Adding pass-through option to MultiTrackSelector (for LST tracking / seeding) - LST-internal only#247
Adding pass-through option to MultiTrackSelector (for LST tracking / seeding) - LST-internal only#247
Conversation
|
run-ci: all |
|
The PR was built and ran successfully in standalone mode running on CPU. Here are some of the comparison plots.
The full set of validation and comparison plots can be found here. Here is a timing comparison: |
|
There was a problem while building and running with CMSSW. The logs can be found here. |
mmasciov
force-pushed
the
passThroughMTS-SL
branch
from
4ae573e to
2f5c077
Compare
|
run-ci: all |
|
The PR was built and ran successfully in standalone mode running on CPU. Here are some of the comparison plots.
The full set of validation and comparison plots can be found here. Here is a timing comparison: |
|
The PR was built and ran successfully with CMSSW running on CPU. Here are some plots. OOTB All Tracks
The full set of validation and comparison plots can be found here. |
| (trackingPhase2PU140 & (trackingLST | seedingLST)).toModify(highPtTripletStepSelector, passThroughForAll = True) | ||
| # Passthrough selector to satisfy the TrackListMerger requirement for selector values | ||
| highPtTripletStepSelectorLSTT4T5 = RecoTracker.FinalTrackSelectors.multiTrackSelector_cfi.multiTrackSelector.clone( | ||
| src = 'highPtTripletStepLSTT4T5Tracks', | ||
| passThroughForDisplaced = True, |
There was a problem hiding this comment.
which of these is making a change in plots_building_highPtTripletStep ?
Now that the selector functionality allows to select T4/T5 from the same collection, can we already use a merged collection?
There was a problem hiding this comment.
which of these is making a change in plots_building_highPtTripletStep ?
Nothing should change at all, because to my understanding the tested workflow does NOT include LST procModifiers.
EDIT: actually, apparently the cmsDriver command is modified wrt. the listed workflow, looking at the logs. This is confusing - at least to me. Will investigate further.
Now that the selector functionality allows to select T4/T5 from the same collection, can we already use a merged collection?
Yes. I will try to simplify this part.
There was a problem hiding this comment.
Nothing should change at all, because to my understanding the tested workflow does NOT include LST procModifiers.
https://github.com/SegmentLinking/TrackLooper-actions/blob/main/cmssw/run.sh has
--procModifiers trackingIters01,trackingLST \
There was a problem hiding this comment.
Nothing should change at all, because to my understanding the tested workflow does NOT include LST procModifiers.
https://github.com/SegmentLinking/TrackLooper-actions/blob/main/cmssw/run.sh has
--procModifiers trackingIters01,trackingLST \
Later noticed in the logs, hence my "EDIT" above.
I believe I now understand why we see a difference at "building" level in LST.
The reason is that the tracks being monitored when LST is enabled are not the product of a TrackProducer, but LST tracks that undergo already the high-purity selection:
https://github.com/cms-sw/cmssw/blob/master/RecoTracker/IterativeTracking/python/HighPtTripletStep_cff.py#L293
2 participants
As per title.
FYI: @bucket420, @slava77, @bdanzi