reconstruction tools for the study of short-lived
DESCRIPTION
Reconstruction tools for the study of short-lived resonances in ALICE pp collisions at the LHC startup. F.Riggi Dept. of Physics and Astronomy and INFN, Catania. Outline. The ALICE experiment @LHC Short-lived resonances in ALICE pp collisions: preparing the tools for LHC startup - PowerPoint PPT PresentationTRANSCRIPT
Reconstruction tools for the study of short-lived resonances in ALICE pp collisions at the LHC startup
1. The ALICE experiment @LHC2. Short-lived resonances in ALICE3. pp collisions: preparing the tools for LHC startup4. Results from local and distributed computing analysis
Outline
F.RiggiDept. of Physics and Astronomy and INFN, Catania
ACAT 2007, Amsterdam, 23-27 April, 2007
Z
The Large Hadron Collider (LHC) @ CERN
4 major experiments waiting for the first beam…
ALICE Detector @ Point2
Point 2
ALICE: A Large Ion Collider Experiment
ITS
TPC
TOF
PT > 1 GeV/c
ALICE will track and identify products from pp and nucleus-nucleus collisions in a large multiplicity environment
Main tracking detectors: ITS and TPC
The silicon Inner Tracking System (ITS)
Drift (SDD) 133,120 channels
Strips (SSD) 2,608,128 channels
Pixel (SPD) 9,830,400 channels
Time Projection Chamber (TPC)
Channels 557 568
Gas Ne/CO2 90/10 or + 5%N2
Volume 88 m3
Drift length 2.5 m
Drift field 400 V/cm
Drift velocity
2.84 cm/ms
Max drift time
88 ms
Arrival of the TPC in the ALICE cavern
Possible ALICE physics at the LHC startup (pp @0.9 TeV?)
Physics aspects Compare pp and pp data at 900 GeVCompare pp and pp data at 900 GeV Improve the precision in comparison with existing pp data at 900 GeVImprove the precision in comparison with existing pp data at 900 GeV Prepare the detector (calibration, alignment)Prepare the detector (calibration, alignment) Provide a reference for data at different cm energies (2.4 TeV, 5.5 TeV?, 14 TeV). Provide a reference for data at different cm energies (2.4 TeV, 5.5 TeV?, 14 TeV). Use first pp events to study global event properties:Use first pp events to study global event properties: Charged Multiplicity, Pseudorapidity distributions, pCharged Multiplicity, Pseudorapidity distributions, pT spectra spectra
First strange particle studiesFirst strange particle studies Short-lived resonances ?Short-lived resonances ?
Ks0 P P
Yield perevent
0.1 0.01 210–4 10–5 0.4 0.4
Statisticsneeded
104 104 104 104 104 104
PP eventsneeded
105 106 108 109 104 104
Resonance Life-time [fm/c] 1.3++ 1.7 f0(980) 2.6 K*(892) 4.0 (1520) 13 ω(783) 23(1020) 45
Short-lived resonances have typical lifetimes comparable to that of the hot and dense matter created in AA collisions
Time
ch
em
ical fr
eeze-
ou
t
signal lost
kin
eti
c f
reeze
-ou
t
signal measured late decay
signal measured
rescattering
regeneration
e+
e-
signal measuredearly decay
+
-signal measured late decay
e+
e-signal measuredlate decay
Observation of short-lived resonances is affected by two competing effects: rescattering vs regeneration
Resonance K*(892) Φ(1020) *(1520) Δ(1232)Decay channel (B.R.) K (~100%) K+K- (49%) N K (45%) Nπ (100%)
Width [MeV/c2] 50.8 4.5 15.6 100 Life time [fm/c] 3.9 44 13 1.7
Study of short-life resonances in ALICE
Main observables:
● Extraction of the signal/yields● Mass and widths of resonances● Transverse momentum and transverse mass spectra● Particle ratios● Elliptic flow● Nuclear modification factors: RCP and RAA
pp collisions @900 GeV and @14 TeV
● LHC running scenario at startup: 1-2 days for physics collisions@900 GeV ?
● Expected no. of events: 0 to 106
● To what extent is the study of short-lived resonances feasible?
