[Hps-analysis] FEE Track Matched Clusters

Wed Oct 7 16:11:16 EDT 2015

Something is wrong with the GBL track states, in that the default track state (target) points to the hit in the last layer better than the track state at the last layer.  Obviously this is wrong, although I’m not sure yet whether the problem is with the track state at the target or the one at the last hit, or both.  I’ll be looking at this, although I’m amidst a broader review of the entire track reconstruction (with plans to overhaul major portions of it) which is taking up my time at the moment.

In any case, extrapolation of the target track state is the best thing we currently have and extrapolation errors at the ECal are small compared to ECal cluster position resolutions anyway, so this shouldn’t be a serious issue (and certainly not one to explain Holly's E/p discrepancies).

-Tim

> On Oct 7, 2015, at 11:23 AM, Omar Moreno <omoreno1 at ucsc.edu> wrote:
> 
> Final state particles are made out of both GBL and seed tracks.  The type of the particle (ReconstructedParticle of HpsParticle) is set to the type of the track so you should be able to select whichever tracks you want to use.  The momentum of the FS particle is equal to the momentum of the track associated with it.
> 
> A word of caution, though; when extrapolating GBL tracks to the Ecal, the GBL track parameters at the target are used.  This is incorrect and, as a result, FS particles using GBL tracks may be mismatched to a cluster. What we really want to do is extrapolate the track using the GBL track parameters at the last layer with a hit in the SVT.  This requires some additional code which, I believe, Tim is working on.  Furthermore, we still haven't checked the agreement between the GBL tracks in Java and those obtained using the C++ refit.  Matt said the GBL tracks look fine so I'm assuming they are OK, but we should do a thorough comparison.
> 
> --Omar Moreno
> 
> On Wed, Oct 7, 2015 at 11:01 AM, Sho Uemura <meeg at slac.stanford.edu> wrote:
> You could try making the plots with seed tracks and with GBL tracks. The GBL tracks should show better top-bottom agreement on momentum scale, and they should also show better resolution (tighter E/p distribution). If you see neither of those differences, there might be something wrong with the track type code.
> 
> Yes, the new pass3 detector should make seed tracks and GBL tracks agree better. But bad things will continue to happen if somehow your analysis is using seed tracks instead of GBL, or a mixture of both.
> 
> On Wed, 7 Oct 2015, Holly Vance wrote:
> 
> So I re-ran my code checking that the track type is GBL, and the plots look
> the same. It seems that these effects will be specifically addressed in
> this upcoming pass.
> 
> On Wed, Oct 7, 2015 at 12:25 PM, Nelson, Timothy Knight <
> tknelson at slac.stanford.edu> wrote:
> 
> So the GBL track isn?t used for this collection? (I though GBL ?healed?
> 
> the v2 momentum scales).
> 
> T
> 
> On Oct 7, 2015, at 9:18 AM, Sho Uemura <meeg at slac.stanford.edu> wrote:
> 
> SeedTracks in v2 geometry have big momentum shifts like what you're
> seeing. GBLTracks have better momentum in v2 than in v1.
> 
> We think this is because v1 has systematic misalignments across the
> detector (e.g. L1-3 vs. L4-6), and v2's misalignments are due to
> measurement error and are less correlated from sensor to sensor. SeedTracks
> are very sensitive to alignment of the first layers of the SVT, so the
> momentum measurement is worse in v2. GBLTrack momentum is in a sense
> "averaged" over the length of the track, so the momentum measurement is
> better in v2.
> 
> SeedTracks are very sensitive to misalignment in
> 
> On Wed, 7 Oct 2015, Nelson, Timothy Knight wrote:
> 
> Hi Holly,
> 
> It turns out the v2 (survey) alignment used in pass 2 is no better (and
> possibly worse) than the v1 (as designed) alignment (the assembly error
> appears to have been as small as or smaller than the survey error, where 50
> microns is a big effect).  Nonetheless, you shouldn?t be seeing those
> effects with either one, so something is obviously wrong.    I?m not sure
> which track collection is used for the FP Particle collection, but our best
> fits for FEE should have errors at the 1% level, not the 10% level, in both
> top and bottom.
> 
> Sho?  Omar? Pelle?
> 
> Tim
> 
> On Oct 7, 2015, at 8:20 AM, Graf, Norman A. <ngraf at slac.stanford.edu>
> wrote:
> 
> Good Morning Holly,
> I?m sure the experts will also chime in, but I?d like to point you to
> the presentations at
> yesterday?s SVT meeting where there was quite a bit of discussion of
> alignment.
> https://confluence.slac.stanford.edu/display/hpsg/10.6.2015+Weekly
> Omar had also shown some slides (not yet posted) from his analysis of
> the v3
> detector, which I assume he will be showing either tomorrow or Friday.
> Norman
> 
> From: Hps-analysis [mailto:hps-analysis-bounces at jlab.org] On Behalf
> Of Holly Vance
> Sent: Wednesday, October 07, 2015 7:24 AM
> To: hps-analysis at jlab.org
> Subject: [Hps-analysis] FEE Track Matched Clusters
> 
> Hi,
> 
> I analyzed the skimmed FEE events from pass 2 studying track matched
> clusters at the Ecal. I made a very short summary in the attached pdf, and
> I was wondering if anyone from tracking can comment on why the momentum for
> these tracks is 10% high/low.
> 
> How well do we know the SVT planes positions? I assume the survey
> cannot tell us this precisely.
> 
> -Holly
> 
> 
> _______________________________________________
> Hps-analysis mailing list
> Hps-analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hps-analysis
> 
> 
> _______________________________________________
> Hps-analysis mailing list
> Hps-analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hps-analysis
> 
> 
> 
> 
> --
> BEGIN-ANTISPAM-VOTING-LINKS
> ------------------------------------------------------
> 
> NOTE: This message was trained as non-spam.  If this is wrong,
> please correct the training as soon as possible.
> 
> Teach CanIt if this mail (ID 01PqgpVwH) is spam:
> Spam:
> https://www.spamtrap.odu.edu/canit/b.php?i=01PqgpVwH&m=e987565e4a79&t=20151007&c=s
> Not spam:
> https://www.spamtrap.odu.edu/canit/b.php?i=01PqgpVwH&m=e987565e4a79&t=20151007&c=n
> Forget vote:
> https://www.spamtrap.odu.edu/canit/b.php?i=01PqgpVwH&m=e987565e4a79&t=20151007&c=f
> ------------------------------------------------------
> END-ANTISPAM-VOTING-LINKS
> 
> 
> 
> _______________________________________________
> Hps-analysis mailing list
> Hps-analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hps-analysis
> 
> _______________________________________________
> Hps-analysis mailing list
> Hps-analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hps-analysis