sPHENIX tracking discussion

[Anthony Frawley <afrawley AT fsu.edu>]

Some notes from our tracking software meeting this morning.

Tony

The main topic was a discussion about trying to get ahead of some problems that may occur in the reconstruction of real data next year. Updated slides posted at:

https://indico.bnl.gov/event/20668/

The action items related to that are summarized below.

----------------------------------------------------------------------

1) Put as much survey data as we can into the initial tracking alignment constants.

  INTT is complete and ready to go.  (thanks Joseph)

  MVTX has relative chip and ladder positions, we will not have clamshell positions. (thanks Cameron)

  TPC survey data not ready yet, Ross is pushing on it with a student.

2) Develop procedures to handle missing TPC modules or groups of layers. (David Stewart volunteered)

  The goal is to not have missing modules (or groups of layers)  wipe out the track acceptance in that phi range.

  TPC seeding will need some development.

3) Restore the option to find the bunch crossing in pp running without INTT clusters in the track. (Tony)

  If we lose some INTT acceptance in 2024, recover tracks by assigning beam crossing geometrically.

  Some code exists, will need afterburner to get precise match geometrically.

4) Develop a way to find silicon track seeds with < 3 MVTX hits. (Joe)

  The goal is to avoid losing tracking acceptance if there is a missing MVTX layer in some part of the detector.

  Requires modified silicon seeding algorithm that does not begin with 3 MVTX hits.

5) Create static distortion files with misaligned E and B fields for testing. Create correction files made from distortion files with different TPC orientation. (Ross et. al.)

  The goal is to see the effect of static distortions on tracking when we do not have the static corrections right.

6) Start exercising tracking reconstruction with zero field cosmic simulations. (who?)

  The goal is to prepare for analyzing real cosmic data when online/offline chain is done.

7) Look at (and tune) the performance of TPC seed ghost rejection. (Derek)

  The present ghost rejection algorithm misses ghost tracks.

8) Develop on-the-fly modification of Acts transforms for track-based distortion corrections. (Tony & Joe)

  Try to mock-up 3D tracking in the TPC by moving fake surface to corrected cluster position.

   See if this resolves the problem that Genfit tracking gives better radial correction results.

9) Optionally implement alignment corrections in the G4 model used in simulations

  Would obviously be desirable.

  Requires a lot of work.

  Make this a goal, but not an urgent one.

We discussed two other topics:

----------------------------------------

Implementation of David's truth/reco matching scheme.

  Agreed that it is time to adopt this as the default for users who are not tracking experts.

Neural net clustering. 

  This is merged. It is relatively slow.

  Suggestion was to consider using this in afterburner mode for special cases (such as Upsilon).

  Discuss with Zhongling.