Iss864 #865
Iss864 #865
Conversation
Remove HelicalTrackHitDriver as it should not be needed for the Kalman Filter.
Remove unused calls to hitToStrips and hitToRotated
Indentations.
Fix imports
Switch to MinDist track cluster matcher
Don't skip monster events.
normangraf
requested review from
cbravo135,
omar-moreno,
pbutti,
alspellm and
bloodyyugo
|
Can you show some checks that this did not change the behavior of PhysicsRun2019_pass0_recon_evio.lcsim? |
|
The two collections that were being accessed were never being used. Commenting out those calls should not affect the results of any other Drivers. |
| @@ -87,7 +89,7 @@ | |||
|
|
|||
| <driver name="TrackerHitDriver" type="org.hps.recon.tracking.DataTrackerHitDriver"> | |||
| <neighborDeltaT>8.0</neighborDeltaT> | |||
| <saveMonsterEvents>false</saveMonsterEvents> | |||
| <saveMonsterEvents>true</saveMonsterEvents> | |||
There was a problem hiding this comment.
Why is this changed from false to true?
There was a problem hiding this comment.
I don't believe the current way of handling these events is correct, and simply skipping these events may be giving us a false impression of how well we're doing. I'd like to see what effect this has in combination with enabling the event flag filter.
There was a problem hiding this comment.
The event flag filter is not currently activated in this steering file.
There was a problem hiding this comment.
If this is your intention, please add it to this steering file and I will approve this PR.
|
I think at a minimum you should run on one file and show that these collections are indeed removed as expected. |
|
Also, if we plan on moving forward with this steering file now, you should add the EventFlagFilter driver and turn on the "svt_readout_overlap_good" flag. |
|
Seems to be a "Fitting with: Simplex" debug printout left in the code. Ooops! Added this comment to the wrong PR. |
|
The collections aren't (yet) being removed. I was simply removing an extraneous call to access those collections, which requires them to be in the event. Once we drop the SeedTracker and GBL track finding and fitting these collections will no longer be in the event. No behavior has changed. |
I removed superfluous code in ReconParticleDriver that accessed the rotated helical track hits for no reason.
One can now run the Kalman Filter track finding and fitting without having to create HelicalTrackHits.
The KF only steering file has been modified to remove that Driver.
It has also been updated to use the distance-based track cluster matcher.
It also now keeps monster events instead of zeroing out the 1D strip hit collection.