QCD Meeting
Feb 25,1999 3 PM Black Hole
AGENDA
NEWS/ANNOUNCEMENTS
1. General news and announcements/Run II news 20 min
-discussion of upcoming QCD meetings
PREBLESSINGS
2. Run 1b inclusive photon cross section Dana Partos 30 min
BLESSINGS
3. 630 GeV jet cross section/xT scaling Alex Akopian 30 min
4. alpha_s from inclusive jet Christina Mesropian 20 min
5. Photon + muon (rebless) Koichi Kurino 10 min
6. MLLA comparisons to jet fragmentation Alexei Safonov 20 min
UPDATES
7. Minimum bias studies Niccolo Moggi 30 min
1. General news and announcements
1.1 upcoming meetings: the next month will be extremely busy with
a number of preblessings and blessings needed for LaThuile, Moriond etc; we
will have meetings on March 11 (special topics week; details still not
worked out) and March 18 (off-week).
1.2 Movement of qcd files to cdfsga complete thanks to Simona and Liz.
1.3 EPS abstracts due on April 1 (no fooling); think about any new results
That may be available in that time frame.
1.4 We need speakers for DIS99; sign up and help Frank not have to give all
Of the talks himself.
1.5 QCDWB workshop will be held at Fermilab on March 4-6
PREBLESSINGS
2. Run 1b inclusive photon cross sections Dana Partos
Dana prsented the long awaited work on the Run 1b inclusive photon
cross section. The results are summarized in CDF note 4910. As a reminder,
the Run 1A results indicated good agreement
with NLO QCD predictions at moderate to high ET, but an indication of a
devation at the lowest values of ET. This deviation was interpreted as an
another indication of the effects of soft gluon radiation (kT).
There are two methods used to separate photons from the neutral meson
backgrounds:
-shower shapes using the CES (used below 36 GeV/c)
-conversion probabilities using the CPS(used above 36 GeV/c)
The photon fraction is calculated by a statistical background subtraction.
The CES efficiency is defined as the fraction of events with a chisquare of <4
(out of the sample which has a chisquare of < 20). The CPR efficiency is defined
as the fraction of events which have a CPR hit. This is a very powerful tool
that two techniques can be used to check the photon cross section; the two
techniques should give equivalent results. They did (to within 2% in Run 1A) but
do not in Run 1B. The profile method gives a result considerably higher than
the conversion method at low ET.
As a calibration tool, pi0's from rho decays were used to check
background efficiencies. Events were chosen for this study from the 10 GeV
and 23 Gev isolated photon samples. ELES clusters were combined with tracks
with a requirement that the pi0 PT be greater than 23 GeV and track pT
be greater than 1.4 GeV with cos(theta*)<-0.92 (where theta* is the angle
between the track in the rho CM frame and the rho in the lab frame).
The mass distribution is fit to a Breit-Wigner + background curve.
The rho peak region is from 0.6-0.95 GeV and the background region
from 1.7 to 2.0 GeV. A further cut on the pi0 pT of 25-36 GeV is made.
The expected CES results come from the QFL rho simulation; the expected CPR
results are obtained by modifing the GetCprweight routine by updating the
UE and pi0 only background source. A signficant difference is observed
between the data and the expected values.
The expected efficiency (fraction of time there is a CPR hit) for
pi0's is 0.838+/-0.006 while the observed efficiency is 0.865+/-0.007.
The expected efficiency for the CES (fraction of events with chisquare <4)
for pi0's is 0.472+-0.005 while the observed efficiency is 0.453+/-0.01.
The reason for the discrepancies is not known. No pT dependence to the
efficiency discrepancy is observed.
The effect(s) on the signal efficiencies is unknown. The best quess
at this moment is that the signal efficiencies are changed in the same way
as the background. Thus, the background and signal efficiencies in the
weighting routines must be changed to agree with the rho data.
When both signal and background efficiencies are modified in the above
manner, the resulting photon cross sections agree among the two techniques
and with the Run 1A results.
