCandidateProducer.cc modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/plugins/TICLGraph.h modified in PR(s): Add directories for TICL-EGM code development #48319

cmsbuild · 2025-12-18T21:14:57Z

Pull request #49652 was updated. @AdrianoDee, @DickyChant, @Martin-Grunewald, @Moanwar, @antoniovagnerini, @cmsbuild, @davidlange6, @fabiocos, @ftenchini, @jfernan2, @mandrenguyen, @miquork, @mmusich, @srimanob can you please check and sign again.

mmusich · 2025-12-18T22:51:28Z

@cmsbuild, please test

RecoHGCal/TICL/plugins/GNNInterpretationAlgo.cc

cmsbuild · 2025-12-20T16:54:12Z

+1

Size: This PR adds an extra 76KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-553e9c/50356/summary.html
COMMIT: 777302a
CMSSW: CMSSW_16_0_X_2025-12-18-1100/el8_amd64_gcc13
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/49652/50356/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 19 lines to the logs
ROOTFileChecks: Some differences in event products or their sizes found
Reco comparison results: 9 differences found in the comparisons
Reco comparison had 4 failed jobs
DQMHistoTests: Total files compared: 53
DQMHistoTests: Total histograms compared: 4280393
DQMHistoTests: Total failures: 84
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 4280289
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 52 files compared)
Checked 227 log files, 198 edm output root files, 53 DQM output files
TriggerResults: no differences found

GNN Track-Tracksters Linking

7c1112d

cmsbuild added this to the CMSSW_16_0_X milestone Dec 17, 2025

cmsbuild added reconstruction-pending hlt-pending operations-pending pending-signatures tests-pending orp-pending pdmv-pending code-checks-pending labels Dec 17, 2025

cmsbuild added the requires-external label Dec 17, 2025

cmsbuild added tests-started and removed tests-pending labels Dec 17, 2025

cmsbuild added code-checks-rejected and removed code-checks-pending labels Dec 17, 2025

code format

d9b6b73

cmsbuild added tests-pending code-checks-pending and removed tests-started code-checks-rejected labels Dec 17, 2025

cmsbuild added tests-started and removed tests-pending labels Dec 17, 2025

cmsbuild added code-checks-approved and removed code-checks-pending labels Dec 17, 2025

cmsbuild added tests-pending and removed tests-started labels Dec 18, 2025

adding wf .7521 to the ph2_hlt matrix

6ddbf85

cmsbuild added code-checks-pending and removed code-checks-approved labels Dec 18, 2025

cmsbuild added code-checks-approved and removed code-checks-pending labels Dec 18, 2025

mmusich reviewed Dec 18, 2025

View reviewed changes

Moanwar commented Dec 18, 2025

View reviewed changes

fix the indentation

777302a

cmsbuild added code-checks-pending and removed code-checks-approved labels Dec 18, 2025

cmsbuild added code-checks-approved and removed code-checks-pending labels Dec 18, 2025

cmsbuild added tests-started and removed tests-pending labels Dec 18, 2025

waredjeb reviewed Dec 19, 2025

View reviewed changes

RecoHGCal/TICL/plugins/GNNInterpretationAlgo.cc Show resolved Hide resolved

mmusich mentioned this pull request Dec 19, 2025

Remove cutTk condition from interpretationDescPSet #49680

Merged

cmsbuild added operations-approved tests-approved and removed operations-pending tests-started labels Dec 20, 2025

GNN Track-Tracksters Linking in TICLv5 #49652

Are you sure you want to change the base?

GNN Track-Tracksters Linking in TICLv5 #49652

Uh oh!

Conversation

Moanwar commented Dec 17, 2025

Uh oh!

cmsbuild commented Dec 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Moanwar commented Dec 17, 2025

Uh oh!

Moanwar commented Dec 17, 2025

Uh oh!

cmsbuild commented Dec 17, 2025

Uh oh!

Moanwar commented Dec 17, 2025

Uh oh!

cmsbuild commented Dec 17, 2025

Uh oh!

mmusich commented Dec 18, 2025

Uh oh!

Moanwar commented Dec 18, 2025

Uh oh!

cmsbuild commented Dec 18, 2025

Uh oh!

cmsbuild commented Dec 18, 2025

Uh oh!

mmusich Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Moanwar Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

mmusich Dec 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Moanwar Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

mmusich Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Moanwar Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Moanwar Dec 18, 2025

Choose a reason for hiding this comment

Uh oh!

cmsbuild commented Dec 18, 2025

Uh oh!

cmsbuild commented Dec 18, 2025

Uh oh!

mmusich commented Dec 18, 2025

Uh oh!

Uh oh!

cmsbuild commented Dec 20, 2025

Comparison Summary

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

cmsbuild commented Dec 17, 2025 •

edited

Loading

mmusich Dec 18, 2025 •

edited

Loading