QCD MEETING
                        April 8, 1999 at 4 PM
                                Black Hole
 
NEWS/ANNOUNCEMENTS
 
1.      General news and announcements/Run II news
 
 
BLESSINGS
 
2.      Some photon plots                       Steve Kuhlmann  10 min
 
UPDATES
 
3.      Rapidity gaps in minimum bias events    Mary Convery    30 min
 
4.      Comparisons of jet cross sections with theory
                                                Steve Kuhlmann  20 min
 
 
1. News/announcements
 
1.1 EPS abstracts due on May 3, 1999 (note change of date)
 
Possible abstracts include:
 
-inclusive jet cross sections at 1800 and 630 GeV
-inclusive photon cross sections at 1800 and 630 GeV
-differential dijet cross sections
-dijet mass cross sections
-photon + muon cross section
-NLO 3 jet comparisons
-W + jets (including R10, differential jet
results)
-MLLA comparisons to jet fragmentation
-diffractive dijet production at 1800 and 630 GeV
-diffractive heavy flavor and J/psi production
-your topic here...
 
1.2 Change of QCD meeting schedule
 
To emphasize the growing concentration of the collaboration on Run II
preparations, there is a proposal to have Thursdays devoted to these
topics with the physics meetings on Fridays. The QCD meeting would be
from 1-4 PM on Friday. This would take place on June 1,1999.
 
There may also be some restructuring of the physics groups with the
QCD group taking on more responsibility.
 
Note that the next on-week is a Special Topics week and there will be
no QCD meeting.
 
1.2 Papers of the week
 
hep-ph/9904232: Testing TeV Scale Quantum Gravity Using Dijet Production
at the Tevatron; Prakesh Mathews, Sreerup Raychaudhuri, K. Sridhar; Dijet
production at the Tevatron including effects of virtual exchanges of
spin-2 Kaluza-Klein modes in theories with large extra dimensions is
considered. The experimental dijet mass and angular distribution from
the Tevatron are used  to set a stringent limit (~1.2 TeV)
 
hep-ph/9904237, 38, 39; Diffractive light quark jet production at
hadron colliders in the  two gluon exchange model; Diffractive
gluon jet production at hadron colliders in the two gluon exchange model;
Large transverse momentum direct photon production i
 
 
1.3 Run II news
 
As onset of blessings is mostly over and physics analyses are winding down,
we will be trying over the course of the rest of the year to step up
reviews of Run II triggers/software during the QCD meetings.
 
If you have a student who can work on Run II QCD related issues
(for example on the Run II physics analysis debugging),  please
contact Anwar and Joey.
 
If you could use  an Italian summer student for the same, please
contact Anwar and Joey.
 
 
 
 
BLESSINGS
 
2.      Some photon plots                       Steve Kuhlmann
 
        Steve presented two plots for blessing that showed the Run 1B photon
cross section (previously blessed) compared to some theory predictions.
Included in the theory predictions were comparisons to the NLO QCD prediction
using the CTEQ5M pdf and
to NLO QCD predictions with accompanying parton showers (Baer/Reno). As
noted before, there is a shape discrepancy between the data and NLO theory
predictions at low pT. The predictions with the parton shower included do
change the shape of the prediction, leading to an increase at  low ET. The
shape deviation in the data (from NLO QCD) falls off more rapidly with
increasing pT, though, than does the deviation produced from the addition of the
parton shower. It was suggested to Steve that he try comparing to the Gaussian
kT smearing curve in the recent paper (hep-ph/9808367), which uses an
average kT per incoming parton of 3.5 GeV/c (determined from CDF diphotons).
That curve  was later shown to fit the shape of the data almost perfectly.
 
        The plots were blessed.
 
UPDATES
 
3.      Rapidity gaps in minimum bias events    Mary Convery
 
        Mary Convery described the work she has done with Dino Goulianos
on extracting a soft double diffraction signal.
In Regge theory/phenomenology, in single diffraction in p-bar(p)
interactions, one of the interaction particles emits a "pomeron" and
dissociates. There is a rapidity gap between this cluster and the p (p-bar).
In double diffraction, the p and p-bar exchange a pomeron and both
dissociate,
resulting in two clusters of particles with a rapidity-gap in between.
Note that the same topology is observed in the jet-gap-jet events where
these clusters have a jet each. Soft-double diffraction study may also
help us understand the survival probability of in the jet-gap-jet events.
Gap generation in both (with or without jet) classes of events may have
the same physical origin.
 
Mary used run 1A and 630 GeV minimum bias data with 0,1 vertex.
After all quality cuts she has 1021k events at 1800 GeV and
1670k events at 630 GeV.  Using a 2D plot of eta-max left and
eta-max right, the DD signal is clearly visible.
 
To extract the DD cross section, she defines a gap to be a region
without any a track or a tower above threshold. The gap is required to
trasverse the eta=0 point. The analysis is done for
different gap size requirements and results are compared with
Regge predictions.  The measured rate is about factor of 10 lower
than the Regge prediction using standard pomeron flux and is
in reasonable agreement if the renormalized flux is used. For example,
requiring a gap>1.6 both on +/- eta side, the measured rate is
0.975+\-0.033 compared to 9.7 +2.5 -2.0 using standard flux and
0.72 +0.18 -0.15 using the re-normalized flux.
 