The photon data come from the 10 GeV isolated photon trigger
(86.92 pb**-1) nominal prescale value of 75, calculated prescale of
77.2; used below 30 GeV/c),the 23 GeV isolated photon trigger (86.92 pb**-1,
no prescale, used between 30 and 55 GeV/c) and the 50 GeV non-isolated
photon trigger (80.05 pb**-1, no prescale used above 55 GeV/c).
All cuts (except for the isolation cut) are the same as in Run 1A.
Efficiency Sys uncertainty
-abs(eta)<0.9
-chisquare<20
-pho94 fiducial cuts 0.64
-2nd CES cluster < 1 GeV 0.893 3.1%
-no tracks 0.943 1.0%
-missing ET/ET<0.8 0.983 1.3%
-abs(Zvertex<60 cm 0.925 3.1%
-energy in cone 0.4<1 GeV 0.897 0.3%
In addition, there is a luminosity uncertainty of 7% (using old ntuples;
will be reduced). The total systematic uncertainties from the above sources is
8.4%. In addition, there are systematic errors associated with the background
subtraction. One of the largest error sources is whether the same
efficiency corrections should have been applied to the signal as to the
background. The systematic error is taken as the difference between applying
the correction to
the background only or to the signal and background both. The total error
associated with CES/CPR efficiencies and backgrounds is 9% at 16 GeV/c
increasing to 13% at 100 GeV/c. The net effect of the systematic errors is
a normalization uncertainty of 14%.
When compared to "Run 1A theory", the data show an excess at low ET
with agreement above 30 GeV/c. The theory has evolved though since the Run 1A
wesults. Changing the isolation cut results in a change in the cross section
by 10%. Lowering the cutoff of the 2->3 body interactions in the code
changes the normalization scale by 5%. Going from CTEQ2M to CTEQ4M changes
the slope at low ET by 15%. In addition, Vogelsang has made some additions
to the theory, adding NLO fragmentation and letting the renormalization,
factorization and fragmentation scales floata.
When the data is compared to the (new) Owens predictions with
CTEQ4M, mu=pT, a slope is still evident at low pT but the data is about
15% below the theory at high pT. Vogelsang can flatten out some of the slope
at low pT
by the inclusion of the NLO fragmentation contributions and the scale floating.
The latter indicates that the theory has this flexibility, not necessarily
that these are the appropriate choices of scales.
BLESSINGS
3. 630 GeV jet cross section/xT scaling Alex Akopian
Alex presented the blessing talk for another long-awaited result.
To bring the reader to date, a concern had arisen late last year about the
response function for 630 GeV at low ET. The response function at 1800 GeV
had been studied in some detail and the assumption was made that the response
was similar at 630 GeV. There may be some differences because of the different
fractions of quark and gluon jets in the two samples, but there was evidence
(such as the energy loss plot) that indicated that the functions were similar
at the two energies. The response function at 630 GeV was calculated from Herwig
and found to be significantly softer than the 1800 GeV data at low ET. The
softer response function led to larger corrections to the cross section at
630 GeV and the un-anomalying of the xt scaling anomaly.
The Herwig response function was also examined at 1800 GeV and found
to have the same soft response at low ET. In addition, a study was made of
the fragmentation at 630 GeV and the energy loss parameter (sum over all
charged particles in a jet of (P_i-Cal(P_i)/Jet ET where P_I is the momentum
of a charged particle and Cal(P_I) is the calorimeter response to that
charged particle. The results were found to be the same at both energies
in the data.
So conclusions from this:
1. Herwig response functions do not depend on sqrt(s).
2. Calorimeter response to jets in data does not depend on sqrt(s).
3. Herwig response functions are softer than SETPRT for low ET(<50 GeV).
So SETPRT response functions should be used in the unsmearing procedure
for the 630 GeV data and the xt scaling anomaly remains. The results are
summarized in CDF note 4890.
The changes since last blessing:
-BADRUNS removed
-integrated luminosity calculated as 576 nb**-1
-EM corrections at 630 GeV changed to 1.85%
-systematic uncertainties evaluated for 630 GeV
-new Run 1B jet cross section used for xT comparison
The data in the 630 GeV analysis comes from two triggers, the Jet5
and Jet15. The Jet5 trigger efficiency is determined from MINBIAS events
and the JET15 trigger efficiency from Jet5. The integrated luminosity for
the JET5 trigger is 3.2 nb**-1 and the integrated luminosity for the JET15
trigger is 483.8 nb**-1.