Results from:
● Local analysis via LSF (@900 GeV): 2 x 105 events
● Distributed GRID analysis (@14 TeV): 1.5 x 106 events
Tuning of reconstruction tools: ● Study of the combinatorial background ● Perfect and realistic Particle IDentification
Detailed examples for K*(892), other resonances treated in the same way
Reconstruction of short-lived resonances requires optimal performance on:
primary and secondary vertex reconstruction
Efficiency
An example of D0-> K-π+
decay in a large multiplicity environment
σ ~ 40 μm for pp
σ ~ 5 μm for PbPb
3D resolution
Reconstruction of short-lived resonances requires optimal performance on:
primary vertex reconstructionprimary vertex reconstruction
tracking efficiency
Kalman filter strategy allows a good tracking performance down to very low momenta
Methods based on neural networks also implemented for stand-alone tracking in the ITS: 10% improvement
ITS + TPC
Reconstruction of short-lived resonances requires optimal performance on:
primary vertex reconstructionprimary vertex reconstruction
tracking efficiencytracking efficiency
momentum resolution
pT resolution
The inclusion of ITS, TPC and TRD results in pT resolution as good as 3 % up to 100 GeV/c
Reconstruction of short-lived resonances requires optimal performance on:
primary vertex reconstructionprimary vertex reconstruction
tracking efficiencytracking efficiency
momentum resolutionmomentum resolution
track impact parameter
Information on impact parameter mainly provided by the ITS
Transverse impact parameter resolution
Reconstruction of short-lived resonances requires optimal performance on:
primary vertex reconstructionprimary vertex reconstruction
tracking efficiencytracking efficiency
momentum resolutionmomentum resolution
track impact parametertrack impact parameter
particle identification
with ITS and TPC at low momenta…
and TOF at high momenta…
w(i | s)=r(s | i) Ci
Σk r(s | k) Ck
Bayesian PID for each detector
Inside same event, correlations between
K+ and π- candidates
K- and π+ candidates
Evaluate invariant mass spectrum
Combinatorial background
Signal extraction (unlike-sign)
Mixed-event technique
Like-sign technique
Example: K*(892) Kπ (~100%)
The AliROOT framework
STEERBase classes, overall control
AliReconstruction ESDReconstruction
Event Summary Data
Particle generation
DPMJET
HIJING
HBTP
PYTHIA PDF
ISAJET
Transport of particles
GEANT3 GEANT4 FLUKA
Analysis
HBT JET PWG0-4
Detectors
ZDCPHOS
EMCALHMPID
TOF
TRD
TPC
ITS
T0 V0PMDMUON
Response
Geometry Calibration
Alignment
Selection of primary and identified tracks
Size reduction by 200
AOD:
Analysis Object Data
Analysis Cuts,histograms,…
From generation and reconstruction
ESD: Event Summary Data
Kinematics
…
Local analysis of 200,000 pp min bias events via
LSF (Load Sharing Facility, A general purpose distributed computing system)
For such application:
● Cluster of 60 multiprocessors CPU (up to 30 simultaneous jobs running)
● Generation and full reconstruction of events (1 job= 100 events)
● CPU time/event: 110 s @ 900 GeV ( 230 s @14 TeV)
● Total CPU time: approx. 200 days (about 10 effective days)
No event selection
Only events with Δm<5 and Δzv < 3 cm mixed
Effect of event selection (multiplicity and vertex location) on mixing procedure
Selection in multiplicity is more effective than in primary vertex location
(True Background)
(Mixed events background)
True background = (Signal) – (True pairs)
Like-sign technique
Signal
Background
Effect of particle identification on the results
PID information from TPC and TOF used in the present analysis, with Bayesian procedure
Different scenarios
● Perfect PID● Realistic PID with no threshold on max_probability● Realistic PID with improvement both in K and pion identification● Realistic PID with improvement in K identification /no PID on pions
K*(892) invariant mass with perfect PID
Improving kaon identification in the Bayesian algorithm by a threshold on max.probability
From max_prob > 0 to max_prob > 0.7
No thresh on maxprob
Maxprob > 0.7 (K)
No PID (π)
Maxprob > 0.7 (K)
Maxprob > 0.7 (π)
Perfect PID Realistic PID
Summary of PID influence on K*(892) reconstruction True
Found
S/B = 0.11S/√B = 20.28
S/B = 0.111S/√B = 20.01
Φ(1020)
Λ*(1520)
Similar analyses also done on other short-lived resonances
Yields and particle ratios uncertainties expected at 900 GeV with 200K pp events
Particle ratios Statistical
K*/K- 1.6 %
Λ*/Λ 9 %
Φ/K* 8 %
Φ/Λ* 12 %
K*/Λ* 9 %
REAL DATA (MB) MonteCarlo (MB)
Raw ESD AOD Ev. Catalog Raw ESD
pp / event 1 0.04 0-004 0.01 0.4 0.04
pp / year (x 109) 1.1 0.18 0.4 0.1
HI / event 12.5 2.5 0.25 0.01 300 2.5
HI / year (x 109) 1.4 1.0 3.0 0.9
More detailed analyses require a large number of events
Distributed analysis of 1.5 Mevents pp min bias events @ 14 TeV
from PDC06 on the GRID via AliEn (ALICE Grid Environment)
Submitting and monitoring jobs on the GRID via AliEn
Distributed computing on the GRID:
~1.5 Mevents (pp @14 TeV) processed on the grid via AliEn middleware
Monitoring the jobs on the GRID in ALICE
Typical results at 14 TeV from distributed analysis on the GRID
pT-analysis with realistic PID
pT= 0 - 0.5 pT= 1.5 - 2
pT= 3.5 - 4
Correction matrix
pT=0-4 GeV/c (8 bins)
Y = -1.5 – 1.5 (4 bins)
Summary
● Study of short-lived resonances in pp and PbPb collisions @ LHC energies could be studied in ALICE from the very beginning
● With a small sample of events [O(105)] and realistic PID:
● Extraction of yields at least for K*(892), Φ(1020), Λ*(1520)
● Rough pT - distribution for K*(892) up to 1.5-2 GeV/c
● Particle ratios Φ/K*, Λ*/K*, Φ/Λ* measurable
● Analysis of O(106) pp events at 14 TeV fully reconstructed on the GRID
● Resonance yields with large statistics
● pT-analysis
● Correction matrix (y,pT)
● Extension to other resonances in progress