Transcript Energy Flow Studies Steve Kuhlmann Argonne National Laboratory

Energy Flow Studies
Steve Kuhlmann
Argonne National Laboratory
for Steve Magill, U.S. LC Calorimeter Group
Introduction/Outline
Detector is the “Small” Detector
(Si-W EM Cal, 5 mm X 5mm, R=127 cm, 17%/E)
(Fe-Scint HAD Cal, 1 cm X 1cm, R=144cm, 60%/E)
Software is JAS2 and GIZMO simulation
Conversion to Geant4 “soon”
Real Track Pattern Recognition Included
Will Discuss:
Brief Photon Review and Plans
Initial work on the Real Challenge: Neutrons/KLongs
Resolution components of Hadronic Z Decays at s = 91 GeV
Assuming Perfect Identification in
this Detector Configuration
•Neutrons+KLong 2.9 GeV
•Photons 1.4 GeV
•Tracks 0.25 GeV
Put together in Tesla TDR
in Energy Flow algorithm
Hadronic Z Decay
Simple 3 cut analysis
1. Reject EM Clusters if within Delta-R<0.03
from Track (0.2% loss of real photons)
2. Shower Max Energy > 30 MeV (MIP=8 MeV)
3. Reject EM Cluster if Delta-R< 0.1 AND E/P<0.1
Java code is available at:
www.hep.anl.gov/stk/lc/uta/
Will be put in CVS Server “soon”
Total Photon Candidate Energy
Hadronic Z Decays at s = 91 GeV
Total Hadron Level Photon Energy (GeV)
Hadronic Z Decays at s = 91 GeV
Mean=0.25 GeV,
Width=2.8 GeV,
Perfect EFLOW Goal
is 1.4 GeV.
Total Photon Energy - Total Monte Carlo Photons (GeV)
Energy Fragments from a Single 10 GeV -
Current Photon Work
1. Reject EM Clusters if within Delta-R<0.03
from Track (0.2% loss of real photons)
2. Shower Max Energy > 30 MeV (MIPS=8 MeV)
3. Reject EM Cluster if Delta-R< 0.1 AND E/P<0.1
Replace these two cuts with SLAC
NNet-based ClusterID package.
(Worked on technical difficulties
with Bower after UTA, not solved)
Neutron/K0L Content of Hadronic Z Decays at s = 91 GeV
Neutron/K0L Energies in Hadronic Z Decays at s = 91 GeV
Neutrons/K0L,
Mean E=4.35
Neutrons/K0L,
Mean E=4.4 GeV
Study of >2 GeV Neutron/K0L
overlapping >2 GeV Tracks
Study of >2 GeV Neutron/K0L
overlapping >2 GeV Tracks
Study of >2 GeV Neutron/K0L
overlapping >2 GeV Tracks
Separation between Track
and Closest N/K0L
Separation between
N/K0L and Closest Track
Overflow bin
Overflow bin
Angular Separation (radians)
10% overlap within Sep<0.2
23% overlap within Sep<0.4
Angular Separation (radians)
48% overlap within Sep<0.2
77% overlap within Sep<0.4
Overlapping Showers
Other Tracks
from
Separation between random >2 GeV Track
and Closest >2 GeV Track
Angular Separation (radians)
16% overlap within Sep<0.1
41% overlap within Sep<0.2
59% overlap within Sep<0.3
72% overlap within Sep<0.4
Single 10 GeV Charged Pions:
Basic Shower Widths
Angular Separation (radians)
Single 10 GeV Charged Pions:
Means and Widths
Mean
Width
Width
All Hits
8.3 GeV
19%
60%/sqr(E)
Cone<0.4
8.1 GeV
21%
67%/sqr(E)
Cone<0.3
7.9 GeV
22%
68%/sqr(E)
Cone<0.2
7.5 GeV
22%
70%/sqr(E)
Cone<0.1
6.4 GeV
25%
80%/sqr(E)
Cone<0.075
5.8 GeV
28%
88%/sqr(E)
Single 10 GeV Charged Pions:
All Hits
Cone<0.2
EM+HAD Energy (GeV)
EM+HAD Energy (GeV)
These plots are with analog hadron cal, very similar with digital
Select Charged Pions isolated from
other tracks in Z Decays,
look
for Neutron Overlap
No overlap from
particle list
Overlapping
Neutron/K0L
Cal Energy/Track P
Two approaches being investigated:
1) Put calorimeter and track
properties into neural net.
List of calorimeter
variables put into
ClusterID Net:
Tesla TDR approach
2) Careful removal of
track depositions from
Calorimeter. Used in
European package called
“Snark”. Results similar
to Tesla TDR, but larger
resolution tails.
Reminder, the Questions we
eventually need to Answer
Detector Size and Hadron Calorimeter Resolution?
Digital or Analog Hadron Calorimeter?
Optimized segmentation for physics/costs?
Backup Slides
Question from Jeju
and Calor2000:
Will Hadronization or
Jet Clustering Ruin
Resolutions?
No, at least if
backgrounds are small
Particle Energies in Hadronic Z Decays at s = 91 GeV
Charged,
Mean E=2.85
Photons,
Mean E=1.0
Neutrons/K0L,
Mean E=4.35
Tracking cannot be assumed to be perfect,
forward tracking and “curlers” are issues
Tesla TDR,
is fine if
achieved
Effect of ignoring charged
particles below certain thresholds
Track Reconstruction Efficient Down
to Pt=0.5 GeV in Barrel Region
EM Cluster Energy (GeV)
Single 10 GeV -
EM Clustering -- Cone 0.04
Delta-R from EM
Cluster to Track
Reduce charged particle fragments with
3-layer shower max energy > 30 MeV
ddd
2 GeV Electron
Also reduces
neutron/K0L
clusters
ddd
2 GeV -
MeV
EM Cluster Energy (GeV)
Single 10 GeV -
Now With Shower Max Cut,
will be improved with more
detailed information on
lateral/longitudinal profile
Delta-R from EM
Cluster to Track
Effect of possible Photon threshold on
Hadronic Z Decays at s = 91 GeV
Photons are soft,
Mean E=1.0
Sum of all Hadron Level energy
except photons < 0.2 GeV.
Won’t apply such a cut (yet).
Hadronic Z Decays at s = 91 GeV
Simple photon finder:
Remove EM Clusters within
0.03 of Track, unless track
was MIP in all 30 layers.
Then remove if within 0.01.
Hadronic Z Decays at s = 91 GeV
Probability of Overlapping
Photon Close to a Track,
0.1% within DR<0.02,
3.3% within DR<0.1,
11% within DR<0.2
Determining Charged Particle Depositions
Energy deposited in last EM layer
(within 0.60 of track)
• Easy to recognize MIP
• Easy to determine 1st
layer of pion shower
Interactions
Zeros
Overflows
Tail
Single 2 GeV -
Single 2 GeV Muon
Determining Charged Particle Depositions
Single 2 GeV -
Energy weighted
Effect of Neutrinos in Hadronic Z Decays
EM Cluster Energy/Track E
One more cut motivated by
Single 10 GeV -, now
either an Energy Ratio
Delta-R from EM
Cluster to Track