sPHENIX discussion of software

[Anthony Frawley <afrawley AT fsu.edu>]

Hi All,

Below are my notes from the discussion we had on Thursday morning about preparations for the January workfest from the tracking side.

There will be further discussion in the simulations meeting on Tuesday.

Cheers

Tony

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

Who do we know will be there to work on tracking?

Tony, Christof, Ross, Hugo, Jin, LANL several people, Sanghoon

Christof: Andy Salzberger is not guaranteed yet (and we learned later he will not be able to come).

Jin said he would work on ACTS, geometry implementation, also HF

LANL will presumably be focused on the MVTX.  Jin suggested getting them involved in multi-vertex finding and Rave tuning, that would be great!

Account access for visitors seems unlikely on this time scale. Christof is trying singularity as a way to get computer access for our visitors.

Before the workfest

  - Some hit finding efficiency has been lost, Tony will try to understand that.

  - Truth tracking is not 100% efficient, Tony will look at that.

  - Other evaluator issues - uninitialized variables.

  - multi-vertex tracking - exists in code now using MVTX vertices, is it good enough for what is needed for HF studies? Check before the workfest that we are happy with existing multi - vertex tracking, so we can generate files in advance (we can do final vertexing later).

       - We need a benchmark to define "happiness" with multi vertexing.

       - Final Rave multi-vertexing needs considerable tuning.

       - Can LANL group take this up?

  - Get iterative tracking back in place - Christof

  - Christof: Track seeding is 95% efficient, track propagation is failing somewhere so we only get 75% in AuAu events.

      - momentum assignment to seeds from Hough gives very low mom sometimes, seed cut at 0.2 GeV that may eliminate real tracks

We need fancier TPC clustering eventually, but it is not a top priority for now

Tracking package interface for ACTS

  - ACTS is really the only package we see as suitable at present

  - Decided that Tony will create shell of parallel infrastructure ahead of meeting.

     - Will need feedback from Andy Salzberger on details, Christof will pursue

  - vertexing in ACTS - need interface from/to current vertexing

TPC distortion simulations - Ross was not called in - talk to him offline

  - what are our goals for workfest?

     - Comparison of sPHENIX model with ALICE model to validate

  - What prep is needed before workfest?

Hugo's work on outer detector - also talk to him offline

  - Christof thinks we will have no alternative to an outer detector, but we will have to quantify the need.

  - We need enough information to try to make a case by the end of the workfest.

    - We need performance parameters defined

  - Talk to Hugo about what can be done in prep for workfest so we can do this

  - Start to think about a real detector?

    - what would it cost?

    - 2 years is already tight for making it

  - John Haggerty commented "A proposal for an sPHENIX upgrade to measure space points  outside the TPC would require thorough justification of the design needs (position resolution, solid angle coverage, segmentation, number of points), and a detector design that could plausibly meet them.  Such a detector cannot interfere with or compete against the completion of the baseline MIE detector, and the TPC is presently scheduled to be completely instrumented by September, 2021, about 20 months from now. Additional funding sources would be needed for such a detector, since it is unlikely that additional resources can be extracted from the other sPHENIX projects.

Evaluator

  - Look into existing problems (matching to truth tracks, uninitialized variables) before the meeting

  - Fix the current evaluator, don't try to redo it before the meeting.

  - Hugo did some work on it, should we pull his changes in?

Tracking QA

  - Jin would be able to help a volunteer set it up

Calibration and infrastructure - outline a plan in workfest

  - Start to think through how to do it.

  - Involves geometry and data storage containers.

  - When we rotate or translate a detector, all points have to rotate or translate with.

  - Christof asked David Rohr for what they do in ALICE, see what he says.