sPHENIX tracking discussion

[Haiwang Yu <yuhw.pku AT gmail.com>]

Dear all,

I think I figured out what was the cause of the wrong extrapolation.

As we guessed in the tracking meeting it was caused by some bugs I made in the interfaces to GenFit.

Attached is a (GenFit_Prediction - Cluster) / Sigma plots for rphi and z for 2 GeV tracks.

I will have a more complete follow up soon.

I want to send this out as soon as possible just don't want you guys worrying too much. 

I think we are in a good shape.

Best,

Haiwang

On Tue, Apr 25, 2017 at 2:09 PM Frawley, Anthony <afrawley AT fsu.edu> wrote:

Here are my notes from today's tracking meeting.

Cheers

Tony

TPC simulation update - Carlos Perez
----------------------------------------------------
Carlos discussed the developing new TPC simulation scheme. This has one big gas volume, G4 controls the step size using the G4 step limit.
Therefore the cloud diffusion and the effects of space-charge distortions are no longer limited by restriction to a layer, so can move the centroid in radius.
The step size limit is 0.3 cm.

Slide 2: Jin: Please discuss with me naming convention and how to preserve the truth tracing capability. Should have hits in TPC directory, with truth hits also there.

Christof: Pad model here is rectangular, how does this correspond to chevrons?
Carlos: Rectangle used to make the problem analytical. Can play tricks to use this scheme to emulate charge sharing in chevrons.

The tracker should treat the clusters as a black box, hits are corrected for everything so they can be compared with truth.
Christof raised the point that the trackwer needs the errors defined in a plane. Is that information preserved in this process? Carlos: will have to discuss.

Slide 4:
Introducing these details into the simulation may help us to understand occupancy effects.
Takao said that the Samba pulse is too sharp (rise time is really 160 ns), and gaussian shape is not quite right. OK, will change it
Jin: Focus on 10 MHz sampling for now, 20 MHz may be a stretch.

Carlos: The new code is mature nut not yet ready to be merged.

Discussion of existing simulations:
Presently we are seeing ~75 microns cluster resolution. Carlos: that is too small, we expect no better than 100 microns, no worse than 200 microns resolution.
Carlos will push the resolution up to 150 microns in the existing TPC simulation as a priority, should be easy.

Jin: CD1 dry run on June 3rd - would like sim working before that.Probably not possible to have new TPC scheme by then, so should get the existing cylindrical simulation in a good working state for the June 3 review.

Carlos: Inner field cage may be thicker than the 1% we have in now. Need to discuss with Tom.

Updates on PHG4KalmanPatRec - Haiwang:
-----------------------------------------------------------

Slide 4:
Significant fraction of tracks have only ~ 10 hits. Presumably because of a problem with seed -> track stage.
Christof: Is it the track parameter or the error assignment that is spurious? Plot it as absolute distance to see.
Haiwang: from log files, the displacement is 2 cm from layer to layer. That is too large. Sounds like a problem in the seeding interface?

slide 7: very wide search window gives only 70% cluster association efficiency? Seems like it is not sensible.

Christof:
The Kalman fitting efficiency will never be better than the track seeding efficiency, so a likely place to look is the transition from seeds to track fitting. Christof and Haiwang will pursue this offline.

Note: Everbody should check that the dead area for the silicon layers in their macros is set to 100% for now, since we require all 7 si layers for the track seeds.

Iv was suggested that we generate Hijing events with fixed geometery that can be used to test without re-running the G4 part of the simulation, then we can re-run the reconstruction over and over on those files. We will need to freeze the detector setup in the macros first. Also increase the TPC errors to 150 microns. Freeze with the cylinder setup, using 40 layers.

Discussion of freezing macros will continue at the simulations meeting today.

---

From: Frawley, Anthony
Sent: Sunday, April 23, 2017 10:56 PM
To: sphenix-tracking-l AT lists.bnl.gov
Cc: Frawley, Anthony
Subject: Special sPHENIX tracking meeting 10:00 am EDT Tuesday April 25, 2017

 

HI All,

As agreed at the tracking meeting on Friday, we will have a special tracking meeting on Tuesday April 25 at 10:00 am EDT. The link to the agenda page is included below.

At the meeting on Friday Haiwang showed a progress update on his Kalman pattern recognition work, and Carlos gave a verbal report on work on improving the TPC simulation. We will hear more details about both on Tuesday. Haiwang showed that the INTT cluster resolution is screwed up again, and I agreed to look into it. I am discussing a fix with Chris offline.

We agreed that Haiwang should should focus on his top priority, the pattern recognition and the low pT track efficiency, and leave anything else to be resolved by others.

Talk to you Tuesday.

Tony

https://indico.bnl.gov/conferenceDisplay.py?confId=3066

sPHENIX tracking meeting (25 April 2017) indico.bnl.gov To join the Meeting: https://bluejeans.com/822435626 To join via Browser: https://bluejeans.com/822435626/browser To join with Lync: https://bluejeans.com/822435626/lync To join via Cisco Jabber Video: https://bluejeans.com/822435626/jabber To join via Room System: Video Conferencing System: bjn.vc -or- 199.48.152.152 Meeting ID: 822435626 To join via Phone: 1) Dial: +1 408 740 7256 +1 888 240 2560(US Toll Free) +1 408 317 9253(Alternate Number) (see all numbers - http://bluejeans.com/

_______________________________________________
sPHENIX-tracking-l mailing list
sPHENIX-tracking-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/sphenix-tracking-l