Track candidate update nov25 #1073
Conversation
…f the features of this class. Remove reference trajectory and pulls, and add momentum, position and charge as explicit members of the DTrackCandidate class. Add lists of cdc and fdc hits explicitly to the class instead of using associated objects, thereby maintaining the order. Add first and last package indexes for FDC candidates. Add minimum drift time.
…ny methods to try to handle all sorts of edge conditions. Most of the time only one or two of the same subset of methods actually worked; the rest just wasted cpu cycles for marginal or no additional benefit.
|
Test status for this pull request: SUCCESS Summary: /work/halld/pull_request_test/halld_recon^TrackCandidateUpdateNov25/tests/summary.txt Build log: /work/halld/pull_request_test/halld_recon^TrackCandidateUpdateNov25/make_TrackCandidateUpdateNov25.log |
|
For 2017 (30300_000), I see a 3% drop in rho yields for this PR compared to the current master: |
|
The tracks reconstructed with this PR have a significantly lower number of hits associated to them. |
Further cleanup of track candidate code. Replaces multiple CDC/FDC matching methods, most of which were not efficient but wasted CPU cycles anyway, with much more simplified code. Removes dependence on reference trajectory and magnetic field swimmer. Removes reliance on DTrackingData in DTrackCandidate.h.