Information

MC is generated using a gamma filter, modified from one by Mike B.  I use the following cuts:

 mConeRadius = 0.20;//approximately 3X3 cluster
 mSeedThreshold = 7.0;//
 mClusterThreshold = 10.0;//
 mEtaLow = 1.086;//low endcap eta
 mEtaHigh =2.00;// high endcap eta

I am running y2006h geometry tag, Pythia v6.4.22 with Perugia tunes.  I set CKIN(3)=15 and CKIN(4)=25

I look at various plots for gamma candidates with pT>14 GeV and <24.  This is high cut is for two reasons: (1)only the one partonic pT bin above contributes appreciably and (2) I don't need to worry about filter bias this high above the threshold.  I'm also well above the trigger threshold.  Note: initally I planned to go lower, but I bumped up the filter thresholds when I re-ran after the TPC fix in order to get events faster.

I run StGammaMaker, StEEmcSlowMaker and StTriggerSimuMaker.

I generated 100K events.  I got  2011 events with gamma candidates that passed the trigger simulation.  I require candidates with pT between 14 and 24 GeV.   I only count SMD strips with energy >=1MeV.  There must be at least one strip hit in both planes.

Plots

Update 19 January 2010

 

 

 

Update 17 January 2010 II

It's definitely true that there are lower energy strips in the data than in the MC.  I stored the energy of the lowest energy strip in the data and found it to be, on average, 8.8e-05, in MC it is  1.1e-04.   That's a 25% difference.

However, it seems like the problem with the slow simulator flags really made a big difference. 

 

 

Update 17 January 2010

I re-did 100k events with the geometry fixed.  I did not get much better agreement with the bug fix.  The problem with the SMD really seems to stem from the number of hit strips not being correct. 

The plot below shows a histogram of the energy deposited in the u-plane strips.  Cuts are only on candidate pt and the trigger simulation.

both distributions are normalized to 1, so you can see the there are a lot more low energy strips in the data than in the MC.  There are a couple of possibilities here: (1) LOW_EM calorimeter option is not actually being turned on.  I have double-checked this, and all cuts are set at 10 keV as needed.  (2) Problem with slow simulator.  I talked to Pibero.  His understanding is that smd strip energy actually comes directly from GEANT, so should not be affected by slow simulator.  However I did have some settings incorrectly set that could be messing up the trigger simulation so I have changed this in the gamma tree maker macro.  (3)What does a 10 keV tracking cut actually translate to in term so the energy?  There is a point at which the MC will not be realible and we don't know that that is. 

 

In trying to find the appropriate cut in the strip energy, I have redone the strip summation when there is at least 2 MeV of energy in the strips.  The number of strips fired appears below.  Note: at the time I was trying to require at least 5 strips fired.  I have decide to just change this to 1 for now.

 

This is still bad.  I'm re-doing the MC gamma trees with the slow simulator settings fixed and looking at few more plots to determine where the cutoff should be.

 

 

 

Note: Everything below was generated with the TPC bug described here:

http://www.star.bnl.gov/HyperNews-star/protected/get/starsimu/439/1.html

Update:  Required sum of tower pT> =3 and cut out 1 degree around sector edges and re-did the plots.  Also added eta and phi plots.  Binning on phi plots is a little too fine.

 

 

 

 

 

Note: Strange that some data seems to have 0 entries in tower sum plots.  I'm going to re-run and require that the sum of the tower pT > 3.  Any other ideas?