| QCD Agenda |
||||
|
December 17, 1999 1:00 PM Pump Room |
||||
|
The presentations are available by clicking on the speaker.
|
||||
 1. Announcements
|
||||
| ||||
| Blessings (I): |
||||
| ||||
| Updates: |
||||
| ||||
| Blessings (II): |
||||
| ||||
| Minutes: | ||||
|
1.1 QCD Web Pages Webmaster Rick Field has agreed to revamp the QCD web pages; if you have any suggestions please send to him; if there are any blessed plots that need to be posted, please also contact him (as well as Anwar and Joey). 1.2 APS abstracts Deadline for APR00 abstract submission is Jan 14; all of the information for submission is at: http://www-cdf.fnal.gov/internal/talks/apr00/ 1.3 QCD Papers of the Week hep-ph/9912292, Fragmentation and Hadronization, Bryan Webber; the writeup of the talk that Bryan gave at the Lepton-Photon Conference hep-/9912388, Soft gluon resummation and NNLO corrections for direct photon production, N. Kidonakis and J. Owens; resummation of threshold logarithms for direct photon production. Comparisons are made to E706 and to UA6. hep-/9912396, HERWIG 6.1, B. Webber et al., the long-awaited update to HERWIG including supersymmetry (and knobs to adjust the underlying event) 2. Status of Run2 calorimetry and jet software (Pierre Savard) The new EDM (EDM2) allows us to store C++ objects in the events. In the calorimetry there is no "E" or "S" banks (except for strips). The offline reconstruction software that uses the new event data model has been released (3.3.0). The calorimetry software was re-designed with the new EDM in mind and the calorimetry packages have been switched for 3.3.0. (Calor and CalorTest areas). The software has been debugged and extensively tested. The calorimetry simulation is still not fully functional in the new EDM. Jetclu was rewritten to use the new calorimetry package. Small discrepancies between the Fortran Jetclu and the C++ versions were reduced by a factor of 3. Pierre wrote a kT clustering algorithm based on specs by Jay Dittmann. The time needed per event is about 0.3 seconds. This is for the unoptimized version of the executable and should be about 0.1 seconds for the optimized version. Jay will maintain this algorithm. Pierre also wrote a QCD L3 filter module for the Mock Data Challenge, mostly copying what was in Anwar's JetFilter module which produces nice histograms. These modules live in the QCD package maintained by Anwar. The next steps are:
3. Finalizing comparisons of momentum distributions (Alexei Safonov) This analysis is contained in CDF notes 4883, 4886, 4996, 5165 and 5182. Alexei started off with a reminder of the MLLA + LPHD formalism. In the MLLA (modified leading logarithm approximation) formalism, the analytical results are infrared stable. This means the cutoff parameter (Q_eff) can be pushed down to the level of Lambda_QCD (~250 MeV). LPHD (local parton -hadron duality states that the hadronization occurs at the last moment so that the hadrons remember the important features of the parton distributions. The important factor is K_LPHD (N_hadrons/N_partons). Thus with the combination of MLLA+LPHD, there are two parameters to be fitted, Q_eff and K_LPHD. At the Tevatron, though, we have both quark and gluon jets and the fraction of each varies with the dijet mass range.It is assumed that.. the parton multiplicity in gluon jets is related to that in quark jets by a factor r at the same Et. The hadron multiplicity in a jet of a given energy will thus depend both on the relative fraction of quark/gluon jets and on r. Previously the charged particle multiplicity distributions as a function of xi(=log(1/x) have been measured for dijet masses from 82 GeV up to 628 GeV in CDF.There is a good qualitative agreement with some imperfections in the match. (There is some excess and the shapes do not exactly match.) A fit to all of the dijet mass bins and cone sizes shows that Q_eff is not a truly universal parameter (240± 40 MeV is the average) but tends to systematically vary within the error as the dijet mass range or cone size is varied. The simple hadronization assumption may be a bit too naive. So it may be best to think of MLLA as a "mostly correct model" with minor deviations from the data. As mentioned previously, the parton multiplicity in jets (K) depends not only on K_LPHD but also on the fraction of gluon jets in the sample. We need to determine the fraction of gluon jets in the sample. The 9 multiplicity distributions (9 dijet mass bins) were refit with 10 parameters(9 values for K(E_jet) +Q_eff). A fit for K_LPHD and r returns the values of K_LPHD=0.75±0.06 and r=1.8±0.4. The fit was performed using a correlation matrix; the correlation coefficients can not be precisely defined, though, and thus were varied within a reasonable range. The results were blessed. 