Updated SuperclusterDNN v3 for superclustering in TICLv5 by gsokmen · Pull Request #48811 · cms-sw/cmssw

gsokmen · 2025-08-27T12:02:10Z

This PR introduces a new version (v3) of the SuperclusterDNN to be used for superclustering in TICLv5. This PR includes updates to the previous DNN model as well as the superclustering algorithm. The model and algorithm are used for reconstructing electromagnetic objects within TICL.

-This PR needs to be tested with the following cms-data PR: cms-data/RecoHGCal-TICL#9

-It needs to be tested with the following workflows
. 29646.203
. 29846.203

-Validation plots are attached.
Validation_SuperclusterDNN_v3_4Sept_GSokmen.pdf

@waredjeb @felicepantaleo @sameasy

cmsbuild · 2025-08-27T12:02:36Z

cms-bot internal usage

cmsbuild · 2025-08-27T12:05:14Z

-code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-48811/45919

There are other open Pull requests which might conflict with changes you have proposed:
- File RecoHGCal/TICL/plugins/SuperclusteringSampleDumper.cc modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/plugins/TracksterLinkingbySuperClusteringDNN.cc modified in PR(s): Add directories for TICL-EGM code development #48319
- File RecoHGCal/TICL/src/SuperclusteringDNNInputs.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-48811/45919/code-format.patch
e.g. curl -k https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-48811/45919/code-format.patch | patch -p1
You can also run scram build code-format to apply code format directly

felicepantaleo · 2025-08-27T12:10:09Z

allow @waredjeb test rights

felicepantaleo · 2025-08-27T12:10:41Z

allow @sameasy test rights

felicepantaleo · 2025-08-27T12:10:47Z

type hgcal

felicepantaleo · 2025-08-27T12:11:06Z

test parameters:
pull_request = cms-data/RecoHGCal-TICL#9
workflow_opts = -w upgrade
workflows = 29646.203,29846.203

cmsbuild · 2025-08-27T12:59:48Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-48811/45921

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

waredjeb · 2025-09-29T21:30:20Z

Ok including #49023 the HGCAL differences on ticlCandidates are expected (we produce less candidates because of the tigher filter on LCs). The problem with 771 with differences in EGM tracking are still there. The differences could be expected as we are modifying the SuperClusters but it is not clear why they are seen only in 771 and not in other workflows with TICLv5. I'll continue investigating

waredjeb · 2025-10-06T20:47:05Z

test parameters:

workflow_opts = -w upgrade
workflows = 29646.203,29846.203, ph2_hlt
enable = hlt_p2_integration, hlt_p2_timing
workflows = ph2_hlt

waredjeb · 2025-10-06T20:47:37Z

@cmsbuild please test

cmsbuild · 2025-10-06T21:57:27Z

-1

Failed Tests: RelVals
Size: This PR adds an extra 16KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5bb023/48501/summary.html
COMMIT: f70dfb5
CMSSW: CMSSW_16_0_X_2025-10-06-1100/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/48811/48501/install.sh to create a dev area with all the needed externals and cmssw changes.

RelVals

29634.752

----- Begin Fatal Exception 06-Oct-2025 23:45:41 CEST-----------------------
An exception of category 'ConfigFileReadError' occurred while
   [0] Processing the python configuration file named step2_DIGI_L1TrackTrigger_L1_L1P2GT_DIGI2RAW_HLT_VALIDATION.py
Exception Message:
 unknown python problem occurred.
RuntimeError: An exception of category 'FileInPathError' occurred.
Exception Message:
edm::FileInPath unable to find file RecoHGCal/TICL/data/superclustering/supercls_v3.onnx anywhere in the search path.
The search path is defined by: CMSSW_SEARCH_PATH
${CMSSW_SEARCH_PATH} is: /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/48811/48501/CMSSW_16_0_X_2025-10-06-1100/poison:/cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/48811/48501/CMSSW_16_0_X_2025-10-06-1100/src:/cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/48811/48501/CMSSW_16_0_X_2025-10-06-1100/external/el8_amd64_gcc13/data:/cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw-patch/CMSSW_16_0_X_2025-10-06-1100/poison:/cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw-patch/CMSSW_16_0_X_2025-10-06-1100/src:/cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw-patch/CMSSW_16_0_X_2025-10-06-1100/external/el8_amd64_gcc13/data
Current directory is: /data/cmsbld/jenkins/workspace/ib-run-pr-relvals/matrix-results/29634.752_TTbar_14TeV+Run4D110_HLT75e33TimingTiclV5


