Development and vetting of mid rapidity W-reco algo  for 2009 run

updated: 11-24-08, changes 12-5-08

A) Event samples (phase 2 in green):

1. single particle events
   • distribution: eta range [-1.2,1.2], zVertex Gauss(0,30cm), cutOff at |Z|<60cm,
   • geom=y2008
   • 4k events per job, total 100k events (1e6 events)
   • particles: muon, pi-, pi0,proton,electron 
   • fixed pT of: 10,20,30,40,50 GeV
   • custom BFC:
     • chain="trs -ssd y2008 Idst IAna l0 tpcI fcf -ftpc Tree logger ITTF Sti StiRnd  -IstIT  -SvtIt SvtCL,svtDb -SsdIt MakeEvent McEvent geant evout geantout IdTruth  bbcSim emcY2 EEfs bigbig -dstout fzin -MiniMcMk -McEvOut clearmem  VFPPVnoCTB eemcDb beamLine"
     • no pedestals, 100% working detector (ideal status tables)
     • private EmcSimulator with SF should result with :
       • BTOW: electron ET=60 GeV --> ADC=4096, uses default barrel SF of 5%
       • BPRS:  MIP (muon) --< ADC ~30, uses SF=1 and no eta dependence
       • BSMDE,P: electron ET=40 GeV --> peak strip ADC~700 , uses SF=1 and no eta dependence
     • NO private vertex finder 
   • use only scheduler, store output files on data05/ (other permanent disk TBD)
2. Pythia events
   • vertex distribution zVertex Gauss(0,30cm), cutOff at |Z|<60cm,
   • offset vertex transversely by few mm (need special flavor for beam line in BFC!)
   • custom BFC as for single particles
   • type of events:
     
     • QCD background physics events use/modify this kumac inspected by Jim
       • 4 samples with partonic pT [5-10], [10-20], [20-30], 30-40] GeV
     • W-events: kumac ,
   • produce 10k W-events and 100k QCD-backg events, 1k events per job, phase 2 100k &1M events, respectively
   • use scheduler & disk as above

B) Key elements of the W-reco algo, prioritized order

1. assure there is a 2x2 BEMC tower cluster above ET=15 GeV (serves as trigger surrogate) 
2. identify high PT reco primary track
3. assure this track matches 2x2 tower cluster with some margin, no ET to PT matching
4. transverse & longitudinal EM shower, assume continuous detector for eta[-1,1], phi=2pi:
   1. SMD-E, unpack events in to array [301x60]  in eta*phi bins , keep parallel array of dead strips, mark column 151 dead
   2. SMD-P, unpack events in to array [20x960]  in eta*phi bins, mark dead rows 16,31,....,960 (between modules)
   3. PRS and BTOW, unpack in to array [40x120] in eta*phi, mark dead towers independently in both layers
   4. algo must make reasonable decision if limited # of elements in vicinity of high pT track are dead, or discard event
5. isolation cuts around high pT track:
   1. analyze BTOW, BPRS,BSMD energy, hit pattern over variable size square centered on the high pT track  
6. run official  jet finder
   1. match one of the jets to high pT vetted track , develop match criteria, if no match discard event - this will be extension of isolation cut 
   2. develop criteria of missing ET cut using away side jet(s) from the jet finder, decide how to handle 3-jet events

 C) Evaluation technique

1. all reco code runs only on muDst. geant.root file is not used nr needs to be present on the disk.
2. demonstrate W-algo runs on 2008 real-data st_physics runs ( filter barrel trigger(s), run on all ~500 runs) , use ongoing production of st_physics events
   • muDst can be found with mySql query: get_file_list.pl -keys 'path,filename' -cond 'production=P08ie,trgsetupname=ppProduction2008||ppProduction2008-2,filetype=daq_reco_MuDst,runnumber=9067013,filename~st_physics,storage=nfs' -onefile -limit 70
3. post QA plots on designated web area from W-algo for full statistics of _all_ samples described in 1), 2), and 10 real runs, one PDF file per event type 
4. keep consistent record of status of execution of this plan in Drupal blog, possibly 3 times a week