For the cross section analysis I am using a number of Monte Carlo samples along with one data set. The details of each of these sets can be found below:
Pion enhanced, QCD 2->2 sample (full event): aka "T2 Platinum":
This MC sample was produced at Tier 2 by Mike Betancourt for me. It consists of ~200,000 events in each of following partonic pt bins {5-7, 7-9, 9-11, 11-15, 15-25, 25-35} GeV. The events were pre-filtered at the pythia level so every event contains a pythia-level pion with the following kinematic cuts: Pt > 4 GeV, -1.2 < particle eta < 1.2. The code to generate the trees lives on /star/u/ahoffman/T2_maker and is complied in SL08c. The MuDsts and .geant files are archived at HPSS. The trees themselves (600 files) live on /star/institutions/mit/ahoffman/Pi0Analysis/T2_platinum/Trees/. The following parameters were set in the analysis macro:
emcSim->setCalibSpread(kBarrelEmcTowerId,0.15);
emcSim->setCalibOffset(kBarrelEmcTowerId,0.);
emcSim->setCalibSpread(kBarrelSmdEtaStripId,0.25);
emcSim->setCalibOffset(kBarrelSmdEtaStripId,0.0);
emcSim->setMaximumAdc(kBarrelSmdEtaStripId,700);
emcSim->setMaximumAdcSpread(kBarrelSmdEtaStripId,70);
emcSim->setCalibSpread(kBarrelSmdPhiStripId,0.25);
emcSim->setCalibOffset(kBarrelSmdPhiStripId,0.0);
emcSim->setMaximumAdc(kBarrelSmdPhiStripId,700);
emcSim->setMaximumAdcSpread(kBarrelSmdPhiStripId,70);
As of now, only events which pass the software trigger conditions for trigger 137611 (HTTP L2gamma) or 117001 (mb) are saved. These events are weighted properly using Mike B's custom weight calculator for his filtered events. That code can be found /star/u/ahoffman/BetanWeightCalc/
Single Particle Monte Carlo Sets
I have three separate single particle MC samples, single pion, single eta, and single gamma. These were produced using the code located at /star/u/ahoffman/SingleParticle_platinum/. The starsim, bfc, and treemaking code is all there. The .MuDsts and .geant files that result from the bfc jobs are run through a treemaker similar to that for the full pythia monte carlo. These samples are used to estimate the background shapes in in the diphoton invariant mass spectrum. The single gamma sample, for example, is used to model the 'low mass' background, as pion candidates are found from split clusters. The following cuts were set in the macro:
emcSim->setCalibSpread(kBarrelEmcTowerId,0.15);
emcSim->setCalibOffset(kBarrelEmcTowerId,0.);
emcSim->setCalibSpread(kBarrelSmdEtaStripId,0.25);
emcSim->setCalibOffset(kBarrelSmdEtaStripId,0.0);
emcSim->setMaximumAdc(kBarrelSmdEtaStripId,700);
emcSim->setMaximumAdcSpread(kBarrelSmdEtaStripId,70);
emcSim->setCalibSpread(kBarrelSmdPhiStripId,0.25);
emcSim->setCalibOffset(kBarrelSmdPhiStripId,0.0);
emcSim->setMaximumAdc(kBarrelSmdPhiStripId,700);
emcSim->setMaximumAdcSpread(kBarrelSmdPhiStripId,70);
One important difference to note is that these events are not held to the simulated trigger standard. Instead, I only choose events (offline) that have pion candidates with Pt above the trigger threshold. Essentially this assumes that the trigger efficiency is perfect for such events. Obviously this is not true, but in my analysis these samples are used primarily to estimate the background line shapes. The single pion events are not weighted. The other two samples are weighted according to the funcional form given by the PHENIX cross section measurements.
Data
This analysis is only concerned with run 6 data, and only data that satisfies the HTTP L2gamma trigger (online and software.) I restrict myself to good runs between 7139017 and 7143025, as it is the longest period of run 6 with relatively stable tower and SMD status tables. Parts of the barrell are missing in this run range, and this will be accounted for in a geometric acceptance correction, but I believe that the stability of the status tables is more important that having a 100% live detector. Using a stable run range will cut down on the systematic error of the measurement which, given previous measurements, will be larger than the statistical error. The data was produced by Murad using my StSkimPionMaker (which can be found in StSpinPool) as part of the SpinAnalysisChain. The output trees are located at /star/institutions/mit/ahoffman/Pi0Analysis/Murads_Production_2_08/. The following parameters were made in the macro:
As of now, only events which pass either HTTP L2gamma (online and software) or MB triggers are saved. Only the L2g triggered events are used in the analysis. All other cuts are made offline and will be listed in individual analysis sections to follow.
Mixed Event Background:
I should note that I also make use of a 'mixed event' sample that is made by taking photons from different (real data) events and mixing them to form pion candidates. This sample is used to model the combinatoric background as described here.