Direct Photon Analysis

This page documents the work on the prompt photon production in the Standard Model group. The current goal is to have a CSC-like note available as soon as possible.

Meetings in indico: Direct Photon Meetings
Mailing list: Direct Photon HyperNews

-- AndrewHamilton - 19 Sep 2007

Interested People

The list of people who have stated an interested in direct photons at ATLAS:

• Tere Dova (UNLP)
• Martin Tripiana (UNLP)
• Andrew Hamilton (Geneva)
• Francesca Bucci (Geneva)
• John Wilson (Birmingham)
• Mark Stockton (Birmingham)
• Paul Newman (Birmingham)
• Bob Blair (ANL)
• Larry Price (ANL)
• Sergei Chekanov (ANL)
• Valeria Perez-Reale (CERN)

Photon Note

The note is accessible via the DirectPhoton package in CVS. To edit it do the following (email AndrewHamilton if you want commit rights):

• set CVSROOT environment variable

  export CVSROOT=:gserver:isscvs.cern.ch:/local/reps/atlas

  (or just set up ATHENA)
• cvs co -d DIRNAME groups/StandardModel/DirectPhoton/PhotonNote (the DIRNAME directory will be created with content of the CVS PhotonNote directory)
• commit changes and add files with the usual CVS commands:
  • cvs add -m "blah, blah, message" filename (if file doesn't already exist)
  • cvs update
  • cvs commit -m "blah, blah, message" filename
• Please turn "Word Wrapping in Text Modes" off - this helps significantly with the readability of the tex file! If you leave it on, then save the file, it will add newlines with whatever width your emacs window happened to be.

For consistency of plots throughout the note, there is a rootlogon.C and a rootalias.C file on CVS. Please put them in your working directory, ROOT will load them by default.

Relevant Datasets

MartinAANs
MarkAODs

Efficiency Plots

Please post any efficiency plots that you feel are relevant for the note. This way we can have a collection of plots to compare and be sure everybody is getting similar results.

kt kick and pt range

Where should we put the lower threshold of the photon pt to avoid (assuming we want to) theoretical uncertainties in the kt of the photon. The link below is a preliminary summary of some work by Fred Olness. Here is a quote from Fred's email: "The short answer is for gamma/Z production the curves merge at about PT=20 for Tevatron, and PT=30 for LHC. What does this mean for direct photons? At Tevatron, the curves merged in the range of PT=30 to 40 GeV. If I scale by the same factor, I'm probably around 60 GeV at LHC. So, if you are interested in the gluon, then you'll need to definitely look above 40 and as high as 60 or more. Working down to lower PT, this will be an excellent test our our ability to really understand (theoretically) the soft physics." olness_kt.pdf: preliminary study by Fred Olness.

Proposed analysis stategy: Use the region PT > 60 for PDF measurements, use region PT < 60 to look at kT problem. Note that we will also need the diphoton measurement to study the kT problem. Joey Huston has written summary of PDF measurements at LHC which looks at this kT problem, although somewhat old, here it is: huston_pdfsAtLHC_1999.pdf.

Another consideration are trigger thresholds, see TriggerPhysicsMenu. The 10^31 menu contains a prescaled g10 and unprescaled g15, g15i, g20, g20i, g25, g105 thresholds. Maybe Martin has some wisdom regarding trigger efficiencies...

Pi0 rejection

Plotted pi0 variables against photon variables (using 20GeV CSC samples): gamVsPi0.ps

The variables plotted are from the PhotonParamDefs.h:

  etcone       =  ET in a cone of R=0.45 in EM calo
  etcone20   =  ET in a cone of R=0.20 in EM calo
  etcone30   =  ET in a cone of R=0.30 in EM calo
  etcone40   =  ET in a cone of R=0.40 in EM calo
  etaBE2       = eta from the second sample
  et37          =  ET in 3x7
  e237         = uncor. energy in 3x7 cells in em sampling 2
  e277         = uncor. energy in 7x7 cells in em sampling 2
  ethad1      =  ET leakage into 1st sampling of had calo
  weta1        =  corrected width in 3 strips in the 1st samp.
  weta2        = corrected width in 3x5 cells in the 2nd samp
  f1              = fraction of energy found in 1st em sampling
  e2tsts1      = 2nd maximum in strips
  emins1      = energy of strip with minimum between max 1 & 2
  wtots1       = total width in em sampling 1 in 20 strips
  fracs1        = energy outside core (E(+-7)-E(+-3))/E(+-7)

Useful Papers

• Hollin's Thesis: Ian Hollin's 2006 ATLAS Direct Photon Thesis (on CDS)
• diphoton_cdf.pdf: CDF diphoton measurement (NB: qqbar->gamgam is larger contribution than gluglu->gamgam)
• yanwen_thesis.pdf: Yanwen Liu's CDF diphoton thesis (ps version)
• handbook.pdf: The CTEQ Handbook (version 1.1)
• owens_talk.pdf: J. Owens 2004 CTEQ summer school talk
• cdf8708_dipho_mass_prl_v2.ps: CDF high mass diphotons (with RS limit)
• hepph9808467_kTeffects.pdf: Huston et. al. on kT effects on Direct photon production
• Phys. Rev. D70, 074008 (2004).: Carter Hall's CDF measurement of direct photons using conversions - an obscure but interesting attempt
• CDF note 5721 (after fixing broken ps): Dana Partos' CDF thesis explaining the calibration and validation of the photon background evaluation using CPR and CES
• huston_pdfsAtLHC_1999.pdf: Huston PDFs at LHC
• olness_kt.pdf: Olness preliminary work on kT problem

Software to Facilitate Getting High Statistics Background Samples

The package PhotonGenFilter has been checked into CVS (groups/StandardModel/DirectPhoton/PhotonGenFilter). This is a generator filter that finds a cluster of photons and checks to see if they satisfy an isolation cut in both and in the fraction of energy around the cluster. A fragment of python for setting up the filter is here:

#--------------------------------------------------------------
# Filter Options (multi photon filter with isolation for photon bkg. sample)
#--------------------------------------------------------------
GamBkgFilter = Algorithm( "GamBkgFilter" )
GamBkgFilter.RClus = 0.18 # the region of the gamma like cluster
GamBkgFilter.RIso = 0.4  # the size of the isolation cone
GamBkgFilter.IsoCut = 2000. # iso Et cut if Et is less then this in MeV it passes
GamBkgFilter.FracIsoCut = 0.1 # fractional iso cut if fraction is less it passes
GamBkgFilter.FilterMin = 1 # minimum number of candidates
GamBkgFilter.FilterMax = 9999 # maximum number of candidates
GamBkgFilter.CandPtMin = 20000. # minimum pt for a candidate in MeV

#--------------------------------------------------------------
# Pool Output Options
#--------------------------------------------------------------
Stream1 = Algorithm ("Stream1")
Stream1.RequireAlgs += ["GamBkgFilter"]

Note that it passes events that have from "FilterMin" to "FilterMax" candidate photon clusters. In this way it can be used to select events with one or more candidate clusters or as an anti-selection to prevent double counting events when adding a set of filtered events to an unfiltered set (allowing them to be exclusive and permitting weighting of each set by 1/ L).