The STAR experiment

Correction factors are derived from simulation by taking the ratio of the reconstructed primary tracks matched to MC pions divided by the MC pions. Specifically, the following cuts are applied:

Monte Carlo

• |event_vz| < 60.
• |eta| < 1.
• nhits > 25
• geantID == 8||9 (charged pions)

Matched Reco Tracks

• |event_vz|<60.
• |reco eta| < 1.
• |global DCA| < 1.
• reco fit points > 25
• geantID of matched track == 8||9

The track yields and their associated yields are obtained from the minimc files that are produced automatically with each simulation request. I run a separate chain containing StEmcTriggerMaker on the MuDst simulation files to determine if each event would have satisfied EMC and BBC trigger conditions.


There is currently a bug in StDetectorDbMaker that makes it difficult to retrieve accurate prescales using only a catalog query for the filelist. This affects the absolute scale of each cross section and data points for HT1 and JP1 relative to the other three triggers. It's probably a 10%-20% effect for HT1 and JP1. With that in mind, here's what I have so far:


This plot is generated from a fraction of the full dataset; I stopped my jobs when I discovered the prescales bug.

The cuts used to select good events from the data are:

• golden run list, version c
• |vz| < 60.
• Right now I am only using the first vertex from each event, but it's easy for me to change

The cuts used to select pion tracks are the same as the ones used for "Matched Reco Tracks", except for the PID cut of course. For PID I require that the dE/dx value of the track is between -1 and 2 sigma away from the mean for pions.

As always, comments are welcome.

Notes from collaboration meeting

The following are notes I took during the STAR collaboration meeting at MIT, specifically the spin pwg session on 7/13/06. It's not an exhaustive summary,
since I'm just "kicking the tires" on drupal.

Motivation:
Estimates of trigger bias systematic error are derived from simulation. This page compares yields obtained from data and simulation to test the validity of the PYTHIA event generator and our detector geometry model.

Conditions:

• Simulation DB timestamp: dbMk->SetDateTime(20050506,214129). I pick this table up from the DB, rather than from Dave's private directory. Dave changed the timestamps on the files in his directory, so the two files do not match. It turns out that in this case they only differ by one tower (4580), and this tower's status is != 1 in both tables, so there is effectively no difference.
• Data runlist: I use a version of the jet golden run list containing 690 runs. I have heard there is an updated version floating around, but I have yet to get my hands on it
• Simulation files obtained from production P05ih, including larger event samples from Lidia's recent email (http://www.star.bnl.gov/HyperNews-star/protected/get/starsoft/6437.html) but excluding the 2_3, 45_55, and 55_65 GeV samples.
• This is strictly a charged hadron comparison, there is no dE/dx PID cut. The dE/dx dsitributions in simulation are way off.
• Cuts: nFitPoints>25 && |dca|<1. && |eta|<1. && |vz|<60. && pt>2. Also for data I require good spin info and relative luminosity information (these conditions are mostly subsumed by the runlist requirement).

Procedure
Combine PYTHIA partonic pt samples by filling histograms with weight = sample_weight/nevents, using sample_weights

• 3_4 = 1.287;
• 4_5 = 3.117e-1;
• 5_7 = 1.360e-1;
• 7_9 = 2.305e-2;
• 9_11 = 5.494e-3;
• 11_15 = 2.228e-3;
• 15_25 = 3.895e-4;
• 25_35 = 1.016e-5;
• above_35 = 5.299e-7;

Normalized simulation histograms to data, plot yields for MB, HT1, HT2, JP1, and JP2 triggers vs. pt, eta, phi, and z-vertex. Use StEmcTriggerMaker to emulate trigger response in simulation. Also plot eta, phi, and z-vertex yields in slices of pt.

Results:
At the moment I've just linked the raw PDFs at the bottom of the page. The index in the title indicates the charge of the particle being studied. The plots are perhaps a bit hard to follow without labels (next on the list), so here's a guide. Page 1 has pt distributions for the triggers in the order listed above. Pages 2-6 are eta distributions, with each page devoted to a single trigger, again in the order given above. The first plot on each page is integrated over all pt, and then the remaining plots separate the distribution into 1 GeV pt slices. Pages 7-11 repeat this structure for phi, and 12-16 do the same for the z-vertex distributions.

Conclusions:
The agreement between data and simulation appears to be me to be quite good across the board. The jet-patch triggers are particularly well-modeled. A few notes:

• The HT2 pt distributions (page 1, third plot on top row) look funny in simulation. What's with the spike at 6 GeV in the h- plot?
• HT2 eta distribution on the east side for h+ (page 4) has spikes.
• Phi looks good to me
• Vertex distributions for calo triggers in simulation are awfully choppy, but overall the agreement seems OK.

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)