Towards a diffractive trigger stream in CMS

1
Towards a diffractive trigger streamin CMS
Monika Grothe U Turin/ U Wisconsin Manchester Dec
2005
Studies presented in the following were done in
collaboration with TOTEM Part of material is
included in the HERA-LHC workshop
proceedings www.cern.ch/grothe/heralhc/heralhc.ps
All material is preliminary
Lions share of work done by PhD student Richard
Croft of Bristol and Masters student Fredrik
Oljemark of Helsinki
2
What happened since Manchester 04

• Looked into CMS L1 muon trigger and CMS L1 jet
  trigger
• Goal Optimize working point in 2-dim parameter
  space
• bandwidth versus signal efficiency
• Started out with most challenging case 120 GeV
  Higgs decaying to b bbar
• Without fwd detectors, find signal
  efficiencies
• CMS L1 muon trigger 10 (in WW
  decay mode _at_140 GeV 20)
• CMS L1 jet trigger 0
• Find that we can lower CMS L1 jet trigger
  thresholds sufficiently to get
• 15 signal efficiency when combining with
  fwd detector conditions
• Feasibility study evolved into attempt at
  defining a dedicated diffractive
• trigger stream with target output rates of 1
  kHz out of 100 kHz total
• on L1 and 1 Hz out of 100 Hz total on HLT
• Ultimate goal diffractive trigger table that
  covers double-Pomeron
• exchange and single diffractive processes

3
Part I Central exclusive diffractive production
of low mass Higgs
4
H (120 GeV, DPE prod) ? b bbar Why is it so
difficult ?

• L1 jet trigger signature 2 jets in CMS Cal,
  each with ET lt 60 GeV
• Measured L1 jet ET on average only 60 of true
  jet ET
• L1 trigger applies jet ET calibration and cuts on
  calibrated value
• Thus 40 GeV (calibrated) 20 to 25 GeV measured
• Cannot go much lower because of noise
• Use rate/efficiency _at_ L1 jet ET cutoff of 40 GeV
  as benchmark
• L1 2-jet rate for central jets (hlt3) _at_ L1 jet
  ET cutoff of 40 GeV for
• Lumi 2 x 1033 cm-2 s-1 50 kHz , while
  considered acceptable O(1 kHz))
• Need additional conditions to trigger a 120 GeV
  Higgs with L1
• Forward detectors !

5
Central detector jet trigger alone

• 4x4 trigger towers region
• Search for jets with a
• sliding 3x3 regions window
• Jet 3x3 region with local energy
• max in middle
• Reconstructed L1 jet ET on
• average 60 of real jet ET,
• thus need for jet ET calibration
• Jet 144 trigger towers,
• with typical jet dimensions
• Dh x Df 1 x 1

HT condition isolation condition for jets 2
jets in central Cal (hlt2.5) with ?(ET 2
jets)/HT gt threshold HT scalar sum of ET of
all jets in the event with ET(jet)gtthreshold
--gt Provides factor 2 rate reduction
6
Require hits in 220 m Roman Pots (I)

• single-arm 220 m condition
• L 1032 cm-2s-1, i.e. case with no pile-up

Excellent suppression of QCD background rate
reduction 350 at 40 GeV jet threshold for
L1032 cm-2s-1 However
7
Require hits in 220 m Roman Pots (II)
However, rejection factor becomes small when
pileup (contains diffractive component) is
included
single-sided 220m condition without and with cut
on ?
achievable total reduction 10 x 2 (HT cond) x 2
(topological cond) 40
8
Require hits in 220 m Roman Pots (III)

• Possibility of setting a x cut in the RPs, e.g.
  x lt0.1 (recall acceptance is
• 0.02 lt x lt 0.2) to reduce contribution from
  outside diffractive peak
• Topological condition
• Can win additional factor 2 in rejection when
  requiring that the 2 jets
• are in the same ? hemisphere as the RP
  detectors that see the proton

xLP/Pbeam 1-x
9
Require hits in 220 m Roman Pots (IV)
For H (120 GeV, DPE prod) ? b bbar, adding L1
conditions on the RPs at 220m is likely to
provide a rate reduction sufficient to meet the
CMS L1 bandwidth limits at luminosities up to 2x
1033 cm-1 s-1
To go even further up in luminosity need
additional handle to stay within bandwidth
limits ... So what about triggering with the
420 m RPs ? At the current CMS L1 latency of 3.2
?s they are too far away from IP for inclusion
in L1 Note This is a hardware limit - cannot be
changed without replacing trigger pipelines of
CMS tracker and preshower detectors with deeper
ones Should this however happen (under discussion
for SLHC L1 latency 6.4 ?s, determined by ECAL
pipeline depth) then ....
10
Require hits in 220 m and 420 m RPs

