Integrating VectorHits reconstruction for the Phase2 OT - reprise #31314

JanFSchulte · 2020-09-01T15:22:30Z

This is a renewed attempt to integrate the Vector Hits reconstruction in CMSSW. This is a continuation of PR #28101 (#28101), which was closed last year to allow for time to restructure the code.

In particular, the code has been refactored to no longer have an EDProducer as the product of an ESProducer. As before, this PR only integrates the reconstruction code but leaves it deactivated. When activated, this PR adds a new tracking iteration "pixelLess" to the phase 2 tracking targeting displaced tracks.

The code changes of this PR have been tested on 100 Events of the 27434.1 workflow in a recent IB. As expected, the tracking results are identical as the Vector Hits are inactive. This can be seen in the validation results here: https://jschulte.web.cern.ch/jschulte/VH_11_2_X/plots/plots_summary.html

The integration strategy for the VectorHits was last discussed at a TRK POG meeting last year: https://indico.cern.ch/event/847559/contributions/3572893/attachments/1914802/3165228/VectorHits_status_sept25.pdf.
However, the statements in those slides are still valid. A new assessment of the performance of the Vector Hits will be provided with the follow-up PR to activate them.

PR validation:

Compiliation after checking out all dependencies was sucessful
Passes runTheMatrix.py -l limited -i all --ibeos without errors

Final performance after code review, tested on 100 ttbar events at PU 200:

All tracks:
Efficiency: 0.9122 -> 0.9292
Fake rate: 0.1240 -> 0.0324
Duplicate rate: 0.0045 -> 0.0064
High purity tracks:
Efficiency: 0.9054 -> 0.9204
Fake rate: 0.0549 -> 0.0153
Duplicate rate: 0.0038 -> 0.0048

Efficiency gain is mostly located at low pT and efficiencies converge at around 10 GeV
Low pT fake rate is significantly reduced

PR significantly improves sensitivity to tracks with moderate displacement

With the current implementation this comes at a timing cost of about 20% for the tracking-only RECO step

Here the pixelLess iteration accounts for 40% of the total pattern recognition time, which can be improved with improvements to the seeding.

cmsbuild · 2020-09-01T15:22:55Z

The code-checks are being triggered in jenkins.

cmsbuild · 2020-09-01T15:30:33Z

-code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31314/18061

This PR adds an extra 420KB to repository
Found files with invalid states:
- RecoLocalTracker/SiPhase2VectorHitBuilder/python/SiPhase2VectorHitBuilder_cfi.py:
  - Added: dc35aab, c393369
  - Deleted: 03698e6
- RecoLocalTracker/SiPhase2VectorHitBuilder/python/SiPhase2RecHitMatcher_cfi.py:
  - Added: dc35aab, c393369, a6f4890
  - Deleted: 03698e6, 922c8ac
- RecoLocalTracker/SiPhase2VectorHitBuilder/plugins/SiPhase2RecHitMatcherESProducer.h:
  - Added: dc35aab
  - Modified: 2d86c8b
  - Deleted: 03698e6
There are other open Pull requests which might conflict with changes you have proposed:
- File RecoLocalTracker/Configuration/python/RecoLocalTracker_cff.py modified in PR(s): Squash all Patatrack developments on top of CMSSW_11_3_0_pre5 #27983

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

code-checks:
https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31314/18061/code-checks.patch
e.g. curl https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31314/18061/code-checks.patch | patch -p1
You can also run scram build code-checks to apply code checks directly
code-format:
https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31314/18061/code-format.patch
e.g. curl https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31314/18061/code-format.patch | patch -p1
You can also run scram build code-format to apply code format directly

cmsbuild · 2020-09-01T15:41:58Z

The code-checks are being triggered in jenkins.

cmsbuild · 2020-09-01T15:47:44Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31314/18062

This PR adds an extra 424KB to repository
Found files with invalid states:
- RecoLocalTracker/SiPhase2VectorHitBuilder/python/SiPhase2VectorHitBuilder_cfi.py:
  - Added: dc35aab, c393369
  - Deleted: 03698e6
- RecoLocalTracker/SiPhase2VectorHitBuilder/python/SiPhase2RecHitMatcher_cfi.py:
  - Added: dc35aab, c393369, a6f4890
  - Deleted: 03698e6, 922c8ac
- RecoLocalTracker/SiPhase2VectorHitBuilder/plugins/SiPhase2RecHitMatcherESProducer.h:
  - Added: dc35aab
  - Modified: 2d86c8b
  - Deleted: 03698e6
There are other open Pull requests which might conflict with changes you have proposed:
- File RecoLocalTracker/Configuration/python/RecoLocalTracker_cff.py modified in PR(s): Squash all Patatrack developments on top of CMSSW_11_3_0_pre5 #27983

cmsbuild · 2020-09-01T15:48:10Z

A new Pull Request was created by @JanFSchulte (Jan-Frederik Schulte) for master.