At:
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Types.py(1050): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Mixins.py(398): insertContentsInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Types.py(1135): insertContentsInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Types.py(1128): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Mixins.py(398): insertContentsInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Mixins.py(533): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Modules.py(160): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Config.py(1120): _insertManyInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Config.py(1331): fillProcessDesc
  <string>(2): <module>

----- End Fatal Exception -------------------------------------------------

29634.771

----- Begin Fatal Exception 06-Oct-2025 23:45:49 CEST-----------------------
An exception of category 'ConfigFileReadError' occurred while
   [0] Processing the python configuration file named step2_DIGI_L1TrackTrigger_L1_L1P2GT_DIGI2RAW_HLT_VALIDATION.py
Exception Message:
 unknown python problem occurred.
RuntimeError: An exception of category 'FileInPathError' occurred.
Exception Message:
edm::FileInPath unable to find file RecoHGCal/TICL/data/superclustering/supercls_v3.onnx anywhere in the search path.
The search path is defined by: CMSSW_SEARCH_PATH
${CMSSW_SEARCH_PATH} is: /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/48811/48501/CMSSW_16_0_X_2025-10-06-1100/poison:/cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/48811/48501/CMSSW_16_0_X_2025-10-06-1100/src:/cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/48811/48501/CMSSW_16_0_X_2025-10-06-1100/external/el8_amd64_gcc13/data:/cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw-patch/CMSSW_16_0_X_2025-10-06-1100/poison:/cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw-patch/CMSSW_16_0_X_2025-10-06-1100/src:/cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw-patch/CMSSW_16_0_X_2025-10-06-1100/external/el8_amd64_gcc13/data
Current directory is: /data/cmsbld/jenkins/workspace/ib-run-pr-relvals/matrix-results/29634.771_TTbar_14TeV+Run4D110_NGTScoutingAll


At:
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Types.py(1050): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Mixins.py(398): insertContentsInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Types.py(1135): insertContentsInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Types.py(1128): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Mixins.py(398): insertContentsInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Mixins.py(533): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Modules.py(160): insertInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Config.py(1120): _insertManyInto
  /cvmfs/cms-ib.cern.ch/sw/x86_64/nweek-02910/el8_amd64_gcc13/cms/cmssw/CMSSW_16_0_X_2025-10-05-0000/src/FWCore/ParameterSet/python/Config.py(1331): fillProcessDesc
  <string>(2): <module>

----- End Fatal Exception -------------------------------------------------

HLT P2 Timing: chart

mmusich · 2025-10-07T06:44:35Z

@waredjeb I think we missed to add pull_request = https://github.com/cms-data/RecoHGCal-TICL/pull/9 to #48811 (comment).

mmusich · 2025-10-07T06:45:00Z

test parameters:

pull_request = Adding new version of SuperclusterDNN model v3 cms-data/RecoHGCal-TICL#9
workflow_opts = -w upgrade
workflows = 29646.203,29846.203, ph2_hlt
enable = hlt_p2_integration, hlt_p2_timing
workflows = ph2_hlt

waredjeb · 2025-10-07T09:10:54Z

@cmsbuild please test

waredjeb · 2025-10-07T09:11:34Z

@waredjeb I think we missed to add pull_request = https://github.com/cms-data/RecoHGCal-TICL/pull/9 to #48811 (comment).

thanks marco!

cmsbuild · 2025-10-07T15:18:30Z

+1

Size: This PR adds an extra 16KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5bb023/48511/summary.html
COMMIT: f70dfb5
CMSSW: CMSSW_16_0_X_2025-10-06-2300/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/48811/48511/install.sh to create a dev area with all the needed externals and cmssw changes.

