Dijets in high Q2 DIS at HERA-II
Jörg Behr, Thomas Schörner-Sadenius, Thorben Theedt (UHH/DESY)
|
Introduction
The aim of the analysis is to provide dijet cross-sections at high values of
the photon virtuality, Q2, above 125 GeV2. On the one
hand side these data provide precise tests of the predictions of perturbative
QCD and of the concepts of factorisation and the universality of the parton
distribution functions (PDfs).
On the other hand these data can be used directly in NLO QCD fits of the PDFs
to constrain these quantities further.
The analysis outlined here follows very closely an analysis of dijet
cross-sections in HERA-I data that was published recently (see below for a
link to the corresponding web page and paper). Relevant changes concern
- The data set: We use 127.XY pb-1 of data from the 2004/05
running period. During that time, HERA was operated with positrons. In
contrast, the HERA-I analysis used about 81 pb-1 of electron and
positron data from the years 1998-2000.
- The electron calibration: For the HERA-II data, we apply an electron
calibration that is based on a comparison with the DA measurement between data
and MC and a correction back to the true level. Such a calibration was not
applied in the HERA-I data.
- The hadronic final state: For the HERA-II data, a scale correction is
applied that improves the agreement between the hadronic energy scales in data
and MC and thus reduces the scale uncertainty. In addition, different jet
nergy corrections than for the HERA-I data are applied.
Plots stamped preliminary
Single-differential
Cross-sections and ratios to NLO for Mjj.
Cross-sections and ratios to NLO log10ksi.
Cross-sections and ratios to NLO etaprime.
Cross-sections and ratios to NLO for Q2.
Cross-sections and ratios to NLO for ET,mean.
Cross-sections and ratios to NLO for xBj.
Double-differential
Cross-sections for log10ksi in bins of Q2.
Ratios of data over NLO for log10ksi in bins of Q2.
Here is a table of cross sections for HERA-1/HERA-2 combined (PDF).
Presentations
Data Selection
The selection corresponds to that applied in the HERA-I measurement. To
summarize:
|
Inclusive Sample
|
|
Q2 [GeV2]
|
125-5000
|
DA method for reconstruction
|
|
cos(gammahad)
|
(-0.65) - (0.65)
|
reconstructed from ZUFOs.
|
|
Eel [GeV]
|
> 10
|
after energy calibration
|
|
isolation
|
standard cone cut
|
|
|
zvtx [cm]
|
(-32) - 31
|
|
|
pT/sqrt(ET) [sqrt(GeV)]
|
< 2.5
|
|
|
E-pz [GeV]
|
45-62
|
|
|
Dijets Sample
|
|
ET,Breit,1 [GeV]
|
12
|
Applied after hadronic scale and jet energy correction
|
|
ET,Breit,2 [GeV]
|
8
|
Applied after hadronic scale and jet energy correction
|
|
etaBreit
|
(-2) - 1.5
|
|
|
ET,lab [GeV]
|
> 3
|
|
|
etaT,lab
|
< 2.5
|
|
|
|
|
Removing events if jets too far back or too close to electron.
|
Link to paper on HERA-I data
You can find the final version of the paper as accepted by the journal here.
Link to web page for HERA-I analysis
You can find the web page containing the accompagnying material for the
dijets analysis of HERA-I data here.
Control plots
Inclusive sample
The following plots show a comparison between the HERA-II data after event
selection and the two MC samples (ARIADNE and LEPTO).
Variable set 1 linear and logarithmic scale.
Variable set 2 linear and logarithmic scale.
Dijet sample
The following plots show comparisons between the dijet distributions in the
data and in the two MC samples (ARIADNE and LEPTO). On the left is shown the
comparison for HERA-I (HERA-I analysis redone by Jörg), on the rigt for
HERA-II.
Q2, Mjj and log>sub>10ksi on linear and logarithmic scale.
Mean ET, xBj and etaprime on linear and logarithmic scale.
log>sub>10ksi in different bins of Q2 on linear and logarithmic scale.
We paid particular attention to the eta distributions in the dijets sample
(again on linear and logarithmic scales).
Purities, efficiencies, corrections, and migrations
Migration matrices for the 6 inclusive dijet variables and for
the double-differential distributions.
The acceptances, purities, efficiencies, QED and hadronisation corrections are
shown here for the 6 inclusive dijet variables and for
the double-differential distributions.
A comparison of acceptances in HERA-I and HERA-II can be found here.