The chisquare for the comparison of the measured to the smeared cross
section is 3.15/DOF (with two points at around 25 and 35 GeV/c contributing
a fair portion of the chisquare).
The corrected cross section at 630 GeV is less than the EKS prediction,
except perhaps at the highest ET values. When the scaled cross section is
formed with the 1800 GeV jet cross section, the result is the famous xT
scaling plot. The result is in agreement with the previous result from 546
GeV and in disagreement with the NLO QCD predictions.
The systematic errors were determined from the standard 8 sources
for both the 630 GeV jet cross section alone and also the xT scaled cross
section.
The result was blessed.
4. alpha_s from inclusive jet Christina Mesropian
Christina presented the blessing talk for an analysis to determine the
value of alpha_s from the inclusive jet cross section using JETRAD. The value
of alpha_s is determined at each value of jet ET using the JETRAD analytic cross
section (for a given parton distribution) and a global value of alpha_s(M_Z)
is determined by a fit over the jet ET region of 40-250 GeV/c. This result
shows the running of alpha_s over a wide kinematic range.
One of the complications is the coupling of the value of alpha_s with the gluon
distribution in the relevant kinematic region. This result is
summarized in CDF note 4892.
The prime result is obtained with the CTEQ4M pdf. The value of alpha_s
determined is 0.1129 with a statistical error (fit only) of +/-0.0001. The
input value of alpha_s for the CTEQ4M pdf is 0.116. Comparisons have also
been made for a number of other pdf's such as the CTEQ4A series, and the MRST
and MRSA' series.
The experimental systematic errors were evaluated using the
standard sources of jet systematic errors.
The result was blessed.
5. Photon + muon (reblessing) Koichi Kurino
Koichi came back for a reblessing of the result previously blessed on
12/17/98. The results are summarized in CDF note 4822. The major change
was the addition of a new theoretical prediction for photon + b production.
One might ordinarily think that photon + b production would be small compared
to photon+charm production due to the heavier b quark mass and the charge of 1/3
compared to 2/3 for charm. However, the requirement for the signal (in
addition to the photon) is the presence of a muon with a pT greater than
4 GeV/c. B quarks are more efficient at producing high pT muons than are
c quarks which makes the b quark contribution non-negligible. A series of
plots were blessed showing the comparisons to photon + c and photon + b
production.
6. MLLA comparisons to jet fragmentation Alexei Safonov
Details of this analysis can be found in the minutes of the QCD meeting
from 2/18/99 and in the two CDF notes 4883 and 4886.
There are two stages of jet evolution: a perturbative phase with
kT>Qo and a non-perturbative hadronization phase. One of the goals of the
MLLA formalism is to extend the perturbative phase as low as possible.
The predictions based on MLLA calculations describe a wide range of
intra-jet features with essentially no free parameters: Q_eff (the
perturbative cut-off scale) and a constant = N_hadron/N_parton (the number
of hadrons per parton). Most of the distributions are related to the
variable E_jet*theta/Q_eff, where theta is the opening angle around the jet
axis.
The advantages of performing this analysis at CDF is that the dijet
masses go up to 600 GeV, greatly increasing the range of study. In CDF, we
can also explicitly check E_jet*theta/Q_eff scaling. In addition, there is
the possibility to analyze data samples enriched either by quark or gluon jets.
The complications for this measurement to be done in CDF is that the
background environment is more complicated due to the underlying event and
extra MB interactions. In addition, we have to worry about the CTC
inefficiencies, especially inside the core of the jet.
Some of the event selection criteria are described below:
-# of jets allowed in the event from 2 to 4
-if 3 jets, ET_jet3<0.05*(ET_jet1ET_jet2)
-if 4 jets, (ET_jet3+ET_jet4)<0.05*(ET_jet1+ET_jet2)
-abs(eta_jet1,eta_jet2)<0.9
-well balanced two leading jets
-# of vertices of class 12 no more than 2
A question came up in the preblessing regarding this cut: Alexei
showed that the distributions looked essentially identical if only 1 good
vertex were allowded.
