Transcript Document
Top physics during ATLAS
commissioning
Ivo van Vulpen
Wouter Verkerke
Structure of the talk:
Reminder you of the goals of the study
and main results presented in Rome
Overview new results since Rome
Goals for top physics during comissioning:
1) Can we see the top peak in the
LHC commissioning run ?
With 300 pb-1
Without b-tagging
W boson
CANDIDATE
TOP quark
CANDIDATE
2) Can we help commission the ATLAS
detector using these events ?
Calibrate light jet energy scale
Calibrate missing ET
Obtain enriched b-jet sample
Cross section
Simple (standard) top quark selection:
Missing
ET > 20 GeV
1 lepton
PT > 20 GeV
Selection efficiency = ~5 %
4 jets(R=0.4) PT > 40 GeV
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 3
Main results shown in Rome:
3-jet mass distributions m(jjj),
with and without cut on Mw
Hadronic 3-jet mass
Hadronic 3-jet mass
m(Whad)
L=300 pb-1
(~1 week of running)
Cut on Mw
Mjjj (GeV)
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Mjjj (GeV)
Slide 4
What’s new since Rome
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 5
What’s new since Rome: focus on concerns
1) Trigger
Effect of electron trigger: 2e15i+e25i+e60
2) New background estimate from W+jets
Addressing concern about phase space coverage
A7 sample (W+jets) used for Rome analysis
New estimate using Alpgen+MLM matching
3) 100 pb-1
More realistic estimate for integrated luminosity
during LHC commissioning run
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 6
Trigger Performance
“How much ‘good’ electron events do we lose by including the trigger ?”
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 7
Impact various selection criteria on ttbar selection efficiency
Fraction of events
passing cuts
• Jets:
4 reconstructed jets with Pt > 40 GeV
Losses mainly due to hard analysis cut on jet kinematics
13.4%
• Electrons
At least 1 reconstructed electron wth Pt> 20 GeV
Losses mainly due to reconstruction
62.0%
• Missing Et > 20 GeV
91.8 %
Electron trigger important for event selection
and cross section measurement
Need to understand differences between ttbar
and clean Ze+e- or Weν events
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 8
Scope of trigger plots
Data
Trigger
Reconstruction
Analysis
“How much ‘good’ electron events do we lose by including the trigger ?”
Investigate trigger performance:
step 1: Require reconstructed good e- (with/without Pt cut)
step 2: Require e- to point back to MC truth e- from W decay
step 3: Look at trigger decision
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 9
Trigger efficiency versus Pt (no pt-cut)
Note:
Events with a reconstructed electron (no Pt-cut) that
matches the electron from the W decay (Monte-Carlo truth)
Same as white, but have ‘yes’ trigger decision
83.9 %
e- (Rec+match)
e- (Rec+match
+ Trigger)
Remaining questions:
What object triggered the
events with low-Pt e-‘s ?
Why do we lose electrons
Pt = 100 GeV in barrel ?
MC truth electron Pt (GeV)
Ivo van Vulpen (NIKHEF)
MC truth electron Pt (GeV)
ATLAS physics week (October 2005)
Slide 10
Trigger efficiency versus Eta (Pt > 20 GeV)
Note:
Events with a reconstructed electron (Pt>20 GeV) that
matches the electron from the W decay (Monte-Carlo truth)
Same as white, but have ‘yes’ trigger decision
e- (Rec+match)
e- (Rec+match
+ Trigger)
MC truth electron Eta
Ivo van Vulpen (NIKHEF)
MC truth electron Eta
ATLAS physics week (October 2005)
Slide 11
Background estimate from W+jets
“Do you cover the full phase space contributing to 4 reconstructed jets?”
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 12
What did we have in Rome: the A7 sample
• What is the A7 sample
A7 = ‘Alpgen+ 4 jets’:
= W+4-partons L.O. Matrix Element
+ (Herwig) parton shower
Wlν
W
ln
• Possible concern about the A7 sample
Do we cover the full phase space that contributes
to 4 reconstructed jets. Probably not.
