Title: I%20primi%20108%20trigger%20in%20ATLAS
1I primi 108 trigger in ATLAS
- Aleandro Nisati, per ATLAS
2Some assumptions
- ltLgt 1031 cm-2 s-1
- Event Filter rate 100 Hz (108 events
collected) - ? T 106 s
- ? L 10 pb-1
- Not so much for physics
3outline
- Trigger menu
- Calibrations
- Data processing
- Computing
- Physics
- more emphasis on italian commitments
4Example of a trigger menu (L6x1031 cm-2
s-1)
- Physics signatures are selected with cuts that
are less stringent than the ones for the low
luminosity runs. - Rates are indicative!
- mu 6 GeV (2) 10 Hz
- mu 20 GeV 4 Hz
- E/gamma 15 GeV 25 Hz
- Jet pTgt 150 GeV 6 Hz
- Missing Energy(40 GeV)jet 40 GeV 30 Hz
- Minimum Bias (prescaled) Monitor 10 Hz
- Overall trigger rate 90 Hz
5ATLAS Detector at the start-up
- Stringent construction requirements and quality
controls - Equipped with redundant calibration/alignment
hardware systems - Prototypes and part of final modules extensively
tested with test beams (allows also validation of
G4 simulation) - Calibration in situ of the detector (accounts for
all effects not studied with the test beams)
includes - Test pulses
- Radiactive sources
- Cosmic runs (end 2006beg 2007)
- Beam-halo and beam-gas events during single beam
period - What can be done with the first pp data
sample ?
6Inner Detector
- Pixels 1700 modules, 100 million cells middle
barrel and disks initially staged - SCT 4000 modules, 6 million channels
- TRT 370k straws, 420k channels C wheels
initially staged - Extremely complex systems months of
commissioning even before physics data
7Position accuracy
- Individual modules located on supports to
17-100?m in r-? - Support structures (layers/disks/modules)
positioned to 20-200?m - X-ray and FSI should reduce these uncertainties
for SCT - Whole ID positioned to within ?3mm of theoretical
beamline - Possible rotation up to ?1 mrad wrt beamline, lt 1
mrad to solenoid axis - Start system debug and alignment with cosmics and
beam-gas interactions
8Trackfinding with as installed alignment
- How well will tracks be found initially ?
- Use standard iPatrec track finding
- Misalign all modules (SCT/pixel) by local
installation precision (module? structure) - Misalign all barrels/disks by RMS 100 ?m
- Reasonable estimate of installed precision
- Four examples different misalignments
- Run 10k 6 GeV muons (dc1 002207)
- Study track finding efficiency wrt perfect
alignment - 94 efficiency for local misalignments
- 40-60 efficiency for installed precision
- Tracks can still be found (with std cuts)
- Should really run with relaxed tolerances
- With 500 ?m RMS, serious degradation
- Sometimes very few tracks found
- Important to build as precisely as possible
9alignment precision estimates - 2
- Calculate r-? alignment precision from one day of
low luminosity running (here L1033 cm-2s-1 was
assumed) - Use all tracks in modules, or only overlaps
(assume 1 - 1.5-3 in DC 1 geometry) - Results given for middle pixel barrel, and 2nd
SCT barrel
The same is for L1031 cm-2s-1 in the case of
hadrons scale by sqrt(10) for muons
- Statistics to align pixels to 1-2 ?m and SCT to
2-3 ?m using 1 day of data taking - Limited by data recording rate rather than
luminosity - But systematics will also be important can make
a start with little data
10Alignment perspectives
- Statistics is not the issue plenty of data with
initial collider running - Plus useful data from cosmics and beam-gas events
- Systematics and understanding will be the key
issue - Bringing together the knowledge from survey,
tracking and ongoing FSI monitoring - Detector (thermal) stability will become
important below 100 mm (initial running
conditions will probably be unstable)
11Solenoid field mapping
- Understanding magnetic field is important for
mass scale - W mass requires overall field integral to lt
0.05, other physics processes 0.1 - Shape more difficult understand local
variations to lt 0.3. - Strategy for B-field determination
- Mapping field before ID installation using
mapping machine with Hall probe array - Intrinsic precision better than 0.1, measure
many points (few days mapping) - Fit field map using Maxwell equations
- Field shape can be described to 0.02 using
Bessel function expansion - May improve precision by constraining Hall probe
measurements - Monitoring using NMR probes during running
- 4 NMR probes installed outside barrel TRT at z0
(region with low field gradient) - Use for checking overall scale and monitoring
time stability (probe intrinsic precision of 5
ppm frequency measurement) - Final check using particles of known masses (J/?,
?, Z) - But this also brings in alignment and material
effects corrections hopefully small
12Electromagnetic calorimeter
Pb-liquid argon sampling calorimeter with
Accordion shape, covering ? lt 2.5
H ? ?? to observe signal peak on top of huge
?? background need mass resolution of 1 ?
