1. How to… (a.k.a. Photon PID and Pizero Analysis) G. David, April 20, 2000 A report on the status and advertisement of the tools http://www.phenix.bnl.gov/phenix/WWW/photon/howto/howto.html Methods geared toward heavy ions, low pt Goals: service (providing tools) and education (including my own) Group: subset of Thermal Photon PWG (but anybody is welcome!) Principal players up to now: Shunji, Ken, Saskia, Craig, Gabor Focus on: photon PID Basic tool: EMCal evaluator chain, ntuples Pion spectrum: reconstructed via event mixing (one particular method) Comparison of GEANT and measured

2. How to… (a.k.a. Photon PID and Pizero Analysis) G. David, April 20, 2000 Prerequisites Processing (simulated) PRDF files with the evaluator detailed and short documentation available tested on 20,000 + 70,000 events outputs available on dataxx disks nonstandard output – GEANT and measured OR Processing DST files (currently no evaluation, just measured) in the works (Saskia) documentation coming OR Processing standard reconstruction/evaluation chain output planned both GEANT and measured

3. How to… (a.k.a. Photon PID and Pizero Analysis) G. David, April 20, 2000 The “Nishimu-Oyama method” Framework (compiled code) gg2pi to establish photon PID cuts and to check their efficiency use GEANT PID or PID based upon measured quantities Boolean (yes/no) cuts primary cut: slewing corrected TOF with E-dependent sigma secondary cut: charge overlap tertiary cut: an empirical combination of cluster characteristics compared to an ASCII database easy to change and tune output: ntuple with all accepted photon pairs (single and pair data) can provide output with event mixing applying the same cuts mltslice split gg2pi input according to event multiplicity mkhist generate invariant mass histograms from gg2pi output in bins of pt manipulate the output yourself (example provided)

4. How to… (a.k.a. Photon PID and Pizero Analysis) G. David, April 20, 2000 Example: how to check the efficiency of your photon PID? - bit 1 is set if GEANT PID is “photon” - bit 2 is set if cluster passes your TOF cut - example only: doesn’t mean you cannot do better! Nishimu-Oyama method

5. How to… (a.k.a. Photon PID and Pizero Analysis) G. David, April 20, 2000 Nishimu-Oyama method Example: 2000 events, PID with measured quantities - processed with gg2pi and mkhist, plotted with pi0_pac_plot.C pt range 1.0 – 1.5 GeV/c 2.0 – 2.5 GeV/c

6. How to… (a.k.a. Photon PID and Pizero Analysis) G. David, April 20, 2000 Nishimu-Oyama method 20,000 events GEANT PID, deposited energy GEANT PID, reconstructed energy measured PID, reconstructed energy

7. How to… (a.k.a. Photon PID and Pizero Analysis) G. David, April 20, 2000 The “Szillasi method” Framework (CINT macros to help establish probabilistic “combined” cuts) ntread.c select a set of measured quantities A, B… appropriate for PID plot their distributions for photons and non-photons get the probability that for a given value a,b… it’s a photon p(a), p(b)… assign P = w(A)p(a) x w(B)p(b) x … to each cluster as “probability of being a photon” (currently w(A)=w(B)=…=1.0) your main job will be to find proper A,B quantities and w weights you can now use this probability P to tag your photons the histoes and an ntuple with the data of all particles is written ntopti.c using the output histoes of ntread.c it calculates the efficiency and the cleanness of your cuts for all values of P The code is very transparent: easy to read and modify