The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic:

@CMSTrackerDPG Enabling SiPixel digi morphing from 2025 onward #49065

You can see more details here:
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5bb023/48511/git-recent-commits.json
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5bb023/48511/git-merge-result

HLT P2 Timing: chart

Comparison Summary

Summary:

You potentially added 18 lines to the logs
Reco comparison results: 2090 differences found in the comparisons
DQMHistoTests: Total files compared: 64
DQMHistoTests: Total histograms compared: 4293829
DQMHistoTests: Total failures: 2281
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 4291528
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 63 files compared)
Checked 263 log files, 223 edm output root files, 64 DQM output files
TriggerResults: no differences found

waredjeb · 2025-10-08T08:03:37Z

@mmusich looks like we don't have the reproducibility problem anymore after the merging of #49023
what do you think?

mmusich · 2025-10-08T08:10:18Z

what do you think?

there is too much churn going on in tests (due also to #49073). Let's wait to have the solution to that merged and run a final round of tests.

waredjeb · 2025-10-08T08:11:33Z

what do you think?

there is too much churn going on in tests (due also to #49073). Let's wait to have the solution to that merged and run a final round of tests.

ok perfect, thanks!

mmusich · 2025-10-12T08:52:49Z

@cmsbuild, please test

cmsbuild · 2025-10-12T13:58:12Z

+1

Size: This PR adds an extra 16KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5bb023/48618/summary.html
COMMIT: f70dfb5
CMSSW: CMSSW_16_0_X_2025-10-12-0000/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/week1/cms-sw/cmssw/48811/48618/install.sh to create a dev area with all the needed externals and cmssw changes.

HLT P2 Timing: chart

Comparison Summary

Summary:

You potentially removed 1 lines from the logs
Reco comparison results: 18 differences found in the comparisons
DQMHistoTests: Total files compared: 64
DQMHistoTests: Total histograms compared: 4293829
DQMHistoTests: Total failures: 59
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 4293750
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 63 files compared)
Checked 263 log files, 223 edm output root files, 64 DQM output files
TriggerResults: found differences in 1 / 62 workflows

waredjeb · 2025-10-13T08:58:55Z

Hi @mmusich , looks like we don't have the differences anymore, what do you think?

mmusich · 2025-10-13T09:03:46Z

TriggerResults: found differences in 1 / 62 workflows

there are changes in the trigger results (in L1 algos?). Probably not related though.

mmusich · 2025-10-13T09:05:18Z

+hlt

cmsbuild · 2025-10-13T09:05:48Z

This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @sextonkennedy, @mandrenguyen, @ftenchini (and backports should be raised in the release meeting by the corresponding L2)
Notice This PR was tested with additional Pull Request(s), please also merge them if necessary: cms-data/RecoHGCal-TICL#9

mmusich · 2025-10-14T14:44:36Z

there are changes in the trigger results (in L1 algos?). Probably not related though.

for completeness I re-run the tests in a vacuum at #49063 (comment) and I don't see the same. I wonder if it reproduces here.

waredjeb · 2025-10-14T14:53:43Z

there are changes in the trigger results (in L1 algos?). Probably not related though.

for completeness I re-run the tests in a vacuum at #49063 (comment) and I don't see the same. I wonder if it reproduces here.