Comparison first analysis (J. Behr) versus second analysis (T. Theedt)
Here you can see the comparison of cross-sections for the inclusive quantities. J. Behr's analysis in this case does not contain the additional hadronic energy scale correction. The agreement is very good.
Comparison HERA-I versus HERA-II
In the following two plots, you can see the comparison of J. Behr's analysis to the published HERA-1 data.
- Here we show the ratio between J. Behr's cross-sections and the published data for the inclusive variables and the double-diff. distributions. The comparison is done separately for using LEPTO (green) or ARIADNE (blue) for the acceptance corrections. The errors are statistical only, and all corrections and calibrations have been applied to the HERA-II data. The agreement is satisfactory, with only about a third of all points being separated by more than one sigma.
- Here we show the ratio between J. Behr's cross-sections and the published data for the inclusive variables and the double-diff. distributions. The comparison is done using ARIADNE for the acceptance corrections. The errors are statistical and systematic added in quadrature, and all corrections and calibrations have been applied to the HERA-II data.
- Here we show the acceptance corrections for J. Behr's analysis and a re-analysis of HERA-1 data (done by J. Behr) for the inclusive variables and the double-diff. distributions. The corrections are shown for LEPTO (green) or ARIADNE (blue) and for HERA-1 (lines) and HERA-2 (dots). There is a small change of about 5 percent between the two data taking periods, and the agreement between ARIADNE and LEPTO is slightly better for HERA-2 than for HERA-1.
Theoretical uncertainties and further details
For all questions concerning the theory please refer to the web page for the
HERA-1 analysis since all theory input is identical to the one used then
(including hadronisation and QED corrections).
Cross sections for HERA-2(for test)
Single-differential
Cross-sections and ratios to NLO for Q2 and Mjj.
Cross-sections and ratios to NLO for ET,mean and log10ksi.
Cross-sections and ratios to NLO for xBj and etaprime.
Double-differential
Cross-sections for log10ksi in bins of Q2.
Ratios of data over NLO for log10ksi in bins of Q2.
Here is a table of cross sections for HERA-2 (PDF).
Cross sections for HERA-1/HERA-2 combined (stamped preliminary)
Single-differential
Cross-sections and ratios to NLO for Mjj.
Cross-sections and ratios to NLO log10ksi.
Cross-sections and ratios to NLO etaprime.
Cross-sections and ratios to NLO for Q2.
Cross-sections and ratios to NLO for ET,mean.
Cross-sections and ratios to NLO for xBj.
Double-differential
Cross-sections for log10ksi in bins of Q2.
Ratios of data over NLO for log10ksi in bins of Q2.
Here is a table of cross sections for HERA-1/HERA-2 combined (PDF).
More detailed checks
Trigger efficiency
The trigger efficiency of the combination DIS03 and DST bit12 was checked
using events preselected without a trigger requirement and checking the
fraction of selected events that were triggered. The trigger efficiency was
found to be flat at 100% over the whole Q2 range, as can be seen here. The red line is for data, the
dashed black line for the ARIADNE MC at detector level.
Electron calibration
A proper electron energy calibration was applied in order to compensate for
the MC mismodellings and miscalibrations. This was motivated by the observed
bad agreement between MC and data for the electron quantities pT, energy and
E-pz. at the same time, good agreement between the DA quantities in MC and
data was observed, so that it was assumed that the DA method would allow for a
correction of the energy scale in the MC. In a second step, the true energy
information in the MC was used to correct both MC and data to the true
level. As a result, the energy spectra agree pretty well between data and MC
after the correction. The correction was performed in bins of theta for 4
different phi regions. The following plots illustrate the procedure and the
result of the corrections applied.
- In these plots, the description of data by the MC can be seen for DA
quantities (left) and electron-method quantities (right): pT and energy, and
hadronic and EM E-pz.
- These plots show, before the correction:
the ratio of DA energy over EL energy for data
and MC as a function of theta in 4 phi bins,
the double-ratio data / MC of the quantity (DA energy)/(EL
energy) as a function of theta in 4 phi bins, and
the ratio of EL energy (MC) over true MC
energy as a function of theta in 4 phi bins. The bad description of data
by the MC is clearly visible, for example, in the double-ratio deviations from
1. One can also observe that the true energy deviates from the MC
reconstructed energy.
