QCD Meeting
** Minutes **
May 2, 2002
2:00 PM Pump Room
1. News/announcements

a. Simona is now Dr. Murgia.

b. The Les Houches accord #2 is now finished. The program and documentation are available at pdf.fnal.gov in a form that is very easy to incorporate into MC/ME programs. Lynn Garren will make it available as a ups product so that the accord can be easily incorporated into any Fermilab product.

c. Abstracts submitted to Amsterdam

QCD radiation of ttbar events
Inclusive jet production at CDF
Inclusive photon production at CDF
Run 2 jet algorithms at CDF
Direct photon cross section from conversions
Quark and gluon jet fragmentation at CDF
Underlying event in hard scatterings
Multi-gap diffraction
Charm meson cross sections

d. Papers of the Week

hep-ph/02040244, Matching NLO QCD Computations and Parton Shower Simulations, S. Frixione and B. Webber; a method is proposed for matching NLO QCD calculations with MC simulations using a subtraction technique. An example is given for WW production.

hep-ph/0204316, The QCD/SM Working Group Summary Report from Les Houches; 153 pages, summary of work on higher order calculations, pdf's and uncertainties, resummation, Monte Carlos, photon cross sections, jet algorithms; includes description of both Les Houches accords (MC's and pdf's)

hep-ph/0205006, QCD as a theory of Hadrons, outline for a book to be released by Cambridge University Press

2. Looking at MC/looking at data - Frank Chlebana

Frank showed the results of some work he has been doing in comparing the shapes of jet distributions generated with MC. The Herwig generator and the runMC executable in the mcProduction package were used. Distributions were generated with different pdf's (but not with the newer pdf's since a newer PDFlib is not linked in). 500,000 events with pT>20 GeV were generated. Plots of jet eta1 vs eta2 showed shape differences when generated with different pdf's. Two particular distributions were compared for which the pdf-derived differences are somewhat larger: ratio of same side jet cross section to opposite side cross section, and the ratio of the central to forward jet cross section. The next steps include: combining MC samples in order to cover a wider ET range, trying to quantify the shape differences between pdf's, and finally comparing the results to data. Frank also showed some results from looking at the data. He updated the inclusive jet ET plot using 13.7 pb^-1 of data (Physics table PHYSICS_0_02[6,xxx] and later, since these have the CLC requirement removed from the jet triggers). Simple cuts were applied to clean up the sample (Etot < 2000, Metsig < 6, |Vx| < 60). The jet ET distributions before vertex and jet energy corrections go out to ~400 GeV. Frank plotted the ratio of the lead jet ET distribution for the forward and central jet eta regions as well as the jet ET distribution with the requirement that there are two jets on the same side/opposite side of the detector.

3. Cosmic ray background and HadTDC spectra - Igor Gorelov

Igor presented an analysis of data from the Central Hadron Calorimeter taken in special single beam runs several weeks ago. The CHA has a background at high ET from cosmic rays depositing energy in the detector. The CHA has timing information from TDC's that allow for discrimination against the cosmic ray induced depositions.

The data sample has been stripped selecting JET70 or JET100 trigger paths only, corresponding to ET_jet >70 or 100 GeV. Events with ET_miss/sqrt(ET) > 7.0 are classified as cosmic ray induced whereas events with ET_miss/sqrt(ET)<4.0 are classified as physics. Only towers with E(tower) > 0.75 GeV are considered. Igor presents plots of the TDC start times for events in the two categories: cosmics (ET_miss/sqrt(ET) > 7.0) and physics (ET_miss/sqrt(ET) < 4.0). The distribution of start times for physics events is cleaner and more narrow than the corresponding distribution for cosmics. Furthermore, the timing distribution for cosmics bears a resemblance to data from a cosmic ray run with no beam in the Tevatron. There is, however, a feature in the distribution that appears when beam is present (could this be evidence of beam halo?)

Igor describes the importance of "time slewing" and how this affects the T0 resolution: 1) it is an instrumental effect of discriminators in ASD cards, 2) it is an effect of the threshold discrimination point depending from an amplitude of a signal, 3) it is revealed as a typical banana shape of 2D amplitude vs time plot. The E(tower) > 0.75 GeV cut reduces the effect, and an additional slewing correction has been applied.

When an overall slewing factor correction is applied, a Gaussian fit to T0 for physics events yields T0 = 553.2 +/- 0.02 ns and a width of sigma = 4.86 +/- 0.02 ns (although the chisq of the fit is poor). When slewing corrections are applied on an event-by-event basis, the offset in T0 is removed and one obtains T0 = 2.08 +/- 0.01 ns and sigma = 3.77 +/- 0.01 ns.

Summary: 1) The cosmic background contributes to QCD jet triggers at the level of 10-20%. 2) The hadron TDC spectra reveal a distinct pattern of cosmic background that is uniform in time over 132ns. 3) The T0 time resolution is improved by ~1.2ns with proper channel by channel TDC offset/slewing corrections.

Igor's results are very encouraging and he is planning to pursue additional studies of the efficiency and rejection of applying a timing window. He also plans to look at data from WHA and PHA.


Run 1 Update

4. Study of inclusive double pomeron exchange in Roman pot triggered events - Ken Hatakeyama

Motivation: The suppression of the diffractive structure function at the Tevatron relative to that at HERA is generally attributed to additional partonic interactions which spoil the diffractive signature of the event. In cases of multi-gap events, if a single gap survives, additional gaps may be unsuppressed. This can be tested by looking at the rate of an additional forward (or central) gap on the proton outgoing side in events with a leading antiproton.

For events triggered on a leading antiproton in Roman pots, the xi_p value can be reconstructed from the calorimeter and BBC counter information. The DPE process is expected to yield an enhancement in the low xi_p region.