4. NLO 3 jet production (Alex Brandl) This analysis is described in CDF note 4948. The purpose of this analysis is the measurement of alpha_s by comparing the energy partitioning in multi-jet events to a NLO QCD prediction from Kilgore and Giele. The data were extracted from the CDF sum ET Stream B data banks and correspond to a total integrated luminosity of 86.4 pb^-1. The analysis considers systems of three or more massless jets with the three leading jets in the laboratory frame being used as the basis of transformation into the 3-jet rest frame. In this frame the jets are re-ordered according to their energies. The 3-jet system can be characterized by the 3 jet mass, the Dalitz variables x_3 and x_4 and by the two angular variables , cos(theta*) and psi*. The conjecture is that the variation of density in the x_3 vs x_4 plane provides a measure of alpha_s. The calculation of the 3-jet cross section to NLO provides a theory that is relatively independent of the renormalization scale. The standard event quality cuts were applied to remove cosmic rays, calorimeter malfunctions, etc. Events were selected that had 3 or more jets in within a rapidity of ±2. In order to exclude data for which the acceptance is less than 95%, a cut on cos(theta*), depending on the mass of the two leading jets, is applied. The data and the NLO calculation use the same set of cuts before evaluation of the cross section. Thus the efficiencies of the offline cuts should be the same in the data and in the Monte Carlo. The analysis must account for L2 trigger inefficiencies, though. It was found that the data with sum ET greater than 320 GeV had 100% effiency at L2. This is the data that was used. Due to the detector resolution and energy mismeasurement, the jet energies are smeared. The data need to be corrected (unsmeared) for comparison with Monte Carlo calculations. There are two approaches, the Monte Carlo approach and the analytical approach. The latter is still in progress and the first is discussed in this presentation. In the Monte Carlo approach, a Herwig Monte Carlo sample was generated where all of the events before and after the application of the detector simulation (QFL) pass all cuts. The true (before QFL) and measured (after QFL) distributions in the x_3-x_4 plane were determined and a correction factor K (measured/true) was determined for each bin. The Giele-Kilgore prediction comes in two parts: a hard emission result with all of the partons well-resolved and an infrared result with one or more partons in the soft or collinear limits. The two parts are added algebraically with the theoretical errors added in quadrature. The total theoretical error results from the neglect of higher order terms and from the limited Monte Carlo statistics. Each calculation takes weeks of running to obtain reasonable statistical errrors. For this analysis 10 template Monte Carlo data sets were generated, each one equivalent to 86 pb^-1 and each with a different value of alpha_s. The Monte Carlo data sets were compared to the data. Both the data and Monte Carlo are binned in x_3-x_4 space in 0.02X0.02 bins (about 220 nonzero bins). The binned Monte Carlo cross sections are multiplied by the luminosity for the data set to obtain the expected number of events in each bin. The data is compared to the Monte Carlo using the method of maximum likelihood. For each of the pdf Monte Carlo sets, a chisquare is calculated taking into account the difference between the data and the Monte Carlo value in each bin. The chisquare values are plotted as a function of alpha_s and a minimum is searched for. The result is relatively flat from alpha_s values of <0.1 up to 0.1125 with a small dip near 0.116. It's not clear why the chisquare distribution has the shape that it does and not a quadratic shape. The next steps:
