20 March 2006 19:15:26

Plan for improving vertex finding in Cu+Cu
Anthony looking at ways of telling whether vertex is likely to be the correct one based on event info: ZDC, (apparent) reference multiplicity, CTB sum (correlations between these)
I will work on get more vertex info available by modifying and re-running the vertex finder and Anthony will look at these.

• 0. ZDC vertex info. This should or could already be there. Place holder in StZdcTriggerDetector which is stored in StEvent and picked up by MuDst. Problem is that value is unfilled (according to Ant). There is a maker which can accomplish this task (StZdcVertexMaker). This may not have been given go-ahead to run in chain as it is 'too small' since it produces a single vertex.  It also depends how early in chain StTriggerData is created as at the moment it uses StEvent to get StTriggerData. Maybe the functionality should be put somewhere else - hopefully in some way that the code is not duplicated. We could compare the ZDC vertex to the TPC one and/or use it as a seed (Minuit vertex finder has a hook for this).
• 0.5 pVPD vertex
• 1. Flags from StMinuitVertex. In principle can be used to check what stage vertex-finder reached. This information is not making it out into the data. Looks like flag is set=1 to conform to current flagging scheme. There is a CVS comments about making it like this because FTPC (quite correctly) expects this.
• 2. SVT standalone info. Idea from Rene to use the SVT vertex reconstructed from SVT hits only (tracklet method?)
• 3. Match tracks. Either run PPV or borrow methods for matching tracks

Threads. Starsoft
cucu200 datasamples to study vertex finder at high luminosity (started by  Lidia Didenko Feb 13 ends Feb 24)
http://www.star.bnl.gov/HyperNews-star/protected/get/starsoft/6313.html

Cu+Cu: Investigation of pile-up (started by Dan Magestro Feb 21 ends Feb 28)
http://www.star.bnl.gov/HyperNews-star/protected/get/starsoft/6318.html

Beamline constraint in PPV (started by  Jamie Dunlop Feb 22 single mail)
http://www.star.bnl.gov/HyperNews-star/protected/get/starsoft/6320.html

CuCu vertex finder: different EVR vs Minuit (started by  Jamie Dunlop Feb 24 single mail)
http://www.star.bnl.gov/HyperNews-star/protected/get/starsoft/6323.html

Everyone (except for Jan) who replied to those posts is on vertex-hn list (once I rechecked later.) Couple of other interested people, Joern and Anthony, not on.

Looked into creating pages on drupal to document the known issues. Can have (slightly) more formal 'book' type document under Offline Computing or one can create a more informal 'blog'.
For now the latter seems best to just include links to material already there.

22 March 2006 17:22:52
Discussed vertex finding in our weekly meeting. Anthony plots refmult vs CTB sum and see a second class of events with a small CTB sum but a range of 'too large' refmults.
This second population apparently grows in the higher luminosity sample.
This technique can be used to cleanup certain centrality classes.

We came to the conclusion that the first thing to try would be using one of the fast detectors that can provide a vertex position as the seed for the vertex finder.
This could be pVPD which probably has the best resolution but maybe this isn't there for the whole run?
BBCs which must be there because they are used for triggering (check this idea though).
ZDCs were also 'always on' presumably - resolution worse but could win for non-central since they are relatively better there.

Want to try to implement something which does not depend on run due to various detectors (or components thereof) dropping in and out of run.
Even though methods employing match to finer granularity fast detectors (incl. SVT) could be better they may lead to a very inhomogenous dataset where the (centrality dependent) vertex finding efficiency depends on detector status. Would imply run-by-run vertex-finding efficiency corrections. Also potential difficulties in running embedding because e.g EMC embedding is done as a post-pass (I think).

28 March 2006 09:59:07
Post from Marco to vertex-hn list. He has been busy coding to improve seed finding.
http://www.star.bnl.gov/HyperNews-star/get/vertex/64.html  (March 27th to ...)

Extracted links from all threads to date:
Plot location on AFS (Janet Seyboth) /afs/rhic.bnl.gov/star/users/jcs/public/P05ih_ittf (actually no longer there)

Dan's page:
http://www.star.bnl.gov/protected/highpt/magestro/cucu/
- event characteristics

Jan's pages on PPV weights:
http://www.star.bnl.gov/protected/spin/balewski/2005-PPV-vertex/shortOverview/PPV-C/Slide6.GIF
http://www.star.bnl.gov/protected/spin/balewski/2005-PPV-vertex/zStabReal2005/

Jamie's 'original' plots for Cu+Cu
http://www.star.bnl.gov/protected/common/common2005/trigger2005/cu200_centrality_stability.ps
http://www.star.bnl.gov/protected/common/common2005/trigger2005/refmult_zdcx.ps

Jamie's plot from Minuit vertex finder events
http://www.star.bnl.gov/protected/highpt/dunlop/cucu200/high_vs_low/high_vs_low.ps
showing effects on refmult distribution (black is low luminosity, red is high)

Jing's page with PVPD to TPC vertex comparisons
http://www.bonner.rice.edu/~liuj/pvpd/vertex/
- was agreed that plot needed to be remade with only the minbias triggers as currently including high towers