Diphoton Cross Section Measurement at CDF II
R. Blair, R. Culbertson, J. Huston, S. Kuhlmann, Y. Liu, X. Wu
We have used 207 pb-1 of data collected by CDF II to study the diphoton QCD production and compared with next-to-leading order QCD predictions : DIPHOX and ResBos. The background from neutral mesons is estimated using a statistical method based on the difference in the EM showers initiated by photons and background.
Reference to DIPHOX program can be found in Eur. Phys. J. C 16,311-330. Reference to ResBos program can be found in Phys. Rev. D 57, 6934-6947,1998 [hep-ph/9712471]. Both calculations agree very well with data in the mass distribution, except for the very low end.
Click here to see the comparison.
In the diphoton system pT (usually called qT) distribution, as common to any fixed order calculation, DIPHOX suffers infrared divergence. In particular, there is a singular point at physical region in the qT distribution due to the isolation implemetation : the isolation cut makes the qT distribution of gamma-jet final state, with a photon embedded inside the jet, a step function at Leading Order, because in this case, the isolation of the photon in jet is equal to the diphoton qT by transverse momentum conservation. At NLO, the step function is convoluted with the probability of a gluon emision. The integral of the convolution, unfortunately, is divergent. This divergence is discussed in detail in the DIPHOX reference (Eur. Phys. J. C 16,311-330). A more general analysis can be found in JHEP: 9710:005(1997). ResBos, which resums the effects of soft and/or collinear gluon emissions to all orders, predicts a smooth qT distribution.
See here for the comparisons :
DIPHOX curve is unstable at around 4 GeV, due to the divergence mentioned above - in DIPHOX, the isolation cut is implemented to parton level. The cut is set to 4 GeV in the calculation, to be compared with 1 GeV isolation cut with data. The looser isolation cut in the calculation has no significant impact on the numeric results, but improves the stability of the calculation.
At large qT, we notice some enhancement causing a shoulder structure in DIPHOX curve. This is related to the fact that DIPHOX preditcts higher rate than ResBos at small mass region in the previous comparison.
The difference is more pronounced in the dphi (azimuthal angle between the two photons) distribution.
See here for the comparisons.
At small dphi, the ResBos prediction is lower than DIPHOX in order of magnitude. In discussing with the authors of the two programs, we learnt that the discrepancies between the two predictions are expected: DIPHOX includes NLO fragmenation contributions, which populates the low DeltaPhi region, while ResBos includes fragmenation contribution only to LO.
The NLO effects in DIPHOX were discussed in Phys. Rev. D63 (2001) 114016.
We compare the mass distribution with Pythia MC here. The data points are significantly above the curve from MC.
In this plot , we scale the Pythia curve up by a factor of 2.21, and overlay it with the data points to compare the shapes, and find them pretty close.
The gluon-gluon contribution is of LO in DIPHOX for all the comparisons above. Higher order corrections for gluon-gluon contribution is now available[hep-ph/0206194]. We find it makes little difference repacing the gluon-gluon contribution in DIPHOX with the higher order one. See here .
The differential cross sections from data/Pythia/DIPHOX/ResBos are compared numerically in the table .
There were some minor changes made during the godparenting process. And it's agreed that we include three plots for publication. see the three plots here.
For questions please contact Yanwen Liu
Last modified: Tue Aug 12 16:40:41 CDT 2008