If I understood correctly the differences, they are shown in workflow 29634.0 which is the default (no procModifiers) with TICLv4. This PR does not touch anything in TICLv4 so I would guess it should not be related to this PR. But if you have doubts I can try to run additional tests

mmusich · 2025-10-14T14:54:43Z

@cmsbuild, please test

cmsbuild · 2025-10-14T19:26:43Z

+1

Size: This PR adds an extra 16KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5bb023/48680/summary.html
COMMIT: f70dfb5
CMSSW: CMSSW_16_0_X_2025-10-14-1100/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/week1/cms-sw/cmssw/48811/48680/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 12 lines to the logs
Reco comparison results: 12 differences found in the comparisons
DQMHistoTests: Total files compared: 64
DQMHistoTests: Total histograms compared: 4293829
DQMHistoTests: Total failures: 14
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 4293795
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 63 files compared)
Checked 263 log files, 223 edm output root files, 64 DQM output files
TriggerResults: no differences found

mandrenguyen · 2025-10-15T19:46:28Z

+1

gsokmen and others added 5 commits August 26, 2025 19:00

DNN v3 with timing info, corrected seed pT

ad12e52

Fixing superclustering logic

e48b573

changing scls config

cb9937e

modified SCS and minor changes in the SuperclusteringDNNInputs

65e32db

Update model name in config file

6307d9d

cmsbuild added this to the CMSSW_15_1_X milestone Aug 27, 2025

cmsbuild added reconstruction-pending pending-signatures tests-pending orp-pending upgrade-pending code-checks-pending labels Aug 27, 2025

cmsbuild added code-checks-rejected and removed code-checks-pending labels Aug 27, 2025

cmsbuild added the allow-waredjeb label Aug 27, 2025

cmsbuild added hgcal allow-sameasy labels Aug 27, 2025

cmsbuild added the requires-external label Aug 27, 2025

Adding code format

1526f0d

cmsbuild added code-checks-pending and removed code-checks-rejected labels Aug 27, 2025

cmsbuild added code-checks-approved and removed code-checks-pending labels Aug 27, 2025

cmsbuild added the tests-approved label Sep 29, 2025

cmsbuild removed the requires-external label Oct 6, 2025

cmsbuild added tests-started and removed tests-approved labels Oct 6, 2025

cmsbuild added tests-rejected and removed tests-started labels Oct 6, 2025

cmsbuild mentioned this pull request Oct 13, 2025

Adding new version of SuperclusterDNN model v3 cms-data/RecoHGCal-TICL#9

Merged

Conversation

gsokmen commented Aug 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cmsbuild commented Aug 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cmsbuild commented Aug 27, 2025

Uh oh!

felicepantaleo commented Aug 27, 2025

Uh oh!

felicepantaleo commented Aug 27, 2025

Uh oh!

felicepantaleo commented Aug 27, 2025

Uh oh!

felicepantaleo commented Aug 27, 2025

Uh oh!

cmsbuild commented Aug 27, 2025

Uh oh!

waredjeb commented Sep 29, 2025

Uh oh!

waredjeb commented Oct 6, 2025

Uh oh!

waredjeb commented Oct 6, 2025

Uh oh!

cmsbuild commented Oct 6, 2025

RelVals

Uh oh!

mmusich commented Oct 7, 2025

Uh oh!

mmusich commented Oct 7, 2025

Uh oh!

waredjeb commented Oct 7, 2025

Uh oh!

waredjeb commented Oct 7, 2025

Uh oh!

cmsbuild commented Oct 7, 2025

Comparison Summary

Uh oh!

waredjeb commented Oct 8, 2025

Uh oh!

mmusich commented Oct 8, 2025

Uh oh!

waredjeb commented Oct 8, 2025

Uh oh!

mmusich commented Oct 12, 2025

Uh oh!

cmsbuild commented Oct 12, 2025

Comparison Summary

Uh oh!

waredjeb commented Oct 13, 2025

Uh oh!

mmusich commented Oct 13, 2025

Uh oh!

mmusich commented Oct 13, 2025

Uh oh!

cmsbuild commented Oct 13, 2025

Uh oh!

mmusich commented Oct 14, 2025

Uh oh!

waredjeb commented Oct 14, 2025

Uh oh!

mmusich commented Oct 14, 2025

Uh oh!

cmsbuild commented Oct 14, 2025

Comparison Summary

Uh oh!

mandrenguyen commented Oct 15, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

9 participants

gsokmen commented Aug 27, 2025 •

edited

Loading

cmsbuild commented Aug 27, 2025 •

edited

Loading