Add executable to compare truth vs. CUDA seeding #906
Conversation
This commit adds a new executable that compares the truth seeding to the CUDA seeding, emitting a CSV file with matched seeds allowing further analysis with e.g. Python.
stephenswat
added
feature
|
| const measurement& bot_meas = | ||
| truth_track_candidates.at(i_trk).items.at(0); | ||
| const measurement& mid_meas = | ||
| truth_track_candidates.at(i_trk).items.at(1); | ||
| const measurement& top_meas = | ||
| truth_track_candidates.at(i_trk).items.at(2); | ||
|
|
||
| const std::map<particle, std::size_t>& bot_ptc_map = | ||
| evt_data.m_meas_to_ptc_map.at(bot_meas); | ||
| const std::map<particle, std::size_t>& mid_ptc_map = | ||
| evt_data.m_meas_to_ptc_map.at(mid_meas); | ||
| const std::map<particle, std::size_t>& top_ptc_map = | ||
| evt_data.m_meas_to_ptc_map.at(top_meas); |
There was a problem hiding this comment.
@beomki-yeo I've noticed that if I get the bottom, middle, and top spacepoint positions like this, I get non-sensical
There was a problem hiding this comment.
It is not expected and the track candidates do not have to be sorted.
How did you get the spacepoint positions from bot_meas, mid_meas and top_meas?
There was a problem hiding this comment.
After going through event_data.cpp, it is possible that we may need to sort the track candidate if csv_measurements is not sorted in the timing order:
traccc/performance/src/utils/event_data.cpp
Line 230 in 6fee5ca
There was a problem hiding this comment.
I found that measurement.csv of geant4 odd data does not have truth timing information (all of them are set to zero).
We will have to find the truth time from the corresponding hit object in hits.csv, which has truth timing information, via measurement-simhit-map.csv
Then you can run sorting on measurement (track candidate) vectors w.r.t. truth timing
This commit adds a new executable that compares the truth seeding to the CUDA seeding, emitting a CSV file with matched seeds allowing further analysis with e.g. Python.