20131018_STEP - University of Birmingham

Download Report

Transcript 20131018_STEP - University of Birmingham

Improving ATLAS hard
diffraction measurements
with the STEP award
Hardeep Bansil
University of Birmingham
18/10/2013
Diffractive Dijets at ATLAS
Look for single diffractive events (pp → pX)
Involves a large rapidity gap due to colourless exchange with
vacuum quantum numbers: “pomeron ( IP )”
Search for diffraction with a hard scale set by 2 jets
Described by diffractive PDFs + pQCD cross-sections
IP
Key variables
ΔηF - largest forward gap from either side of ATLAS detector
acceptance
ξ - fraction of momentum of proton transferred to pomeron
Transverse momentum (pT) and pseudorapidity (η) of jets
Quantify amount of rapidity gap survival probability
Work presented in thesis includes full systematic treatment
and represents first measurement at ATLAS
Still could benefit from improvements before being put
forward as official ATLAS measurement for publishing
2
Diffractive Dijets at ATLAS
Rescatter with p?
Look for single diffractive events (pp → pX)
Involves a large rapidity gap due to colourless exchange with
vacuum quantum numbers: “pomeron ( IP )”
X
Search for diffraction with a hard scale set by 2 jets
Described by diffractive PDFs + pQCD cross-sections
Key variables
ΔηF - largest forward gap from either side of ATLAS detector
acceptance
ξ - fraction of momentum of proton transferred to pomeron
Transverse momentum (pT) and pseudorapidity (η) of jets
IP
Quantify amount of rapidity gap survival probability
Work presented in thesis includes full systematic treatment
and represents first measurement at ATLAS
Still could benefit from improvements before being put
forward as official ATLAS measurement for publishing
3
Improvement 1: Monte Carlo production
Current analysis only uses one set of Monte Carlo samples, generated using PYTHIA8, to provide
description of non-diffractive and diffractive dijets
Initial plan was to use two different models (PYTHIA8 and HERWIG++)
Test effect of parton showering / hadronisation models on rapidity gap distributions and gap
survival probability
HERWIG++ could not be used: Problems with
underlying event model produce events with very
large gaps
PYTHIA8 diffractive samples use unconventional
pomeron flux parametrisation (black) based on past
experiments (blue/green)
PYTHIA8 - generate new samples or correct existing
diffractive samples
Generate new HERWIG++ samples with improved
underlying event
Allow rapidity gap survival probability to be measured
Also generate new samples using other generators e.g. POMWIG
Further investigate model dependencies, allow zIP to be unfolded
Expected time for generation: 2 months
4
Improvement 2: Unfolding
Transforms data to underlying true distributions by correcting for effects of finite resolution and
limited acceptance of the detector
Correction process uses RooUnfold software to unfold individual distributions separately but there
are two exponentially falling distributions in this analysis, ∆ηF and jet pT
Both produce a net migration to larger rapidity gaps
Currently controlled by optimising jet pT scale for current analysis
Improve by using simultaneous unfolding procedure for both variables
Migrations will be better controlled and there should not be a necessity to correct the jet pT scale as
was done in this analysis
Currently investigating modified version of RooUnfold used by other analyses and Imagiro software
to perform this
Speaking to authors of both software packages
Expected time for updating unfolding process - 1 month
5
Improvement 3: Trigger
Current trigger method selects jets using a hardware based jet trigger (L1_J5) and minimum bias
trigger (L1_MBTS_1) and per-jet strategy to check whether both jets are capable of triggering event
Selection via jet trigger dependent on pT and η of jet - allowed to drop down to 80% to increase
statistics but reduces precision of analysis
Potentially effects measurement
L1_MBTS_1 fully efficient but prescaled (accepts 1 in every 50 events triggered)
Analysis uses multiple data streams – overlap removal applied but potentially loses useful events
Trialing new trigger strategies
Simplify overlap removal / per-jet triggering strategy to single event level trigger
Keep existing strategy but select jets up to 100% with J5
Use different minimum bias trigger only (L1_MBTS_2)
Unprescaled, fully efficient and has highest trigger rate for all data used
Expected time for changes: 1-2 month(s)
6
Paper to publish + conference
Aim is to get analysis published in e.g. Physics Review Letters
Present results to Soft QCD Working group (21/10/2013) and then to ATLAS Standard Model
working group at earliest opportunity
Get editorial board as soon as possible – even if analysis not fully complete (typical for other
analyses)
Paper should be relatively quick to write now that thesis has been written as it will be more
condensed form of this analysis
Expecting paper to be completed over space of 1 month
Expecting editorial review process to take up to 3 months (common for most ATLAS analyses)
Accounts for rewrites and potential corrections / additional studies
Can start writing a lot of this already with plan to start editorial review process in January 2014
Additional aim is to represent ATLAS at international conference
Currently targeting “Multiple Partonic Interactions @ LHC”, Antwerp, Dec 2-6
Otherwise will look to conferences in early 2014
7
Future projects
Projects to work on in parallel with improving current analysis
Aim not to complete analyses but to lay groundwork for future studies
Diffractive charm
Based on previous work done at HERA (ep collider)
Can work along Birmingham colleagues involved with previous charm studies
Code for D* meson production already available
Need to retrieve files for analysis to do first feasibility studies (2 weeks)
ALFA
Roman pots located 240m from centre of
detector used in dedicated runs
Remove ambiguity for single diffractive events and allow for pomeron structure studies, similar to
previous experiments
Data from 2011/2012 needs to be reprocessed to include jets (maximum 2 weeks)
ATLAS results for elastic process (pp → pp) are close to completion, can work closely with experts
involved
8