- These plots show, after applying to the MC the scale correction derived
from the double-ratio data/MC for (DA energy)/(EL energy) in bins of theta/phi:
the ratio of DA energy over EL energy for data
and MC as a function of theta in 4 phi bins, and
the double-ratio data / MC of the quantity (DA energy)/(EL
energy) as a function of theta in 4 phi bins. The description of data
by the MC is clearly improved a lot, as can be seem from the double-ratios
which now are around 1.
- These plots show, after bot the scale correction and the correction (of
data and MC) to true level using the ratio (EL MC)/(true MC):
the ratio of DA energy over EL energy for data
and MC as a function of theta in 4 phi bins,
the double-ratio data / MC of the quantity (DA energy)/(EL
energy) as a function of theta in 4 phi bins, and
the ratio of EL energy (MC) over true MC
energy as a function of theta in 4 phi bins. Now also the true and EL
reconstructed levels agree well, indicating the full success of the correction method.
In addition, Jürg provided a ROOT class for electron energy
correction at high-Q2 (Q2 above 125 GeV2). Please use with care and
provide feedback!
eta shape in LEPTO for HERA-II
We observe a funny behaviour i the eta prime distribution for LEPTO at HERA-II. in order to investigate this further, we prepared a few plots. more will come very soon.
- Plots showing,
on detector level (lin scale),
on hadron level (lin scale),
and
on hadron level (log scale),
the pseudorapidities in the Breit frame of the hardest and second hardest jet, and
the laboratory pseudorapidities of the forward and the backward jet.
Conclusion: Both on detector and on hadron level
LEPTO at HERA-2 differs from the other 3 models (note that also ARIADNE at
HERA-2 is slightly different from the HERA-1 models for this distribution, but
this is seen mainly in bins with low statistics). The LEPTO-2 discrepancy can
be seen clearest in the shifted peak positions in the hadron level eta(jet1) and
eta(jet2) distributions.
- Plots showing,
on lin scale and
on log scale,
the transverse jet energies for the first and second jet both on detector and
on hadron level.
Conclusion: On hadron level, the ET spectra compare
quite well between HERA-1 and HERA-2, individually for LEPTO and
ARIADNE. There is a difference between both LEPTO models and both ARIADNE
models which is simply due to the modelling of the hard spectrum. On detector
level, the LEPTO model at HERA-2 falls a bit short of the other 3 models
especially at low ET(jet1).
- Plots showing, on detector level and
on hadron level,
the six inclusive dijet distributions for which we measure cross sections. .
Conclusion: The simulation of kinematics (Q2, xBj) is
different for LEPTO and ARIADNE, to the same extent at HERA-1 and HERA-2. The
only eye-catching difference is (on both detector and hadron level) the shape
of LEPTO at HERA-2 which differs drastically from the other 3 models (which in
turn are well compatible within errors). The strange LEPTO-2 behaviour can be
traced back to the features observed for the individual eta distributions on
hadron level.
Overall conclusion: We note that
- the cross sections for HERA-1 and HERA-2 agree quite well.
- the discrepancies observed between LEPTO (HERA-2) and the other three
models are of the same size at detector and hadron level.
- the discrepancies are limited to the eta-prime distribution.
We therefore conclude that LEPTO (HERA-2), compared to the other 3 models,
has been generated differently, affecting the eta distributions, but not the
other measured distributions. In order to not be affected in our measurement
we treat the eta-prime distribution at HERA-2 different from the other cross
sections: In general, we use ARIADNE to estimate the acceptance correction and
use LEPTO (HERA-2) as a systematic cross-check. However, for eta prime we
still use ARIADNE (HERA-2) for the correction, but we apply the systematic
error derived at HERA-1 since we consider the error of the detector effects to
be of about the same size at HERA-1 and HERA-2.
Hadronic energy scale calibration
The hadronic energy scale was calibrated in the MC files
using specially selected events with
sigle high-energy jets and employing the energy balance between jet and
scattered electron. A few results of the procedure can be seen below. The
correction reduces the scale uncertainty to about 1.5% (for jets between 10
and 12 GeV in the lab frame) and 1% (for jets of higher energy). For softer
jets an uncertainty of 3% is assumed.
- LEPTO hadronic energy scale compared to data for ET,lab
distributions in various eta bins before and after the scale correction.
- Ratio of LEPTO hadronic energy scale to data for ET,lab
distributions in various eta bins before and after the scale correction.
- ARIADNE hadronic energy scale compared to data for ET,lab
distributions in various eta bins before and after the scale correction.
- Ratio of ARIADNE hadronic energy scale to data for ET,lab
distributions in various eta bins before and after the scale correction.
20 March 2007,
TSS