It involves the following packages:

CommonTools/RecoAlgos
DataFormats/TrackerRecHit2D
Geometry/TrackerGeometryBuilder
RecoLocalTracker/Configuration
RecoLocalTracker/Records
RecoLocalTracker/SiPhase2VectorHitBuilder
RecoLocalTracker/SubCollectionProducers
RecoTracker/CkfPattern
RecoTracker/FinalTrackSelectors
RecoTracker/GeometryESProducer
RecoTracker/MeasurementDet
RecoTracker/TkDetLayers
RecoTracker/TkSeedGenerator
RecoTracker/TkSeedingLayers
RecoTracker/TransientTrackingRecHit
SimTracker/TrackAssociatorProducers

The following packages do not have a category, yet:

RecoLocalTracker/SiPhase2VectorHitBuilder
Please create a PR for https://github.com/cms-sw/cms-bot/blob/master/categories_map.py to assign category

@perrotta, @civanch, @Dr15Jones, @makortel, @cvuosalo, @ianna, @mdhildreth, @cmsbuild, @slava77, @jpata can you please review it and eventually sign? Thanks.
@echabert, @rappoccio, @felicepantaleo, @pieterdavid, @robervalwalsh, @mschrode, @abbiendi, @mmusich, @venturia, @makortel, @threus, @JanFSchulte, @jhgoh, @dgulhan, @jdolen, @yduhm, @GiacomoSguazzoni, @rovere, @VinInn, @gkasieczka, @hatakeyamak, @alesaggio, @ebrondol, @mtosi, @fabiocos, @clelange, @riga, @gbenelli, @ahinzmann, @gpetruc this is something you requested to watch as well.
@silviodonato, @dpiparo, @qliphy you are the release manager for this.

cms-bot commands are listed here

slava77 · 2020-09-01T16:03:03Z

@cmsbuild please test

cmsbuild · 2020-09-01T16:03:32Z

The tests are being triggered in jenkins.

CMSSW_11_2_X_2020-09-01-1100/slc7_amd64_gcc820: https://cmssdt.cern.ch/jenkins/job/ib-run-pr-tests/9039/console Started: 2020/09/01 18:04

cmsbuild · 2020-09-01T18:49:55Z

+1
Tested at: f155383
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-7ffe9a/9039/summary.html
CMSSW: CMSSW_11_2_X_2020-09-01-1100
SCRAM_ARCH: slc7_amd64_gcc820

cmsbuild · 2020-09-01T18:49:58Z

Comparison job queued.

cmsbuild · 2020-09-01T20:19:57Z

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-7ffe9a/9039/summary.html

Comparison Summary:

No significant changes to the logs found
Reco comparison results: 8 differences found in the comparisons
DQMHistoTests: Total files compared: 35
DQMHistoTests: Total histograms compared: 2609667
DQMHistoTests: Total failures: 11
DQMHistoTests: Total nulls: 1
DQMHistoTests: Total successes: 2609633
DQMHistoTests: Total skipped: 22
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: -0.004 KiB( 34 files compared)
DQMHistoSizes: changed ( 10224.0 ): -0.004 KiB MessageLogger/Warnings
Checked 149 log files, 22 edm output root files, 35 DQM output files

slava77 · 2020-10-15T22:03:42Z

+1

for #31314 53aad7e

code changes are in line with the PR description and the follow up review: to integrate phase-2 outer tracker pt-module hit pairs as VectorHit (optionally enabled with a processModifier vectorHits)
jenkins tests pass and comparisons with the baseline show no differences; the profiler reports for the default setup show no relevant differences.

