fix: Fix CKF finalization

[andiwand]

Apparently our CKF was not able to smooth multiple tracks for a single seed in some cases. The problem was that we expected the CKF actor to be called a second time after the first smoothing was done but there is a Stepper step in between which could kick you out of the detector because the step length is basically unconstrained.

fixes

• bug: interpolation out of bounds with ATLAS B-field map #1484

[codecov]

Codecov Report

Merging #2299 (9559a4f) into main (cb16df8) will increase coverage by 0.00%.
The diff coverage is 0.00%.

@@           Coverage Diff           @@
##             main    #2299   +/-   ##
=======================================
  Coverage   49.27%   49.28%           
=======================================
  Files         450      450           
  Lines       25407    25405    -2     
  Branches    11728    11728           
=======================================
  Hits        12520    12520           
+ Misses       4549     4547    -2     
  Partials     8338     8338           

Impacted Files	Coverage Δ
...de/Acts/TrackFinding/CombinatorialKalmanFilter.hpp	31.75% <0.00%> (+0.14%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[andiwand]

@timadye if you could test this and/or have a look at the changes it would be great!

[timadye]

@timadye if you could test this and/or have a look at the changes it would be great!

Sure, let me update my build and I'll give it a go. I'm hoping for great things! 😁

[github-actions]

📊 Physics performance monitoring for 897d215

Summary
Full report
Seeding: seeded, truth estimated, orthogonal
CKF: seeded, truth smeared, truth estimated, orthogonal
IVF: seeded, truth smeared, truth estimated, orthogonal
AMVF: seeded, truth smeared, truth estimated, orthogonal
Ambiguity resolution: seeded, orthogonal
Truth tracking
Truth tracking (GSF)

Vertexing

Vertexing vs. mu

IVF seeded

IVF truth_smeared

IVF truth_estimated

IVF orthogonal

AMVF seeded

AMVF truth_smeared

AMVF truth_estimated

AMVF orthogonal

Seeding

Seeding seeded

Seeding truth_estimated

Seeding orthogonal

CKF

CKF seeded

CKF truth_smeared

CKF truth_estimated

CKF orthogonal

Ambiguity resolution

seeded

Truth tracking (Kalman Filter)

Truth tracking

Truth tracking (GSF)

Truth tracking

[timadye]

Hi @andiwand,

This is a HUGE improvement!

I tested with full_chain_itk.py for 100 ttbar_pu200 events, and saw the following changes:

1. All the ~3428/event TrackFinding ERROR messages are gone.
2. Previously 15% of seeds failed in TrackFindingAlgorithm. Now it is exactly 0 seeds failing.
3. The CKFPerforman tracking efficiency is increased from 90.0 → 97.2%. At |η| ~ 2.1 it dropped to 69%, now it is 96.4% (no drop).
4. The job was ~24% slower. I guess much of the extra time is due to successfully processing more seeds.
5. There are still a few remaining ERROR and WARNING messages, but now at the level of ~11/event (not ~3428/event!). These messages (including the FPE) were all there before, so this doesn't speak against this fix. Here are typical remaining messages prefixed by the message count for 100 events:

1097 Propagator     ERROR     Step failed with MagneticFieldError:1: Interpolation out of bounds was requested
  26 Propagator     WARNING   Field lookup was unsuccessful, this is very likely an error
1097 FatrasSimula   ERROR     event 75 particle id=4567375009546240(1|58|221|0|0)|pdg=pi+|q=1|m=0.13957|p=52.5121 failed to simulate with error MagneticFieldError:1: Interpolation out of bounds was requested
   1 Sequencer      ERROR     Encountered 1 unmasked FPEs

Apart from the increase in CPU time, this is all extremely positive. I hope this fix can be merged.

Thanks,
Tim.

[andiwand]

sounds great! it might be that the remaining errors are fixed by #2300 if those are coming from Fatras

[timadye]

[CarloVarni]

one silly question: how is it this PR is not changing performances in the tests but it is for the ITk script? What am I missing?

[andiwand]

one silly question: how is it this PR is not changing performances in the tests but it is for the ITk script? What am I missing?

we do not have a ttbar event in the physmon but only single particles which seem not to be affected

[CarloVarni]

sounds great! it might be that the remaining errors are fixed by #2300 if those are coming from Fatras

@timadye can you check this as well?

[CarloVarni]

oh right, I forgot that paul's PR that adds ttbar test is not in yet. Thanks @andiwand . I'd say we can proceed with this then

[acts-project-service]

✅ Athena integration test results

✅ All tests successful

status	job
🟢	run_unit_tests
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsWorkflow.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsValidateTracks.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsValidateActsCoreSpacePoints.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsValidateActsTrkSpacePoints.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsValidateSeeds.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsValidateOrthogonalSeeds.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsValidateClusters.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsPersistifyEDM.sh
🟢	run_ci_tests: ../athena/AtlasTest/CITest/test/ActsKfRefitting.sh
🟢	run_ci_tests: python3 ../athena/Tracking/Acts/ActsGeometry/test/ActsExtrapolationAlgTest.py
🟢	run_ci_tests: python3 ../athena/Tracking/Acts/ActsGeometry/test/ActsITkTest.py
🟢	run_workflow_tests_run4_mc
🟢	run_workflow_tests_run2_mc
🟢	run_workflow_tests_run2_data
🟢	run_workflow_tests_run3_mc
🟢	run_workflow_tests_run3_data