fix: Solve the CI issue with the ML solver

[Corentin-Allaire]

This ML will try to solve the issue #2023. Unfortunately I couldn't reproduce that issue on my machine so I have to test it directly on the CI.

[codecov]

Codecov Report

Merging #2026 (fa5bbde) into main (8c837ad) will increase coverage by 0.05%.
The diff coverage is n/a.

❗ Current head fa5bbde differs from pull request most recent head 2bcdad3. Consider uploading reports for the commit 2bcdad3 to get more accurate results

@@            Coverage Diff             @@
##             main    #2026      +/-   ##
==========================================
+ Coverage   49.82%   49.87%   +0.05%     
==========================================
  Files         420      421       +1     
  Lines       23955    23898      -57     
  Branches    10857    10843      -14     
==========================================
- Hits        11935    11920      -15     
+ Misses       4388     4357      -31     
+ Partials     7632     7621      -11     

see 12 files with indirect coverage changes

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

[github-actions]

📊 Physics performance monitoring for 2bcdad3

Full report
Seeding: seeded, truth estimated, orthogonal
CKF: seeded, truth smeared, truth estimated, orthogonal
IVF: 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

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

[Corentin-Allaire]

The issue should now be fixed. What was happening was that for the input, I was declaring an Eigen::Array with a size based on the number of tracks in the collection. But the Solver removes the tracks with less measurement than the minimum required, so the size of the Eigen::Array was too large, and the last few elements were just random memory that sometimes was interpreted as nan.

The good news is that we were only reading the entry corresponding to the tracks of interest afterwards, so the performance shouldn't be affected (actually, the code should be slightly faster for a larger number of required measurements).

[andiwand]

nice fix!