Links to trackinOnly tests with PU200 D49 setup with vectorHits enabled for the most recent tests are available in #31314 (comment)

Due to the length of the PR thread, it is not practical to look up some of the performance plots or to summarize what is in the final version of the PR.
@JanFSchulte please update the PR description so that it reflects the most recent updates
and perhaps also include the total time per event spent in tracking in the baseline and the latest variant of the PR using vectorHits for future reference

The longer term points discussed earlier are for some future time

about using looser cuts for the hits used in tracking (after seeding), instead of all combinations
see if something better can be done to avoid direct copies of hits (perhaps the point above will eliminate the need for this optimization)

JanFSchulte · 2020-10-16T00:07:32Z

@slava77 I have updated the PR description with the required information.

silviodonato · 2020-10-16T16:24:49Z

+operations

chayanit · 2020-10-17T05:34:27Z

+1

kpedro88 · 2020-10-17T17:10:54Z

+upgrade

silviodonato · 2020-10-19T08:56:54Z

kind reminder @cms-sw/geometry-l2 @cms-sw/simulation-l2

slava77 · 2020-10-20T12:51:30Z

kind reminder @cms-sw/geometry-l2 @cms-sw/simulation-l2

It would be good to have this PR in pre8.
Is there a time estimate to get the geometry and simulation signatures?

silviodonato · 2020-10-20T17:33:18Z

urgent
for pre8

cvuosalo · 2020-10-20T21:01:01Z

+1

silviodonato · 2020-10-21T18:54:03Z

merge

cmsbuild added this to the CMSSW_11_2_X milestone Sep 1, 2020

cmsbuild added code-checks-pending comparison-pending geometry-pending new-package-pending orp-pending pending-signatures reconstruction-pending simulation-pending tests-pending labels Sep 1, 2020

cmsbuild added code-checks-rejected and removed code-checks-pending labels Sep 1, 2020

cmsbuild added code-checks-pending and removed code-checks-rejected labels Sep 1, 2020

cmsbuild added code-checks-approved and removed code-checks-pending labels Sep 1, 2020

cmsbuild added tests-started and removed tests-pending labels Sep 1, 2020

cmsbuild added tests-approved and removed tests-started labels Sep 1, 2020

cmsbuild added comparison-available and removed comparison-pending labels Oct 15, 2020

cmsbuild added reconstruction-approved and removed reconstruction-pending labels Oct 15, 2020

cmsbuild added operations-approved and removed operations-pending labels Oct 16, 2020

cmsbuild added pdmv-approved and removed pdmv-pending labels Oct 17, 2020

cmsbuild added upgrade-approved and removed upgrade-pending labels Oct 17, 2020

cmsbuild mentioned this pull request Oct 18, 2020

Phase2-gex25 Add a new scenario 2026D74 with V14 HGCal and GE0 #31842

Merged

This was referenced Oct 19, 2020

Patatrack integration - Pixel workflows (12/N) #31854

Merged

Patatrack integration - Pixel local reconstruction (9/N) #31721

Merged

cmsbuild added the urgent label Oct 20, 2020

cmsbuild added geometry-approved and removed geometry-pending labels Oct 20, 2020

cmsbuild added orp-approved and removed orp-pending labels Oct 21, 2020

cmsbuild merged commit 1d50057 into cms-sw:master Oct 21, 2020

silviodonato mentioned this pull request Oct 22, 2020

Segmentation violation in workflows 33034.0 and 33434.0 (geometry D72) #31895

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Integrating VectorHits reconstruction for the Phase2 OT - reprise #31314

Integrating VectorHits reconstruction for the Phase2 OT - reprise #31314

JanFSchulte commented Sep 1, 2020 •

edited

Loading

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

slava77 commented Sep 1, 2020

cmsbuild commented Sep 1, 2020 •

edited

Loading

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

slava77 commented Oct 15, 2020

JanFSchulte commented Oct 16, 2020

silviodonato commented Oct 16, 2020

chayanit commented Oct 17, 2020

kpedro88 commented Oct 17, 2020

silviodonato commented Oct 19, 2020

slava77 commented Oct 20, 2020

silviodonato commented Oct 20, 2020

cvuosalo commented Oct 20, 2020

silviodonato commented Oct 21, 2020

Integrating VectorHits reconstruction for the Phase2 OT - reprise #31314

Integrating VectorHits reconstruction for the Phase2 OT - reprise #31314

Conversation

JanFSchulte commented Sep 1, 2020 • edited Loading

PR validation:

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

slava77 commented Sep 1, 2020

cmsbuild commented Sep 1, 2020 • edited Loading

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

cmsbuild commented Sep 1, 2020

slava77 commented Oct 15, 2020

JanFSchulte commented Oct 16, 2020

silviodonato commented Oct 16, 2020

chayanit commented Oct 17, 2020

kpedro88 commented Oct 17, 2020

silviodonato commented Oct 19, 2020

slava77 commented Oct 20, 2020

silviodonato commented Oct 20, 2020

cvuosalo commented Oct 20, 2020

silviodonato commented Oct 21, 2020

JanFSchulte commented Sep 1, 2020 •

edited

Loading

cmsbuild commented Sep 1, 2020 •

edited

Loading