Level
-1 CSC Trigger Simulation
Level-1
Darin Acosta
University of Florida
May 2001
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
1
Simulation Strategy
Studies have been carried out using the objectoriented software framework of CMS
• Geant3 for hit generation
• “ORCA” for detector and trigger simulation
Entire L1 CSC Trigger scheme coded in C++
• Perfect agreement achieved between simulation and
hardware for millions of events (MPC, SR, SP)
Trigger rate, efficiency, and P T resolution studied
for L = 1034 cm-2s -1
• Pile-up includes 17.3 minimum bias collisions per
beam crossing and neutron hits from much earlier
crossings
• Aim for single muon rate <10 kHz with >90% efficiency
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
2
Signal
MB
Geant3
HEPEVT
ntuples
CMSIM
MC Prod.
Pythia 6
ORCA Production Scheme
Zebra files
with HITS
Objectivity
Database
ORCA
ooHit
Formatter
Mirrored Db’s
Catalog import
Objectivity
Database
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
HLT Grp
Databases
HLT Algorithms
New
Reconstructed
Objects
Objectivity
Objectivity
ytivitcejbO
Database
Database
esabataD
3
(CERN, US, Italy,…)
ORCA
Objectivity
Database
ORCA
Digitization
(merge signal
and MB)
ORCA Prod.
Catalog import
CSC Muon Trigger Scheme
On-Chamber
Trigger Primitives
Strip LCT +
Motherboard
card
LCT
Strip FE cards
3-D Track-Finding
and Measurement
Port Card
Sector
Receiver
Sector
Processor
OPTICAL
FE
SR
PC
SP
LCT
3µ / port card
TMB
FE
2µ / chamber
Wire LCT card
Wire FE cards
RIM
RPC Interface
Module
Combination of all
3 Muon Systems
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
3µ / sector
In
counting
house
RPC
4µ
CSC Muon Sorter
DT
4µ
4µ
Global µ Trigger
Global L1
4µ
4
Geometric Coverage
ME4 descoped
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
5
CSC PTT Resolution
• Track segment in ME1 required to have a reasonable
measurement (resolution worse than 70% otherwise)
• Track segments in three CSC stations required for
optimum PT resolution and rate reduction
(implemented in Sector Processor prototype)
PT < 35 GeV
1.2 < |η| < 2.0
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
6
CSC Trigger Efficiency vs. PTT
Trigger threshold
defined at 90%
efficiency
Sharper turn-on for
better PT resolution
Require ME1 for good PT
resolution
1.2 < |η| < 2.4
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
7
CSC Trigger Efficiency vs. η
Loose: 2 or more stations
including ME1 in endcap,
but any two in DT/CSC
overlap region
~90% efficiency
Tight: 3 or more stations
including ME1 in endcap
and MB1 in DT/CSC
overlap
~70% efficiency
but better PT resolution
Significant acceptance
loss when 3 stations are
required and no ME4
ME4/2
ME4/1
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
8
CSC Single Muon Rate
Weighted minimum
bias sample used to
estimate rate for
L = 1034
Require “tight” track
conditions ( ε ≈ 70%)
to get acceptable rate
from standalone CSC
trigger
5 kHz rate for threshold
set at 25 GeV
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
9
CSC Rate with Neutrons
Previous study did not
include effect of punchthrough or neutron hits
Had used weighted events
with muons
But no significant change
in rate or efficiency (< 10%)
is seen when effects are
included
So real muons dominate
the trigger rate
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
10
CSC Rate without ME4
Trigger rate for “loose”
condition increases by
~50% if ME4 is removed
because of fewer
3-station tracks
⇒ worse PT resolution
Trigger rate for “tight”
condition stays about
the same because
3-station tracks already
required
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
11
Global Muon Trigger Efficiency
Recover some inefficiency by combining DT and CSC (loose) with
the RPC muon system in an optimized way (see next slide)
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
12
Global Muon Trigger Rate
Algorithm Efficiency Ghosts Rate
PT>25
OR
97.6%
1.6%
23 kHz
AND
89.7%
0.07%
7.2 kHz
optimized 96.1%
0.18%
8.1 kHz
Uses “loose” criteria for CSC trigger
The GMT optimized algorithm
gives an efficiency close to the
“OR” of RPC and CSC, but a rate
close to the “AND” of the two.
Optimized ⇒ RPC coincidence
required in certain regions, choose
lowest PT reported by RPC and
DT/CSC when matched
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
13
Conclusions
The previous simulations are the most detailed done to
date of the CMS muon trigger
• FE amplifier response/noise is in, neutrons are in, etc.
• Trigger logic mostly validated against current hardware
designs (some updates needed)
But we need detailed validation against real data
• Testbeam and cosmic ray data
This requires physicist support!
• Postdocs, students
It’s important to design a highly efficient L1 CSC Trigger
• L2 and L3 algorithms use L1 candidates as seeds.
If L1 misses the muon, physics may be lost…
• Must have ME1, and re-scoping ME4 helps at high luminosity
• RPC noise/trigger rate may be unacceptably large to help GMT
US CMS DOE/NSF Review: May 8-10, 2001. D. Acosta, UF
14