Cleanup of tracking core.
Squashed version of !1900 (closed).
- Cleaned up ca::TripletConstructor.
- Removed global ca::TrackFit instance from ca::TripletConstructor.
- Removed access to ca::InputData from ca::TripletConstructor (all needed data is already in ca::WindowData).
- Switched from global to window-based hit index in ca::Branch. Allows simplification of many interfaces (removal of ca::InputData).
- Cleaned up ca::TrackFinder.
- In ca::TrackFinder, replaced separate vector instances which store upper and lower bounds of windows by single vectors that store these bounds together as a pair.
- Cleaned up ca::TrackFinderWindow.
Merge request reports
Activity
requested review from @s.zharko
assigned to @se.gorbunov
@se.gorbunov @s.zharko The class-fields for storage of intermediate data were re-introduced, to avoid memory allocations, as discussed. The transition to using window-based instead of global hit indices where possible was completed. Ready for merge in my opinion.
mentioned in merge request !1900 (closed)
added CodeOwners label
added 13 commits
-
e5f804fe...043446e8 - 12 commits from branch
computing:master - 09d8fa1f - - Cleaned up ca::TripletConstructor.
-
e5f804fe...043446e8 - 12 commits from branch
171 fDoubletData.second.clear(); 172 if (iMC < 0 && matchMc) { 193 173 return; 194 174 } 175 CollectHits(fDoubletData.second, fit, fIstaM, frWData.CurrentIteration()->GetDoubletChi2Cut(), iMC, 176 fParameters.GetMaxDoubletsPerSinglet()); 177 }; 178 179 FindDoubletHits(); 180 FindDoublets(fit); 181 182 //D.Smith 28.8.24: Moving this upward (before doublet finding) changes QA output slightly 183 if (fIstaR >= fParameters.GetNstationsActive()) { 184 tripletsOut.clear(); 185 return; 195 186 } - Comment on lines +182 to 186
@s.zharko @se.gorbunov I found the origin of the small difference in QA output reported yesterday. It was this conditional return statement. I tried to move it upward, before the computation of doublets, to save time, but this had a small effect on QA. Possibly, this happens because the doublet computation changes the contents of the window. Not sure whether this is intended, but I thought I would let you know.
added 10 commits
-
09d8fa1f...a6260228 - 9 commits from branch
computing:master - a3eaed16 - - Cleaned up ca::TripletConstructor.
-
09d8fa1f...a6260228 - 9 commits from branch
@s.zharko @se.gorbunov Rebased to account for !1904 (merged). The necessary changes were applied. The QA output is now different from before, but matches the current master branch. The number of produced triplets is smaller, due to some added clear() statements (discussed in private conversations), but this does not affect the QA plots.
mentioned in merge request !1910 (merged)
added Tracking label
added 9 commits
-
5ef6ddea...584a3864 - 8 commits from branch
computing:master - 4994cf9c - - Cleaned up ca::TripletConstructor.
-
5ef6ddea...584a3864 - 8 commits from branch
Superceded by !1910 (merged). Closing.
Edited by Dominik Smith
