lhcc review, cern, 19/10/99paul bright-thomas, for alan watson 1 lvl1 calorimeter algorithm updates...
TRANSCRIPT
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 1
LVL1 Calorimeter Algorithm Updates
Changes since the TDR:Greater “integration” of e/ and /h triggers:
Numbers of em and tau thresholds now adjustableCommon RoI algorithm for e/ and /h algorithmsSplit e/ hadronic isolation into two regions
“Evolution” rather than “Revolution”:Main effects are to:
Increase flexibilitySimplify implementation
Changes approved by the ATLAS T/DAQ Steering Group in July ‘99
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 2
TDR e/ Algorithm
Object accepted if: 1 trigger cluster cluster
thr
em isolation ET em isol thr
had isolation ET had isol thr
RoI cluster ET = local maximum
Tau algorithm based on same 44
tower window (next slide)
Em Calorimeter Had Calorimeter
had isolationem isolation
RoI clustertrigger cluster
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 3
Current e/ Algorithm
Object accepted if: 1 trigger cluster cluster thr
em “ring” ET em isol thr
had “ring” ET outer had isol thr
had “core” ET inner had isol thr
RoI cluster ET = local maximum
Tau algorithm same except: trigger clusters = 2 em + 22
had towers
only the 2 “ring” sums are used for isolation
(Tau algorithm unchanged from TDR)
Em Calorimeter Had Calorimeter
had isolationem isolation
RoI clustertrigger cluster
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 4
Adjustable Numbers of e/ and /h Triggers
TDR design had fixed numbers of em and tau selections
Proposed Nem = N = 8
Optimum allocations hard to predictWill be luminosity-dependentMay depend on what we find
Better if we could make these numbers adjustableHard to vary total ( Ntot = Nem N)Can adjust Nem & N within this totalRequires small increase in functionality within CP chip
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 5
Adjusting Numbers of e/ and /h Triggers
Em & Tau algorithms very similar See Table
To switch between the two: Use multiplexors to select either
em or tau trigger clusters
Do not set “hadronic core” isolation for tau triggers
Don’t need total flexibility:We propose:
8 sets of em thresholds
8 sets adjustable em/tau
(Even this not fixed if FPGAs used)
Element e/ Trigger /h Trigger
TriggerCluster
21 / 12em towers
21 / 12 em 22 hadronic
EmI solation
"Ring" of 12towers
"Ring" of 12towers
HadI solation 1
"Ring" of 12towers
"Ring" of 12towers
HadI solation 2
22 towersinside "ring"
N/ A
RoI Cluster 22 towers,em+had
22 towers,em+had
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 6
Hadronic Isolation: “Core” & “Ring”
Performance IssuesImbalance between the two regions
“core” provides greater part of jet rejection
“ring” is more vunerable to noise
Separation allows different weighting of the two
same (or slightly better) overall performance
more flexible response to conditions & requirements
ET in 2 regions (signal & background)
Title:/disk/r9a/home/atw/atlas/trigger/results/98/core_anCreator:HIGZ Version 1.23/09Preview:This EPS picture was not savedwith a preview included in it.Comment:This EPS picture will print to aPostScript printer, but not toother types of printers.
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 7
Core & Ring Isolation: Signal vs Background
Signal:Look at electrons + 48 mbias
Pessimistic pileup modelParameterize BCID
More ET in “ring”
Two sums largely uncorrelated
Background:“Core” ET typically larger
Correlations weak
“Ring” contributes to rejection
Title:/disk/r9a/home/atw/atlas/trigger/results/98/2d_compCreator:HIGZ Version 1.23/09Preview:This EPS picture was not savedwith a preview included in it.Comment:This EPS picture will print to aPostScript printer, but not toother types of printers.
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 8
Effect of two-region hadronic isolation
Compare rates for same isolation efficiency look at electrons with
“pessimistic” pileup
compare jet background rates for cuts giving 95%, 98% & 99% isolation efficiency
find some improvement when very high efficiency required
look also with “more realistic” pulse-shape modelling
improvement smaller (few %)no situation in which rate is worse
0
100
200
300
400
500
600
Rat
e (a
rbit
rary
uni
ts)
e >
95%
e >
98%
e >
99%
4x4 had
core+ring
Cluster > 15 GeV, em & had isolation
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 9
A Common RoI Algorithm?
Motivations Simplify implementation
only 1 set of RoI clusters to form & testmakes “selectable” em/tau triggers easier to build
Simplify RoI data
single object produces unique RoI coordinate for both em and tau algorithms
Which algorithm? Em RoI for taus?
Much of tau ET may be in HCAL
May degrade shower containment
Not ideal Tau RoI for e/?
Isolated e/ should deposit little ET in HCAL
Objects for which it makes a difference will fail hadronic isolation anyway
Worth investigating
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 10
Possible Effect: Most Extreme Case
Possible Effects:RoI coordinate may shift by 0.1
Rare (< 1% of e/)
Clusters for which this happens will mostly fail isolation anyway
Possible effects of this:
Moves hadronic ET between “ring” and “core” sums
Very rarely may increase em isolation or reduce trigger cluster
Size of effects: Changes in em sums can be no
larger than hadronic ET deposit
em RoI em + had RoI
em ET deposit hadronic ET deposit
t r igger cluster RoI and windowboundaries
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 11
Effect on Efficiency: 30 GeV e- + pileup
What effects do we see:(with pessimistic pileup model)
No effect in 99% of e RoIs
Small differences in isolation ET have no effect on efficiency.
See same story for: Different electron pT
“ noise/BCID assumptions
“ pileup levels
Title:/disk/r9a/home/atw/atlas/trigger/results/98/roi_isoCreator:HIGZ Version 1.23/09Preview:This EPS picture was not savedwith a preview included in it.Comment:This EPS picture will print to aPostScript printer, but not toother types of printers.
em RoI em+had RoI
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 12
Effect on Jet Rejection
More effect seen in background: O(10%) of background clusters
with ET > 15 GeV have RoIs shifted (cf < 1% e/)
Does it affect jet rejection? Use pessimistic pileup (as
should maxmize size of any effects)
Choose isolation cuts to give 95%, 98% and 99% isolation efficiency
Compare jet rates for 2 RoI algorithms
small reduction in rate seen
0
20
40
60
80
100
120
140
160
180
Rat
e (a
rbit
rary
uni
ts)
e > 95% e > 98% e > 99%
em RoI
em+hadRoI
Cluster > 15 GeV, em & had isolation
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 13
Efficiency in Physics Events
Look at effect in physics events In case more sensitive than
single e/ events
Look at a range of processes containing e/ plus jets
What do we see? For most processes, no effect
Greatest effect in t eb, jjb
1% events have shifted RoIsSmall changes in Had Isoln.No difference in efficiency
Had isoln sums, e from top eventsTitle:/disk/r9a/home/atw/atlas/trigger/results/99/top_roiCreator:HIGZ Version 1.23/09Preview:This EPS picture was not savedwith a preview included in it.Comment:This EPS picture will print to aPostScript printer, but not toother types of printers.
LHCC Review, CERN, 19/10/99Paul Bright-Thomas, for Alan Watson 14
Summary of Changes
Nem and Ntau variableIncreases flexibility of systemNo possible performance drawback
Two-region Hadronic IsolationMay increase flexibility/robustness of performanceSlight improvement in jet rejection.
Common RoI AlgorithmTechnical simplificationNo adverse effect on performance (maybe very small
gain)