[Halld-offline] [EXTERNAL] request for help from tracking group

Simon Taylor staylor at jlab.org
Wed May 6 10:58:23 EDT 2020


The errors on the hits are underestimated for this sample of simulated events.  I am investigating.

From: Halld-offline <halld-offline-bounces at jlab.org> on behalf of Naomi Jarvis <nsj at cmu.edu>
Sent: Wednesday, May 6, 2020 9:57 AM
To: Richard Jones <richard.t.jones at uconn.edu>; HallD Software Group <HallD-Offline at jlab.org>
Subject: Re: [Halld-offline] [EXTERNAL] request for help from tracking group


I think there might be something wrong with the FOMs from the timebased tracks but I'm not really sure what I should expect.   The plots below are for timebased tracks in the CDC with no reaction specified.  The code snippet is below.  It's from the plugin CDC_TimeToDistance, maybe there's a mistake in here.

I have been looking at track DOCA with time and found that selecting larger FOMs made very little difference to this, although so far I only looked at values from 1e-8 to 0.1. I have a few jobs with minimum FOM between 0.1 and 1 still running.

   vector <const DChargedTrack *> chargedTrackVector;


   for (unsigned int iTrack = 0; iTrack < chargedTrackVector.size(); iTrack++){

      const DChargedTrackHypothesis* bestHypothesis = chargedTrackVector[iTrack]->Get_BestTrackingFOM();

      // Cut very loosely on the track quality

      auto thisTimeBasedTrack = bestHypothesis->Get_TrackTimeBased();

      if(!thisTimeBasedTrack->IsSmoothed) continue;

      if (thisTimeBasedTrack->FOM < MIN_FOM) continue;  // was 1e-10



On Wed, May 6, 2020 at 7:39 AM Richard Jones <richard.t.jones at uconn.edu<mailto:richard.t.jones at uconn.edu>> wrote:
Hello all,

I have been looking into differences between geant4 and geant3 simulations of k+ tracks at backward angles in the cdc, and have discovered what appears to be a sudden failure of the tracking past about 120 degrees in the cdc. Please see the final plot and comments here.


>From hdview, I see that the number of hits on the track is very good up through DTrackTimeBased, and visually things look quite clean but the FOM values are terrible and there are no combos being generated. I am not sure, but don't combos need to be generated for these tracks to go further down the reconstruction chain? A set of 100 such tracks is available at:


-Richard Jones
Halld-offline mailing list
Halld-offline at jlab.org<mailto:Halld-offline at jlab.org>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20200506/c7a40d7d/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: resid_vs_fom.png
Type: image/png
Size: 15933 bytes
Desc: resid_vs_fom.png
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20200506/c7a40d7d/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: resid_vs_fom.png
Type: image/png
Size: 16743 bytes
Desc: resid_vs_fom.png
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20200506/c7a40d7d/attachment-0005.png>

More information about the Halld-offline mailing list