The procedure adopted is similar to that used for the dijet mass cross section. The x_3 and x_4 distributions are unsmeared separately. The bin width used for each variable is 0.005. The observed spectrum for each x bin is the convolution of the true spectrum and the detector response function for that bin. In the unsmearing procedure, the energies of all jets in the measured x distributions are corrected using JTC96S. A set of values of x^true (for x_3,x_4) that spans all of the kinematically allowed region are chosen. These events are run through QFL to find the corresponding distributions for x^measured. Normalized to unit area the ratio of the two distributions is the detector response function. The response function is then fit to the sum of 3 Gaussians. The true spectrum is then parameterized and the chisquare between the corrected distribution of x^measured and the convoluted distribution in all bins is minimized, allowing the determination of the parameters of the true spectrum. A correction factor can then be defined for each bin of x_meas as the ratio of the smeared spectrum for that bin over the true spectrum.The data is then divided by this ratio to give the unsmeared , fully corrected values. 5. Soft/hard interactions in minimum bias events (Franco Rimondi) Franco presented an update on the analysis characterizing particle production in soft and hard interaction events at 1800 and 630 GeV. The minimum bias data are divided into two samples based on clustered energy using both tracking and calorimeter information. A calorimeter jet (cal-jet) is given by standard JETCLU (Et>1.1 GeV,R=0.7) and a track-Jet is defined by a seed track of Pt>1 GeV with at least one extra track (pt>0.4 GeV) in an delta eta X delta phi <0.7 X 0.7 region. The two samples are:
The tracking efficiency, measured using Pythia-MB+QFL is constant as a function of \eta (98\%), and is about 1.0 as a function of $P_T$ for $P_T< 4 GeV but drops gradually to 85\% at P_T=10 GeV. The efficiency is about 1.0 up to track multiplicity of 20 but drops to 80\% at mult=40. The data has been corrected for these inefficiencies. Using these data, following conclusions are drawn.
6. Diffractive dijet production (Ken Hatakeyama) This analysis is detailed in CDF note 4920. Ken updated the analysis of the run 1C dijet data taken with Roman Pot detector. The aim of the analysis is to determine the structure function of (anti)Proton participating in a the di-jet events with forward-going particle, presumably an anti-proton. He has already shown that structure of the $\bar p$ factorizes in \ksi and \beta where
\ksi = (p-p')/p where p and p' are momenta of anti-proton
before and after collisions
\beta = p_parton/(p-p') where p_parton is momentum of parton
participating in hard collision
The structure function of anti-proton observed in diffractive
di-jet production is given by
Fjj= const/[(\ksi^m)*(\beta^n)] The rate of jet production is lower the predicted using H1 structure functions indicating that H1 diffractive structure functions are not valid at Tevatron i.e. not universal. Since last blessing, Ken has made a few changes in the event selection criteria and also made following changes.
1. The normalization of SD and ND data are re-evaluated using
current data set. This results in 22\% decrease in SD/ND ratio.
2. For ND parton distribution, GRV98LO is used instead of
GRV92LO and the average Q^2 used is 75 GeV^2 which is closed to
average Et^2 of CDF data.
3. A better parameterizations (to guide MINUIT) were used:
SD events
a) R(x) = ----------- = P_1 (x_pbar/0.0065)^{-P_2) (new)
ND events
= P_1 x_pbar^{-P_2} (old)
b) F^D_{jj} = = P_1 (\beta/0.1)^{-P_2} (new)
= P_1 (\beta)^{-P_2} (old)
Apart from normalization, there is no significant change in
the results, although all the plots have changed because of
minor changes in the event selection chriteria.
The conclusions (as before) from this analysis are: F_jj(\beta,\ksi) is evaluated from the experimental quantity R_{SD/ND} (\x_pbar,\ksi), i.e. the ratio of diffractive di-jet observed to non-diffractive di-jet observed in x_{pbar} and \ksi bins. The x_{pbar} is measured from the Et and \eta of the jets in the event and \ksi is measured from Roman Pot spectrometer. The non-diffractive parton distributions are evaluated using GRV98L at the mean Q^2 of the sample. The F_jj(\beta,\ksi) factorizes in \beta and \ksi and depends on F_jj(\beta, \ksi_i) = A/\beta^n where n is = 1.04\pm 0.01 and independent of \ksi_i. \ksi_i is the ith bin in \ksi. F_jj{\ksi, \beta_i) = B/\ksi^m where m= 0.92 \pm 0.02 and independent of \beta_i. where \beta_i is the ith bin in \beta. The derived F_jj(\beta) is compared with F_jj from H1 fits and found to have similar a \beta distribution but a factor of 20 too low in normalization. The updated plots were blessed. |
Joey Huston - January 29, 2000