A plot of the peak of the xi_pbar variable reconstructed from calorimeter and BBC information versus the same variable reconstructed from the Roman pot track shows a linear relationship at both 1800 and 630 GeV, but the correlation is far from 1:1 (0.59 for 1800 and 0.41 for 630 GeV). If the median is used rather than the peak, then the correlation is closer to 1:1 (0.95 at 1800 and 0.67 at 630 GeV). The smaller coefficient at 630 GeV may indicate that a larger tower ET correction factor (currently 1.6 at both energies) should be used at 630 GeV.

Some changes made after the blessing on Mar 7, 2002:

Events in Run 75713 were removed from the data sample at 1800 GeV. Towers were removed from 6 hot spots in the data sample at 630 GeV.

A bump was observed in the xi_p distribution at a value of ~0.001 (at 1800 GeV). The suspicion is that the bump is due to calorimeter noise. If the threshold in the central region is raised from 200 to 300 MeV, the distribution appears smoother. So the threshold was raised.

The noise was further studied by selecting no-collision events in Run 1a zero-bias events. Several hot spots were found in these events and the towers removed. The noise level was found to decrease with decreasing luminosity.

A comparison between the data and the DPE MC (+noise) distributions for the xi_p variable shows good agreement below and xi_p value of 0.02 (for 1800) and in the range of 3 X 10^-5 < xi_p < 0.02 (for 630). If particle ET values are lowered by 15% in the calorimeter, then better agreement is obtained at 630 GeV for the xi_p range below 0.02.

Using the median values for the xi_p calibration, and the changes described above, results in a ratio of [DPE/(SD+DPE)] of 0.195+/-0.001+/-0.011 at 1800 GeV and 0.170+/-0.001+/-0.012 at 630 GeV for 0.035<xi_pbar <0.095 and xi_p<0.02.

This result is larger by a factor of 5 than the prediction of the renormalized pomeron flux model with independent renormalization for the fluxes from the proton and antiproton, but in good agreement with the predictions of the rapidity gap probability renormalization model.


Run 1 PreBlessing

5. W/Z + photon - Marina Brozovic

Marina presented a preblessing talk for her analysis which looks for W/Z + photon events where the W/Z decay into two jets. The results are summarized in the CDF notes 5698, 5893 and 5915.

The physics motivation for this study is as follows. It serves as a test of the Standard Model Electroweak theory, as a prelude to understanding Higgs production in the analagous channel (ppbar->WH,ZH), as a test of theories which predict more gauge bosons, and with higher statistics it can probe anomalous gauge boson couplings.

The data set is from Run 1b and has an integrated luminosity of 90 pb^-1. The high pt central photon trigger was selected and the GHIB data stream used. The photons are required to be central, within the fiducial volume of the CEM calorimeter, isolated, have ET>23 GeV and pass a number of standard photon quality cuts. Each event is required to have 2 and only 2 jets with ET>10 GeV (with a detector eta of +/- 2.4). For the analysis the requirement on the jet ET values is raised to 15 GeV. Both jets are required to have more than 1 track in the CTC and the jet vertices are required to be within 10 cm of each other. The jets are required to have a separation in eta-phi space greater than 2.

The photon variables are calculated using the vertex information from the two jets. The algorithm to calculate the jet vertex was tested using a sample of Z(->e+e-) + 2 jet events.

The Monte Carlo used to generate the W/Z + photon sample was Herwig, with leading order matrix elements for the subprocesses (from Uli Baur) being added to the program. The cross sections time BR for W(->jet jet) + photon and Z(->jet jet) + photon are essentially the same (about 6.5 pb). About 200,000 events were generated with 8500 passing the cuts on the photon and jets. The shape of the dijet mass distribution from the W+Z fits well to a Gaussian with an rms of about 12 GeV. The combined efficiency for W/Z + photon, where the W and Z decay into two jets, with ET>15 GeV and |eta_d|<3 is on the order of 4.2%.

Backgrounds for these processes come from 2 jet plus photon processes and 3 jet events, where one jet fakes a photon. The 2 jet plus photon backgrounds was generated using the LO matrix elements from Owens + Herwig (+ QFL detector simulation). The 3 jet background was generated from Pythia 2->2 jet processes, where the 3rd jet arises from the radiation of a gluon.

The two backgrounds are of a comparable size for this sample, but the prediction is very sensitive to the choice of the LO renormalization scale. The dijet mass distribution arising from the mixture of the two processes fits well to an exponential (for dijet masses greater than 60 GeV).

The systematics are discussed in CDF note 5915. The total efficiency for the W/Z + photon analysis (with all cuts on the photon plus jet) was found to be 4.2%. The cuts important to determine the systematic uncertainty are:

-jet detector pseudorapidity
-number of 3D tracks allowed in combination with the cuts on track pt and isolation
-trigger efficiency
-cuts on jet ET and detector eta
-jet vertex cuts
-jet-jet delta_R cut

The combined systematics produce an acceptance with uncertainty of 4.2% + 0.5% - 0.4%.

The limit on the cross section is determined by using a 3-part likelihood function. The limit was set by minimizing the chisquare fit with respect to the slope of the background, integrated luminosity, detector efficiency, and W/Z + gamma cross section. The expected data distribution is some combination of the signal and background. The shapes of the signal and background were parameterized (the signal as a double Gaussian and the background as an exponential). The output of the minimization fit was:

-sigma_W/Z + gamma = 19 + 32 ?19 pb
-total_efficiency = 0.042 +/- 0.0015
-luminosity = 90 +/- 1 pb^-1
-background slope = -0.0306 +/- 0.0009

The Standard Model predicts a ~13 pb cross section at NLO. The 68% CL limit cross section is <44 pb; the 95% CL limit cross section is <81 pb.
Joey Huston - May 13, 2002