-abs(z_vertex)<60 cm
-if 2 vertices, abs(z1-z2)<12 cm
-theta angle in center of mass system from 45 to 135 degrees
To take into account the background in the jet cone from secondary
events in the same crossing, underlying event from the jet event, incoming jet
radiation, CTC inefficiency, etc, a complimentary cone subtraction is performed
defining a cone at a right angle with respect to the dijet axis.
High ET jet events were selected from the 20, 50, 70 and 100 GeV jet
triggers. Nine dijet mass bins are defined with the lowest mass value at
72 GeV and the highest at 740 GeV.
If one calculates the multiplicity inside a cone of radius 0.5 with
Herwig, one sees that the multiplicity increases linearly with dijet mass.
The multiplicity for the uncorrected CDF flattens out. This flattening
remains even with the nominal CTC efficiency corrections. This suggests that
The nominal CTC efficiency corrections are incorrect.
A detailed study of the CTC efficiency was undertaken. A real track
was picked from one of the jets in the event. The track was reflected into
the second jet around the vertex position and the track was embedded
with the calculated parameters into the second jet. A determination was
made whether the track was found after retracking, with no constraint for
the track to have the same parameters. This procedure is repeated with
all of the tracks in the event. A histogram is built containing the
parameters of all the newfound tracks with another histogrm being formed
containing the parameters of the embedded track. The ratio of the two
histograms gives the desired correction function.
One variable that can be used to check the new efficiencies is the
fraction of the jet energy carried by the charged particles. With application
of the new efficiency correction factors, the corrected CDF result lies
closer to the Herwig result but does not completely agree. There are indications
that tracks are not completely recoverd at the smaller radii close to the jet.
However, most of the MLLA predictions deal mostly with soft hadrons.
The resulting distributions of the charge particle momenta spectra
(plotted versus xi=ln(1/x_p) where x_p=p_track/E_jet)
are in good agreement with the MLLA predictions. The peak position of the xi
distribution increases linearly with the product of M_jj*theta in good
agreement with the MLLA prediction.
The data is also compared to Herwig (+QFL). The xi distributions are in
good agreement for data and Herwig, if the Herwig (+QFL) distributions are
scaled by a factor of 0.94, i.e. Herwig (+QFL) predicts too high of a charged
track multiplicity.
UPDATES
7. Minimum bias studies Niccolo Moggi
A brief upgrade of the status of the Min Bias analysis done by
Bologna was given.
The Min Bias sample (runs 1A, 1B and 1C) was studied separately for
events that contain no tower clusters of Et>3 GeV and for events with at least
one of such clusters (the clusters were measured both in the Central+Plug
calorimeters and only in the central calorimeter corresponding to the
CTC tracking region).
The dispersion D of the mean event Pt was first studied. "D" is
defined by:
D(<pt>ev) = <pt**2>_mult - <pt>**2_mult / <pt>**2_sample
where <pt**2>_mult is the square of the mean event Pt computed for events of
fixed multiplicity "mult" and <pt>**2_sample is computed for all the events
of the sample (it is a normalization factor).
The dispersion as a function of 1/multiplicity seems to decrease to values
compatible with zero for mult-->infinite only for the sub-sample which does not
contain Et clusters (indicating the absence of non-statistical fluctuations in
the <pt>ev from event to event). The subsample which does contain
Et clusters, viceversa, shows an increasing D with high multiplicities.
This can be finely reproduced by Pythia.
The dispersion of the subsamples with no clusters show a very good agreement
at the two c.m.s. energies of 1800 and 630 GeV (the plots almost perfectly
overlap).
The correlation of track<pt> with charged multiplicity was also studied.
Again, the plots of the no-cluster subsamples overlap at the two energies.
Finally the raw charged multiplicity distribution was analyzed in the KNO form
for the two subsamples separately (the KNO form is obtained by plotting the
probability P(mult) versus z=mult/<mult>).
The multiplicity distribution shows a remarkably good scaling with the c.m.s.
energy limited to the no-cluster subsample and a strong violation in the other
subsample.
All these facts seem to suggest that the "soft" Min Bias component is
characterized by some statistical features which are independent of the total
energy of the system.
Joey Huston - March 19, 1999