Measurement of the Inclusive Isolated Prompt Photon cross section in pp collisions at  √s=1.96TeV, using 9.5 fb-1.

Alessandra Lucà, Ray Culbertson, Fabio Happacher, Anna Mazzacane, Marco Trovato, Costas Vellidis, Tingjun Yang

Abstract

The measurement of the cross section for the production of isolated prompt photons in pp collisions at a center-of-mass energy √s= 1.96 TeV is presented. This analysis uses the full dataset, collected with the upgraded Collider Detector at Fermilab in Run II, corresponding to an integrated luminosity of 9.5 fb-1.  The cross section is measured as a function of photon  transverse energy in the kinematic range between 30 GeV and 500 GeV and in the central pseudorapidity region |η γ| <1. The results are compared to leading-order parton-shower Monte Carlo models and next-to-leading-order perturbative QCD calculations. Next-to-next-to -leading-order perturbative QCD calculations agree well with the measured cross sections as a function of transverse energy.


Documentation


Public Note 11180   

Introduction: Main Motivation

The production of high-ET prompt photons in hadronic collisions is an important testing ground for perturbative QCD, enabling to probe parton distribution function (PDF), and  can also provide evidence for the presence of new physics. Prompt photons also constitute an irreducible background for important searches such as H→ γγ, or SUSY and extra-dimensions with energetic photons in the final state.


Signal fraction


The main background contribution comes from the two photon decays of π0 and η which fake a single photon shower in the calorimeter. We get the signal fraction using an Artificial Neural Network (ANN) which was trained to discriminate between prompt photons and QCD background events.

FIG.1 Artificial Neural Network (ANN) output distribution normalized to unity of signal (red) and background (blue) PYTHIA MC templates, for one photon ET bin: 50< EγT(GeV)  <60 .



For every bin in the photon ET, the ANN output distribution observed in data is fitted to a linear combination of signal and background MC ANN templates, using the ROOT routine TFractionFitter (FIG.2). The signal fraction is provided with an error estimate which takes into account both data and Monte Carlo statistical errors. TFractionFitter errors are corrected based on pseudo-experiments.





FIG.2 Fits to the ANN distributions in bins of photon transverse energy  for 30 < EγT(GeV) < 500. ANN output distributions observed in the data (black points) and the distributions fit result given by the SUM of MC signal (red) and background (blue) templates scaled to TFractionFitter rates. The distributions fit result are normalized to data. 
 


FIG.3 Signal fraction as a function of photon ET . The error bars represent the statistical errors and the azure bands represent the systematic errors.




FIG.4 Total systematic uncertainty and single contributions on the signal fraction (f γ ) as a function of photon ET.



Acceptance x Efficiency

We use the PYTHIA inclusive photon MC sample to unfold data back to hadron level.
These Acceptance x Efficiency factors are computed bin by bin.

FIG.5 Acceptance x Efficiency factors as a function of photon ET. The error bars represent the MC statistical errors and the azure bands represent the systematic errors.




FIG.6 Total systematic uncertainty and single contributions on the Acceptance x Efficiency factors as a function of photon ET.



Results and comparison with theoretical predictions

For comparison we have used

  • PYTHIA 6.216 with CTEQ5L PDFs
  • SHERPA 1.4.1
    • generator with CT10 PDFs. Tree-level matrix element (ME) diagrams with one photon and up to three jets, merged with parton shower (PS) 
  • MCFM 6.8: Fixed-order NLO calculation including non-perturbative fragmentation at LO

Log scale


Linear scale


FIG.7 The measured γ + X cross section compared with three theoretical predictions: PYTHIA, SHERPA and MCFM. The vertical error bars show the statistical uncertainties, while the shaded areas show thesystematic uncertainties. The 6% luminosity uncertainty on the data is not included. A correction to account for extra activity (CUE) is applied to the MCFM theoretical predictions.




Log scale


Linear scale


FIG.8 The fractional systematic errors on the measured γ + X cross section. The continuous line is total systematic uncertainty while the dashed lines correspond to the single contribution.




Log scale


Linear scale


FIG.9 Ratio of the measured γ + X cross section to three theoretical predictions: PYTHIA (upper part), SHERPA (central part) and MCFM (bottom part). The vertical error bars show the statistical uncertainties, while the shaded areas show thesystematic uncertainties. The 6% luminosity uncertainty on the data is not included. A correction to account for extra activity (CUE) is applied to the MCFM theoretical predictions.




Log scale


Linear scale


FIG.10 Data points centered at 1 and Data/Theory ratio (lines) of the inclusive prompt photon cross section as a function of the photon transverse energy, EγT> , in the central ηγ region.The inner error bars on the data points show statistical uncertainties. The full error bars show statistical and systematic uncertainties added in quadrature. The 6% luminosity uncertainty on the data is not included. Pythia has been multiplied by a factor 1.5.




Log scale


Linear scale


FIG.11 From MCFM we can obtain predictions for the single sub-processes contribution. As expected, Compton scattering process dominates for photons with low transverse energies while the annihilation process dominates for high transverse energies.