Fun with J/? ? e e-

1
Fun with J/? ? e e-

• Jan Stark
• LPSC Grenoble

All D0 meeting, February 20th, 2004
2
Introduction
This is a short talk the main motivation is to
flash some cool event displays. What is so cool
about these displays of candidate J/? ? e e-
events ? They are a great illustration of
the performance of the Run II D0 detector, both
at the triggerDAQ level CTT CFT
Calorimeter fancy algorithms at L3 to be added
in a matter of weeks (trigger list
v13) CPS L2 (CTT calorimeter matching)
and at the offline level as above
SMT This performance is the result of year-long
hard work by many collaborators. Nothing is
perfect, and there are also some not yet so
beautiful things to be reported more on this
later. With this channel, one cannot test the
muon system, but I can confirm that they see the
J/? as well.
3
Flashback
Motivation Determine the offset in the electron
energy response (calorimeter). This offset
is an important systematic uncertainty in the
measurement of the MW/MZ mass ratio.
4
Flashback
Dataset A set of special runs at the end of
Run Ia and at the beginning of Run
Ib. Trigger L1 two EM towers above 2.5
GeV with less than 1 GeV in the
hadronic part of the tower (in the CC,
I guess). L2 two L2EM objects above 3 GeV,
isolation lt 0.40 in a 0.4 x 0.4 cone (?/?).

• 21 signal events (significance 6?)
• Mass 3.102 ? 0.035 (stat) GeV
• Resolution 115 ? 25 MeV

5
Run II trigger
A new low-pT di-EM trigger was introduced in
trigger list v12. Summary of the selection
criteria used by this trigger L1 CEM(2,3.)
CEM(1,6.) TTK(2,3.) TTK(1,5.) L2 ?
restriction to central region L3 two road
electrons ( gt 3 GeV and gt 5 GeV, one tight one
loose) (for details, see J. Stark, Low-pT
di-EM triggers status and plans, Trigger
board meeting, October 17th, 2003 and references
therein). This trigger typically runs
unprescaled up to 35E30. Rates ? 190/150/1.3
(0.3 unique) Hz at L1/L2/L3 at 30E30. Comments
on improvements for trigger list v13 last slide.
6
J/? yields
38.4 pb-1
Reconstruction using calorimeter clusters (two
different H-matrix cuts).
Reconstruction using road method (track-based).
727 J/?
2151 J/?
348 J/?
Resolution not quite as good as in Run I
7
An example event
Two tracks, triggered on by CTT. CTT also knows
that these tracks are isolated.
8
Another example
9
One more
10
Summary and Outlook

• J/? ? e e- is a tough test for our sophisticated
  detector/trigger/DAQ system.
• Obtained good results have nice publicity
  material to illustrate that it is working.
• The J/? ? e e- sample is used for a variety of
  detector studies
• central preshower response
• - tracker/CPS/CAL matching
• - efficiency
• - CPS weight in energy determination
• calorimeter resolution and energy calibration
  (room for improvement)
• validate H-matrix and electron likelihood,
• improve quality of simulation for
  discriminating variables
• CTT/CAL matching at L2
• efficiency of the road method which we will
  ultimately use for b-tagging
• (single tags vs. double tags)
• Improvements for trigger list v13 (being
  finalised right now)
• - include CPS,

11
Backup slides
Road method Trigger bias
(for details, see J. Stark, Low-pT di-EM
triggers status and plans, Trigger
board meeting, October 17th, 2003 and references
therein).
12
Different methods to reconstruct J/? ? e e-

• Before we consider the selection at trigger
  level, lets discuss how we can
• reconstruct J/? ? e e- offline.
• One way to reconstruct J/? ? e e- is to proceed
  just like for Z ? e e-
• Look for EM clusters in calorimeter to
  reconstruct electrons.
• Possibly require EM clusters to be matched to
  tracks.
• Combine two clusters to make J/?. Calculate J/?
  mass from cluster energies
• or track momenta.
• This procedure works well for Z ? e e-. As we
  will see, it also works well for J/? ? e e- from
• direct J/? production. J/? from direct production
  tend to be isolated.
• But electrons from b ? J/? ? e e- tend to be
  non-isolated (hidden in jets)
• ? do not find separate clusters.
• An alternative method to reconstruct non-isolated
  electrons is available the road method
• (F. Beaudette, J.-F. Grivaz, D0 notes 3976
  4032).
• Start from reconstructed tracks.
• Look at energy deposited in narrow road around
  the track extrapolated through the

13
Results on Monte Carlo
Use road method and scone clusters to reconstruct
J/? ? e e- in two different p14.3 Monte Carlo
samples J/? from b decays and direct J/?
production. Require one electron track with pT gt
3 GeV, one with pT gt 5 GeV (inspired by L3
criteria of new trigger).
J/? from b decays
direct J/? production
Road method (tight cuts)
J/? mass
radiative tail
GeV
Scone clusters EM fraction gt 0.9 isolation lt
0.15 HMx8 lt 50 track matched to road
electrons (one loose, one tight)
low efficiency, non-gaussian shape because
non-isolated
14
Effect of trigger selection
Take the two Monte Carlo J/? samples (road
method) from the previous slide, and turn on the
trigger requirements (TrigSim). Start with L3,
then add TTK, then CEM. The plots below show the
trigger efficiency w.r.t. offline reconstruction
as a function of pT (J/?).
J/? from b decays
direct J/? production
L3 requirement
L3 TTK
L3 TTK CEM(2,3.)
L3 TTK CEM(2,3.) CEM(1,6.)
pT (J/?) GeV
pT (J/?) GeV
Efficiency of CEM terms tower than for J/? from b
decays. The latter events get help from the rest
of the b jet.
15
Effect of trigger selection
Take the MC sample of J/? from direct production
reconstructed using scones, and turn on trigger
requirements.
offline cuts only
offline L3 TTK
mean 3.295 width 0.262
mean 3.297 width 0.248
offline L3 TTK CEM(2,3.)
offline L3 TTK CEM(2,3.) CEM(1,6.)
mean 3.337 width 0.232
mean 3.366 width 0.229
The CEM requirements increase the mean and reduce
the width of the mass distribution. They eat
into the distribution from the left.
16
Track-mass plot with numbers
38.4 pb-1
Signal model used in the fit (Crystal Ball
function)