progress with the development of energy flow algorithms at argonne josé repond for steve kuhlmann...
TRANSCRIPT
Progress with the Development of
Energy Flow Algorithms at Argonne
José Repond
for Steve Kuhlmann and Steve Magill
Argonne National Laboratory
Linear Collider Workshop, Amsterdam, April 1 – 4, 2003
Simulation Software
Package JAS and GIZMO
Conversion to GEANT4 soon
Detector SD or ‘Small’ detector
Si-W ECAL 5x5 mm2
Scintillator-Fe HCAL 10x10 mm2
Track pattern recognition included
Overview of Algorithm
1st step - Track extrapolation thru Cal
– substitute for Cal cells in road (core + tuned outlyers)
- analog* or digital techniques in HCAL – S. Magill
2nd step - Photon finder (use analytic long./trans. energy
profiles, ECAL shower max, etc.) – S. Kuhlmann
3rd step - Jet Algorithm on tracks and photons - Done
4th step – include remaining Cal cells (neutral hadron
energy) in jet (cone?)
Optimize Cal segmentation for separation of charged/neutral clusters
* V. Morgunov, CALOR2002
Define D-weight: - for every cell - area of ~ 40 cells - D-weight = #/40 E-weight: - simply take energy of cell
cell density weight = 3/40
area ~ 40 cellsred – E fraction for density > 1/#blue – E fraction outside area
# cells in window
digital cal
analog cal
Track Extrapolation/Shower Link Algorithm
D-Weights: in ECAL at Interaction Layer 20
20 layers after π interacts
Response to single π with 10 GeV…
#eve
nts·
wei
ght
If D-weight:
≡ 1 Breit-Wigner
≡ n/40 Gaussian
θ φ
D-Weights: in HCAL at Interaction Layer 10
If D-weight:
≡ 1 Breit-Wigner
≡ n/40 Gaussian
θ φ
E-CAL: Energy in cell vs cell density
MIP signal at 8 MeV
cell density n/40
cell
energ
y (
GeV
)
MIP signal at 36 MeV
what’s this?
cell
energ
y (
GeV
)
cell density n/40
H-CAL: Energy in cell vs cell density
Soft stuff
in
Scintillator???
HCAL: analog E fraction only sum up if Ecell ≥ 1 MIP
ΣE/10 GeVSome losses…
Effect on energy resolution to be investigated…
HCAL: Cell Density (in 40 cell window)
31% single cell windows
mean ~ 4 cells
Seed Cell Distribution defined as cell w/ cell density > 1/40
In most cases
number of seed
cells = 0
Number of seed cells
D-weights: comparison with event display
Blue – allRed – density > 1/40Green – density > 3/40
Track Extrapolation/Shower Link Algorithm
1. Pick up all seed cells close to extrapolated track
- Can tune for optimal seed cell definition- For cone size < 0.1 (~6o), get 85% of energy
2. Add cells in a cone around each seed cell through n layers
3. Linked seed cells in subsequent cones form the reconstructed shower
4. Discard all cells linked to the track
Hadronic Z decays at √s = 91 GeV…
To develop Photon finder
Cuts to select photon clusters and reject anything else
I. EM clusters rejected within 0.03 of extrapolated track within 0.01 if track MIP in all 30 layers
II. EM clusters required to have shower maximum energy > 30 MeV (SME is sum of layers 8,9 and 10) still needs fine tuning
III. Require Ecluster/ptrack > 0.1, if EM cluster within 0.1 of track
Definition of EM cluster
Cluster of calorimeter energies within a cone of 0.04
No requirement on EECAL/EHCAL or EHCAL
Small loss of EM clusters
I. Cut on distance to track
Probability of overlapof γ and track
0.1% within ΔR < 0.02 3.3% within ΔR < 0.1 11% within ΔR < 0.2
Reject EM clusters if ΔR < 0.03 < 0.01 if MIP in all 30 layers of ECAL
2 GeV e-
2 GeV -
MeV
II. Cut on energy at shower maximum
MeV
Require SME > 30 MeV
ΔR from EM Cluster to Track
EM
Clu
ster
Ene
rgy/
Tra
ck E
III. Energy – momentum ratio
Clusters left after previous cuts
10 GeV π-
Require Ecluster/ptrack > 0.1 for ΔR <0.1
Hadronic Z Decays at s = 91 GeV
Total Hadron Level Photon Energy (GeV)
Tot
al P
hoto
n C
andi
date
Ene
rgy
After all cuts…
μEM = 0.25 GeV
σEM = 2.8 GeV
Hadronic Z Decays at s = 91 GeV
Total Photon Energy - Total Monte Carlo Photons (GeV)
Perfect EFA
Goal σEM = 1.4 GeV
Compare to h0 σh = 2.9 GeV
Simulation of different active media
Gas Electron Multipliers
J Li, A White, J Yu: University of Texas in Arlington
Simulation of both detailed and simple GEM geometries
Resistive Plate Chambers
L Xia: Argonne National Laboratory
Single π Resolution
Summary
1. Continuing work on implementation of algorithm
2. Tune to single particles first, then to particles in jets
3. Implement photon finder
4. Deal with neutral hadron energy
5. Compare performance to analog version (SNARK)
6. Use EFA to optimize transverse cell size of DHCAL
7. Compare performance with different active media:
Scintillator, GEMs and RPCs