What about W+1/2/3-partons + hard gluon(s) from PS ?
• ‘Good’ news: A7 cross section wrong on wiki:
Cross section presented on wiki was wrong by factor ~2
Background goes down!
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 13
Towards modeling the full phase space
• ‘Traditional’ approach : W+0jets Matrix Elements(ME) + Parton
Shower (PS):
– Would covers full phase space, but …
– Extremely inefficient for high-Pt jet sample
– Parton shower does not correctly describe hard gluon emission
•
remember: we require 4 jets with Pt > 40 GeV
• Idea for improvement:
– Use parton shower for low-Pt radiation
Parton shower
Matrix Element
– Use matrix element for high-Pt radiation
0
PT-cut
40
100 GeV
• Practical translation:
– Generate separate samples of W + 0,1,2,3,4,5 ME partons
– add arton shower to each sample
– Cannot simply add samples because of double counting from hard parton showers
– Solution: Alpgen + MLM matching (M. Mangano)
In a nutshell: kill events with too high PT-gluons in PS
– After matching can add W + n ME partons samples
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
40
Slide 14
Does MLM matching work ?
• Look at PT distribution of W-boson at Tevatron
(Plot taken from presentation by M. Mangano)
W
W+2/3/4jets
W+0jets
W+1jets
PT W-boson = net PT radiation
– Region of high W-boson transverse momentum described by matrix
element computation
– Sum of MLM-matched W + n ME parton samples describes CDF data well
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 15
Applying MLM to estimate W + 4 reco jet background
•
Generate samples of W + n ME partons + PS sample (n=0,1,2,3,4,5)
•
Look at contribution of each sample to W + 4 reco jets final state
• Constribution of ME parton samples
in selected events (4 reconstr. jets)
#Events
#Reco jets
• # Alpgen ME partons
versus # reconstructed jets
Sample (# of ME partons)
Ivo van Vulpen (NIKHEF)
Sample (# of ME partons)
ATLAS physics week (October 2005)
Slide 16
Applying MLM to estimate W + 4 reco jet background
• Generate samples of W + n ME partons + PS sample
Background dominated by
(n=0,1,2,3,4,5)
W + 4 ME parton sample
• Look at contribution of each sample to W + 4 reco jets final state
• Constribution of parton samples
in ttbar sample (4 reconstr. jets)
#Events
#Reco jets
But other samples also contribute due to
small differences in jet definition in MLM
•matching
# Alpgen ME
andpartons
reconstruction, effects of
versus # detector
reconstructed
jets
simulation
etc…
Does not affect validity of procedure but strong mismatch will
increase number of significantly contributing samples
Sample (# of ME partons)
Ivo van Vulpen (NIKHEF)
Sample (# of ME partons)
ATLAS physics week (October 2005)
Slide 17
Result: W + 4 reco jet background from MLM matching
• Bottom line for W + 4 jets background in 3-jet invariant mass m(jjj)
Add all W + n ME partons samples and normalize sum to 127 pb-1
(luminosity of A7 sample)
A7 estimate (127 pb-1)
MLM estimate (127 pb-1)
W
W
W
W
W
W
+
+
+
+
+
≥
0
1
2
3
4
5
ME
ME
ME
ME
ME
ME
A7 & MLM (unit norm)
part.
part.
part.
part.
part.
part.
Amount of background increases by ~10%
Shape consistent
• Including full phase space adds ~10% background w.r.t A7 samples
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 18
More plots on W+ n ME MLM shape vs A7
MLM: PT of W-boson
pT, h distributions of all jets and
the electron consistent between A7 and MLM
W
W
W
W
W
W
+
+
+
+
+
≥
0
1
2
3
4
5
ME
ME
ME
ME
ME
ME
part.
part.
part.
part.
part.
part.
