Skip to content

Implement fastest reconstuction for selections

Sascha Stahl requested to merge sstahl_fastest_single_container into master

Replaces !944 (closed) .

This MR enables the use of the fastest reconstruction in selections. The difference to the fastest reconstruction implemented in standalone.py is that tracks (Long, Downstream and unfitted T-Tracks) are merged into one container before running the particle identification and protoparticle making. The main reason for this difference are the assumptions currently made in persist reco. Persist reco assumes one container for tracks, protoparticles and PID objects. If it is desired to have separate containers, I suggest that this is done in a follow up MR after consulting with all parties involved.

The signature of one function in the protoparticle making slightly changed to support the default and fastest reconstruction. The fallout is that functors for the track selection are always created. This leads to some failures of qm tests counting how many functors are taken from the functor cache. Some tests are fixed by enabling to ignore the value of this counter. But in the case where the counter was not present, the test still fails and a reference update is necessary.

  • Added test using the fastest reconstruction, includes !985 (merged)
Edited by Sascha Stahl

Merge request reports