The main effort in the data analysis went into understanding the
detector response to low energy particles. Using Suren's low-energy
tower calibration (obtained using underlying event in Pythia ND events
for his diffractive W analysis) did not give a very good agreement in
eta-max distribution.  The data is
biased to lower eta values. This bias is attributed to inadequate simulation
of the shower-spreading and loss of particles in the cracks. Mary moved
the eta-max associated with a MC particle by +/-15 cm randomly to simulate
the shower-spreading in transverse direction. The eta-max selection will
always pick a tower close to \eta=0. The new distribution shows a better
agreement but it is still not perfect. A new set of eta-dependent
calibration constants were determined for the data already shifted in eta.
Using new constants, the data and the MC improved. The systematic
uncertainty of +/- 50% on the calibration constants used to evaluate
the change in the DD cross section.
 
The analysis is described in detail in CDF-4947 and will
be presented for pre-blessing in the next QCD meeting.
 
4.      Comparisons of jet cross sections with theory   Steve Kuhlmann
 
        Steve continued the discussion of the comparison of the jet cross
sections to theory. He pointed out  that there are four main issues that
we have been dealing with:
 
1. The inclusive jet cross section is not a great fit to any smooth
curve; should comparisons be "normalized" to the best that can be
obtained, i.e. the physics function curve?
2. Should the "corrected" statistical errors be modified when the
systematics are moved in the "best-fit" methods?
3. When comparing the jet cross sections to theoretical predictions,
extremely  different conclusions are reached from visual comparisons
and from the two main statistical tests.
4. The two main techniques depend on the the arbitrary choice of bin
sizes.
 
 
        Some of these points were discussed during Eve's talk at the previous
QCD meeting (3/25/99). In this presentation, Steve concentrated on points 3
and 4.
 
        One of  the most interesting plots that illustrates the counter-
intuitive character of the statistical comparisons is to plot (data-theory)/
theory for the inclusive jet cross section compared to the CTEQ4HJ and MRST3
pdfs. One sees generally good agreement with CTEQ4HJ with a small remaining
excess at high pT (13% in the last point) and a small deficit at low pT (-7% in
the first point), leading to a total of a 20% shape difference between data and
theory. MRST3 is much further off in normalization (36%) and shape, with a total
shape difference of 65%. An evaluation of the chisquares for the two pdf's
taking into account the correlated systematic errors, though, results in a
chisquare of 54.8 (for 33 DOF) for MRST3 while CTEQ4HJ has a chisquare of 53.7,
virtually identical results. Two possible conclusions are: (1) we don't know
what we're doing and (2) the systematics must be large and look more like MRST3.
 
        A comparison of the largest systematic at high pT (the high  pT pion
response) to the comparison with the MRST pdf shows that the systematic error
is much smaller than the deviation and not the same shape. A comparison of the
CTEQ4HJ prediction to the data indicates that the deviation at low ET looks
very much like the CDF underlying event systematic.
 
        Steve then went through a simplified example to illustrate point #4.
One has a very high statistics data sample with only  1 systematic, a flat 5%
luminosity uncertainty. Suppose that this data is a flat 10% above one theory
(using PDQ pdf's)and a flat  20% above a different theory (using XYZ pdf's).
Consider what happens in the standard "best-fit" method of  evaluating the
probability. One takes the statistical chisquares and adds to it the sum of
the squares of  the shifts of the systematics after a Minuit minimization.
Assume that the statistical chisquare is approximately the same as the number
of bins. Then if the number of bins is 500, the statistical chisquare is
around 500 and the systematics would add 4 units of chisquare to this, for a
total chisquare of 504 for 500 DOF's. The probability of a chisquare of 500 is
49% while the probability of 504 is 44%. Thus, the effect of shifting the
systematic by 2 sigma has essentially disappeared. But, if a smaller number
of bins had been chosen, the difference between the statistical and total
probabilities would grow. Thus, the final answer depends on an arbitrary choice
of bin sizes.
 
        Consider a modification to the best-fit method. Define a total
probability P_tot=P_stat*P_sys and then minimize -ln(P_tot) using MINUIT.
P_stat is the normal probability from the statistical chisquare for N DOF, while
P_sys =P1*P2*P3*...*Pn, where P1 is the probability for a shift of systematic
error 1 of a given amount or beyond, etc. The statistical probability is
evaluated after the data is shifted
by the n systematic shifts. This technique would give "intuitive" results for
the toy example considered above. When applied to the inclusive jet  data
sample, the results are as follows:
 
PDF             Stat chisq Pstat(%) Psys(%) Ptot(%) Nsigma deviation
 
STD curve               32.2       100      98      99.6        0.0
CTEQ4HJ                 46.7       11       69      8           1.4
MRST3(g down)           39.6       40       0.06    0.023       3.5
 
        This technique does give results that are more intuitive and independent
of the number of bins chosen. It probably works well in the case of having  one
systematic error. It essentially asks the question of how well the theory can
fluctuate to the observed data and beyond.
 
        It's not clear how well it works in the presence of multiple
systematic errors, where there  can be cancellations among the errors leading
to a lower chisquare.