PT of leading jet
A7 estimate (127 pb-1)
Ivo van Vulpen (NIKHEF)
MLM estimate (127 pb-1)
ATLAS physics week (October 2005)
A7 & MLM (unit norm)
Slide 19
Summary on W+jets background
• Evaluated background on full phase space by including
W + 0,1,2,3,4,5 ME partons + PS using MLM technique
- Background level increases by ~10% w.r.t. A7 sample
- M(jjj), pT(jet), η(jet), pT(e-), η(e-) shapes all consistent
between A7 and MLM sample
• To do: study effect of varying MLM matching parameters
– Can e.g. vary PT threshold between PS and ME
– Check that result is not strongly dependent on choice of matching
parameters
• Include Wmν decays in study (need to be generated)
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 20
Results for 100 pb-1
“What are the results of the study when using a more conservative
estimate for the luminosity collected during the commissioning run ?“
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 21
Results for 100 pb-1 (no cut on reconstructed W mass)
Note 1: Background ~factor 2 lower due to initial mistake in A7 lumi
Hadronic 3-jet mass
100 pb-1
L =100 pb-1
electron-only
Mjjj mass (GeV)
Events / 4.15 GeV
Events / 4.15 GeV
Note 2: Error bars now reflect statistical error of 100 pb-1 instead of
statistical error of MC sample as was done for Rome plots.
Hadronic 3-jet mass
200 pb-1
L=200 pb-1
electron+muon
Mjjj mass
(GeV)
estimate
for L=100
pb-1
Mjjj (GeV)
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Mjjj (GeV)
Slide 22
Results for 100 pb-1 (with cut on reconstructed W mass)
Hadronic 3-jet mass
L =100 pb-1
electron-only
Events / 4.15 GeV
Events / 4.15 GeV
Distribution of 3-jet invariant mass after a cut on the mass
of the reconstructed W-boson: 70 < Mjj < 90 GeV
Mjjj (GeV)
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Hadronic 3-jet mass
L=200 pb-1
electron+muon
estimate for L=100 pb-1
Mjjj (GeV)
Slide 23
Relax cut on minimum
PT requirement for jets
“Top peak close to rising edge of background distribution
when using a minimum jet PT-cut at Pt = 40 GeV. “
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 24
Relaxed cut on minimum PT requirement for jets
• Top peak on rising edge background distribution:
Try relaxing cut on minimum jet-PT
Hadronic 3-jet mass
L =100 pb-1
electron-only
Minimum Jet PT = 30 GeV
Events / 4.15 GeV
Events / 4.15 GeV
Minimum Jet PT = 40 GeV
Hadronic 3-jet mass
L=100 pb-1
electron-only
Mjjj (GeV)
Mjjj (GeV)
In Note: investigate stability and effects from changed selection criteria
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 25
Summary
• Focused on concerns after Rome
– New estimate for W+jets background
• Lower estimate due to mistake in A7 lumi
• New procedure Alpgen+MLM matching 10% higher than corrected A7
result
– First results on impact electron trigger
– Preliminary results now quoted for 100 pb-1
• Plan
– Finalize Alpgen+MLM matching study
– Evaluate some outstanding issues (b-tag, calibrations, etc.)
– Write note
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 26
Backup slides
Ivo van Vulpen (NIKHEF)
ATLAS physics week (October 2005)
Slide 27
Electron
(62.0%)
Et-miss
(91.8%)
Jets
(13.4%)
Selection criterium
ttbar events passing all cuts
Electron trigger important for event selection
and cross section measurement
Need to understand differences between ttbar
and clean Ze+e- or Weν events
Ivo van Vulpen (NIKHEF)
Jet Pt-cut
Main loss due
to kinem. cuts
(also # jets)
Pt of 4th jet (GeV)
Number of events
Number of events
100 %
Number of jets
Impact various selection criteria on ttbar selection efficiency
ATLAS physics week (October 2005)
Electron Pt-cut
Main loss due
to reconstruc.
Pt electron (GeV)
Slide 28