• Asymmetric condition
• Require hits on one side in 220m RPs and on one
  side in 420m RPs
• In effect means on opposite sides events where ?
  values of 2 protons are very different
• Can be used either in L1 after increase in L1
  latency, or on HLT

For H (120 GeV, DPE prod) ? b bbar, adding L1
conditions on the RPs at 220m and 420m would
provide a rate reduction sufficient to meet the
CMS L1 bandwidth limits at luminosities up to
1034 cm-1 s-1
11
Require hits in 420 m RPs only

• Double-sided 420 m condition
• Only possible after increase in CMS L1 latency
• Would allow to select events that are
  gold-plated wrt mass resolution
• Note Single-sided 220 m cond and asymmetric
  cond select events with
• worst possible mass resolution
• Sorry, no numbers yet....

12
Background in RPs
Beam-halo/beam-gas level numbers produced by
TOTEM not a problem as soon as central CMS
detector condition is used in L1 Find from
pythia pile-up sample _at_220m 0.012 protons per
pile-up event on average, i.e. at 1034 cm-2s-1
350.0551.93 _at_220m In worst case on average
1.93 tracks from pile-up in addition to track
from signal event _at_420m 0.055 protons per
pile-up event on average, i.e. at 1034 cm-2s-1
350.0120.42 _at_420m In worst case on average
0.42 tracks from pile-up in addition to track
from signal event
13
Part II Single-diffractive dissociation
14
SD production of W, Z, dijets

• Map the parameter space (bandwidth vs efficiency)
  with ultimate goal
• of defining a trigger table for a dedicated
  diffractive trigger stream with
• target output rates of 1 kHz (L1) and 1 Hz (HLT)
• Use Pomwig Monte Carlo
• Recall
• RP acceptances at ? 0.5m
• 220m - 0.02 lt ? lt 0.2
• 420m 0.002 lt ? lt 0.02
• Lowest threshold for L1 jet trigger is ET gt 40
  GeV
• Typical loss of factor 2 in efficiency when
  using 220 m RP condition (RP acceptance)

Pomwig ?-spectra for leading proton in dijet
events
.
Pomeron
Reggeon
15
L1 efficiencies for SD W prod
No RP cond
W Muons (Pom)
W Jets (Pom)
220m S cond
420m S cond
W Muons (Reg)
W Jets (Reg)
Plots normalised to the number of events in the
range 0.001 lt ? lt 0.2 .
16
 
 
1 kHz jet thresholds
Estimated 1kHz Jet Thresholds for various Central
/ RP conditions
S single-sided, D double-sided C ?lt0.1 of the
leading proton
17
Summary

• To trigger on L1 H (120 GeV, DPE prod) ? b bbar
  Reduce to O(1 kHz)
• L1 2-jet rate for central jets (hlt3) _at_ L1 jet
  ET cutoff of 40 GeV
• Triggering in absence of pile-up no problem
• By combining RPs at 220m with jet trigger a QCD
  rate of a few kHz achievable at 2x 1033 cm-1s-1,
  i.e. consistent with CMS L1 bandwidth
  restrictions
• Signal efficiency of order 15
• Using muon trigger increases signal efficiency by
  10
• Have started looking into single-diffractive
  production of W, Z, dijets
• Goal Define trigger table for a dedicated
  diffractive
• trigger stream with target output rates of 1 kHz
  out of 100 kHz total
• on L1 and 1 Hz out of 100 Hz total on HLT

18
RESERVE
19
Forward detectors
CMS IP T1/T2
RPs _at_ 150m
RPs _at_ 220m

RPs_at_420m
TOTEM detectors T1 (CSC) in CMS endcaps T2 (GEM)
in shielding behind HF T1 T2 3 ?
6.8 Roman pots (Si) on 2 sides at up to 220 m
Acc. for nominal LHC optics 0.02 lt x lt
0.2 Under discussion RPs at 420 m Acc. for
nominal LHC optics 0.002 lt x lt 0.02 CMS Castor
calorimeter, downstream of T2
T2
T1
T2
20
Pile-up studies
Soft diffractive elastic events contribute
substantially to pile-up Crucial to study
impact of pile-up on RP L1 condition Prerequisit
e Realistic MC simulation of diffractive
events in pile-up Pile-up in CMS generated with
Pythia, compare to HERA and pp leading proton
data
Diffractive peak
Leading protons
xLP/Pbeam 1-x
of fastest proton in event
21
Pile-up studies (II)
xLP/Pbeam 1-x

• Pythia too low outside diff peak
• Pythia approx ok in diff peak
• after taking shrinkage
• (b b0 4 alphaI ln s)
• into account

xLP/Pbeam 1-x

• Pythia wrong in shape normal.
• outside diff peak ( factor 2-3)
• Pythia approx ok in diffractive peak

Plots Marta Ruspa