Transcript 20111212_SoftQcd - University of Birmingham

Update on Diffractive Dijets
Search
Hardeep Bansil
University of Birmingham
SM Soft QCD meeting
12/12/2011
Diffractive dijets
• Hard diffraction
– Look for single diffractive events (pppX)
involving a large rapidity gap from pomeron
– Then look for dijet system within X
X
• Studied at HERA and Tevatron
– At Tevatron, ratio of yields of SD to inclusive
dijets ≈ 1%
– Measure ratio at LHC (expected to be smaller)
Rescatter
with p?
• Gap Survival Probability
– The chance of the gap between the intact
proton and diffractive system being lost due
to scattering (affects measured structure
function)
– At Tevatron GSF around 10 compared to H1
– Measure this at LHC (Khoze, Martin and
Ryskin predict GSF around 30)
• Study jet and gap properties
(ξ)
Gap destruction by
secondary scattering
2
Gap Finding Algorithm
• Found out that gap finding algorithm is not fully up to date
• Detector gap definition is currently missing pT cut on clusters
associated with cell
Divide calorimeter into 49
rings of 0.2 in η
Detector gap definition
•Calorimeter: no cell
above threshold E/σ > Sth
- probability of noisy cell
in ring smaller than 10-4
(electronic noise only, no
pile-up environment)
• Looked at Rapidity Gaps paper for
comparison of algorithms
– Roughly a 10% difference at small gaps
(ΔηF < 2.0) and then 5% difference
beyond this so will correct for this soon
•Tracker: no good track
above pT > 200 MeV, |η|
< 2.5
Truth gap definition
•No stable particle above
pT > 200 MeV
3
Analysis
• Using Athena version AtlasProduction-16.6.4.2
• Using MinBias stream data10 period A and B ESDs
– Run 153030 (period A) excluded due to noise bursts in TILE
– Total ∫L dt = 8.71 nb-1 - calculated using online iLumiCalc tool with
L1_MBTS_2 ref. trigger
• Average <μ> for selected runs < 0.15  currently ignore pile-up
• Anti-Kt jets with R=0.6 or R=0.4:
– Require >= 2 jets in event
– ET Jet1,2 |η| < 4.5
– ET Jet1 > 26 GeV, ET Jet2 > 20 GeV for asymmetric jet ET cuts (NLO), cut
values suggested based on work by Radek Zlebcik (Prague)
– Jet ET Jet2 limit and η cuts based on jet energy scale systematic
– Currently no requirements to ask about jet quality cuts
• Ask for a forward gap: |ηstart| = 4.9, ΔηF ≥ 2.0
4
Monte Carlo for Analysis
• Currently using POMWIG LO generator as sample of single
diffractive jets
– Modifies HERWIG ep photoproduction so ee+γ becomes pp+IP
– No rapidity gap destruction built in
– Generates QCD 22 process within diffractive system in different pT
ranges (8-17, 17-35, 35-70, 70+ GeV) for SD (system dissociating in ±z
direction)
– Using MC samples generated by myself (4000 events of each
POMWIG sample)
• Have PYTHIA 6 and PYTHIA 8 Jets samples to use as inclusive
jets sample (8-17, 17-35, 35-70, 70-140 GeV)
– PYTHIA 8 J0 sample (8-17 GeV) available only with pile up on Grid
– Generated a small sample for temporary use (5000 events)
Do Pythia jets samples contain any
models of diffraction within them?
• NLO comparison would be ideal
– Need someone to work on this (Radek from Prague not working on
this anymore)
5
Uncorrected Gap Size Distribution
POMWIG SD,
PYTHIA 6 & 8 Jets
weighted relative
to luminosity of
data runs used and
then plotted
against MinBias
Data
MinBias Data
Pomwig SD
Pythia 8 Jets
Pythia 6 Jets
AntiKt6 jets, ET Jet2 > 20 GeV
After Forward Gap Cuts
•
Big factor between PYTHIA and data in the first bin
– Still have it after many checks (see next slide)
– Ratio of MC to Data suggests a GSF of 15-25 in majority of bins but need to
ensure normalisation is correct first
6
Normalisation Issues
•
Make MC samples compatible with data - scale them to luminosity of
data by applying weighting factor to variables (MC run dependent)
•
•
•
•
Ldata – luminosity of data
Ngen – Number of events generated
σgen – Csx of events generated
Nrec – Number of events reconstructed
to run over
•
Ngen , Nrec,σgen all taken from AMI, checked multiple times but still get big
factor between data and MC
AntiKt6 Truth Jets, No Jet ET Cuts, Before Forward Gap Cuts
•
Truth level cross sections plotted
–
–
•
Weight 
Ldata
 N rec N gen 



