Presentation is loading. Please wait.

Presentation is loading. Please wait.

First Thoughts About Backtracking Validation Bruce Schumm UC Santa Cruz October 3, 2007 Inner Detector Software Meeting.

Similar presentations


Presentation on theme: "First Thoughts About Backtracking Validation Bruce Schumm UC Santa Cruz October 3, 2007 Inner Detector Software Meeting."— Presentation transcript:

1 First Thoughts About Backtracking Validation Bruce Schumm UC Santa Cruz October 3, 2007 Inner Detector Software Meeting

2 “BackTracking” = extending TRT segments backwards into SCT and pixels to find non-prompt tracks Backtracking is only now going through its first real performance assessment (see talk next session) Isolated 20 GeV photon conversions (see next session) Isolated 1-50 GeV p  muons To date, studies done with “nightlies”; backtracking not working well in production releases (13.0.30 expected to have operational backtracking) It is a bit premature to do meaningful validation (nightly comparisons), but perhaps time to start thinking about what it might entail.

3 Formal validation studies will be incorporated in automated InDetRecStatistics analysis. Structure of the The InDetRecStat ntuple:

4 TRTSeededTracks (“BackTracks”) are an instance in InDetRecStat, but IndetRecStat designed for validation; capability to do performance studies is limited Tracks originating within TRT will only have TRT segments, which are not “Tracks” from C++ data-format perspective  Are looking into possibility of including segments as another instance in InDetRecStat This will allow validation of the algorithms that reconstruct non-prompt tracks, including those that originate in TRT (conversions, CHAMPS, etc.) One (modest) BackTrack performance study done within InDetRecStat on isolated muons with 1-50 GeV p 

5 Idealize TRT geometry Approximate as straight line in r-z Estimate path-length in TRT Note: Particles can terminate in and emanate from TRT volume Defining “trackable” tracks

6 This crude TRT path-length approximation could easily be improved, but best would be to have the number of TRT, SCT, Pixel?) hits for each MC particle  outside scope of InDetRecStat “Findable” track definition (start restrictively) Outward-traveling TRT path length > 500mm p  > 750 MeV/c eta< 2.0 Radius of originNo cut for now  Sample contains 155 “findable” isolated muons

7 Use nightlies to reconstruct muons; write out InDetRecStat ntuple. Compare standard tracking (XKalmanTracks) to backtracking (TRTSeededTracks) XKalTracks: 154/155 found (99%) TRTSeededTracks: 142/155 found (92%) What about the 12 tracks found by Xkal but not by BackTracking? Number of TRT hits on XKal Track Found by BackTrack Missed by BackTrack

8 Maybe: The particles missed by BackTracking suffered some sort of disruption in or near the TRT? Try looking at ChiSq of the XKalTracks separately for those found and missed by BackTracking… Found by BackTrack Missed by BackTrack ChiSq per DOF for XKal Tracks

9 PP PP   All Findable ParticlesMissed by Backtracking All Findable ParticlesMissed by Backtracking

10 Back to validation… As mentioned, need to explore inclusion of TRT segments in InDetRecStat Thoughts about Validation Material Current data files: single muon, multi muon, Z   all probably useful. Might add 20 GeV photons for monitoring coversion recovery. (SEE: talk by Hongbo Zhu on BackTracking and conversions) Thinking still nascent and we are open to suggestions.


Download ppt "First Thoughts About Backtracking Validation Bruce Schumm UC Santa Cruz October 3, 2007 Inner Detector Software Meeting."

Similar presentations


Ads by Google