 N gen  
gen 

Show smooth transitions between
different pT range samples
Current approach appears correct?
Need to understand and fix this
issue
–
–
Could it be due to running on this
specific data set?
May require checking code line by
line!
7
Tips from CMS Diffractive W/Z search
• CMS have had difficulty in trying to
find diffractive W/Z signals
–
–
–
–
–
η~ = 5 - ∆ηF (similar to forward gap)
300 of 40000 W/Z events have gap
Pythia 6/8 tunes (ND) plotted v data
No clear signal above ND
Something to try with dijets?
• More luck in studying asymmetry
of W being in the same
hemisphere as the gap
– POMPYT (diffractive) incl. with ND
– Determined that 50 ± 10% of events
with a forward gap > 1.9 are
diffractive
– Make jet signed η distributions
• CMS possibly presenting results on
diffractive dijets at EDS 2011
– Should be similar to diffractive W/Z
8
Tips from CMS Diffractive W/Z search
• Signed η distributions
• Cover a much larger range in η
than W/Z search
• All histograms have similar shape
at the moment
– Expect Pythia 6/8 Jets to be flatter like CMS?
9
Noise Study in “Empty” Events
•
•
•
Preliminary look at potential effects of noisy cells destroying gaps in
empty events by using RNDM stream
Define event as empty by having no reconstructed primary vertex (with
5+ associated tracks) + no MBTS counters fired
See that not all events have a full gap (∆ηF=9.8) – 47556/7380809 events
–
Less than 1% of events in stream
Significant Cells η
•
Significant Cells φ
Looking at cells with high energy significance suggests slightly more
activity at +η and φ=-π
– Which events pass “Empty” cuts but still have activity?
– Is there a new noise burst to consider in these runs? Change GRL?
– Slightly different point but will OTX problems need to be accounted for?
10
Ratio of SD to ND Dijets
• Preliminary study to measure the ratio of the single
diffractive to inclusive dijet events based on events passing
jet cuts as well as gap cuts
• Only done on MinBias stream (RNDM stream has very low
statistics in comparison)
MinBias stream Data : 2010 Period A&B (91465899 events total)
AntiKt4 Jets (ND)
89004
AntiKt6 Jets (ND)
191682
AntiKt4 Jets + ∆ηF>2.0 (SD)
173
AntiKt6 Jets + ∆ηF>2.0 (SD)
317
AntiKt4 SD/ND ratio (%)
0.194
AntiKt6 SD/ND ratio (%)
0.165
• On current results, ratio around 0. 2%
• Hard to make direct comparison with Tevatron as cuts are not
similar + corrections not applied to my results yet
• CMS W/Z ratio (on earlier slide) around 0.75%
11
Comparison with Prague
• Group in Prague also looking at diffractive dijets
• Earlier in analysis but already see difference in strategies
Birmingham
Prague
Data used & Stream
MinBias 2010 A&B
L1Calo/JetTauEtmiss - all 2010
Vertex
No requirement
1(+?) vtx (5+ tracks)
Trigger
MBTS_2
MBTS + Jet Pt dependent
Preferred SD MC
Pomwig SD
Herwig++ SD
Preferred ND MC
Pythia 8
Herwig++ ND
• Complementary strategies
• Vertex requirement would force ourselves to smaller gap
sizes (less likely to see forward jets)
• Prague agree very well with inclusive jets analysis for 2010
– How do I test this for only Periods A and B?
12
Comparison with
Prague
•
•
•
Birmingham
Prague
Preferred SD MC
Pomwig SD
Herwig++ SD
Preferred ND MC
Pythia 8
Herwig++ ND
Herwig++ is C++ version of Pomwig but has known “feature”: factor of 3
increase in cross section
Issue with compatibility of hadronisation models between Pomwig
(cluster model) and Pythia (string model)
Herwig++ ND produces events with very large gaps (truth level study)
– Unexpected?
But we want
our samples
to go out further
in gap size
anyway?
Marek’s plots comparing ND
Pythia and Herwig++ Jx samples
•
•
Will compare HERWIG++ & POMWIG SD samples soon
–
Generated EVNT samples, trying to implement Fastjet to make direct comparisons
Will need to get official Monte Carlo production done soon
– Need to settle on stats required / new filter to simulate larger gaps in events
with dijets / move from ESD to dAOD with additional cluster info
13
Next steps
• Main aim: Cross section as a function of
gap size
– similar to the soft diffraction paper, but with
a jet requirement
– Interpretation of GSF can come later
• Correct gap algorithm
• Fix normalisation issues
• Compare different generators at EVNT level with
Fastjet added in
• Decide on appropriate statistics with Prague group
14