response uniformity (i.e. total constant term of
energy resolution) ? 0.7 over ? lt 2.5
13Calorimeter calibration
- The constant term ccL ? cLR
- The local constant term, cL
- Geometry (residual Accordion modulation)
- Mechanics (absorber gap thickness)
- Calibration (with pulse test amplitude
uniformity, etc ) - The long-range constant term cLR (from
module-to-module miscalibration) - The absolute energy scale
- Use test beam measurements, cosmic ray run, pp
collisions
14- Geometry (e.g. deviation from Accordion
modulation) 0.3 - Construction phase thickness of all 1536
absorber plates (1.5m long, 0.5m wide) within
10mm ? response uniformity lt 0.3 - Pulse-Test calibration accuracy of each module
0.4 - Overall local constant term 0.5-0.6.
- Test-beam 4 (out of 32) barrel modules and 3
(out of 16) endcap modules Uniformity over units
of size Dh x Df 0.2x0.4 0.5
lt gt 2.2 mm ? ? 9 ?m
Test-beam data
15- Cosmic muons
- find dead/noisy channels cabling errors compare
with test beam data - check calibrations with lt3 months of cosmics
runs we can correct the calorimeter response
variations vs h to 0.5
Test-beam data
16- Beam-halo and beam-gas
- reconstruct muons in the end-cap
- rate 1 Hz for Etot gt 5 GeV open problem how
to trigger? - measure p0 in EM calo and check shower shapes
- Few usable electrons try to use other tracks to
check calibrations
- If no correction are applied
- cL 1.3
- cLR 1.5
The calorimeter will behave sufficiently well
already at the start-up to allow some physics.
17With pp collisions
- Use min. bias and some electron trigger (pT gt
10 GeV) - Adjust/set-up timing of calorimeters
- Measure overall energy spectrum in EM calo
- Measure EM cluster energy spectrum
- Study response uniformity of calos in ?
- Start tuning/adjusting e-identification procedure
- Check calo shower shapes for electrons
- Combine cluster with tracks
- First E/p measurements
- Study calo/ID alignment
18With pp collisions
- Try to see first Z?ee events
- Start EM inter-calibration
- Calibrate 400 region (?? x ?? 0.2 x 0.4)
- 250 e? per region needed to achieve ctot ? 0.7
- 105 Z ? ee events needed, 104 Z will be
available - Likely, c 1
- worst case scenario ctot ? 2
- cL 1.3 measured on-line
non-uniformity - of individual modules
- cLR 1.5 no calibration with Z ? ee
-
19Hadronic Calorimeters
- Cell calibration
- Reference scale (starting point) for individual
cell calibration EM scale - LAr testbeam and calibration systems about 1
accuracy on EM scale - Tilecal testbeam data, Cs calibration 3.4
precision on EM scale - Cosmic muons, beam-halo muons
- Useful in many aspects
- Largon finding dead channels, cabling errors
- Compare to muon test beam data
- Possibility to trigger with Tilecal under study
- Beam-gas hadrons
- Channel mapping
- Study their properties and how to reject them
20Minimum Bias jet events
- Monitoring detector response stability with
1-8x106 triggers to reach 1 stability - Cell-to-cell calibration
- Using phi-symmetry of MB triggers,
inter-calibrate cells with equal
dimensions/positions (2x64 cells) - Jet calibration based on weights estimated from
Monte Carlo studies ingredients - Jet fragmentation modelling electromagnetic jet
energy fraction, energy and multiplicity of
charged hadrons, etc.. - Hadronic shower models, benchmarked in comparison
with test beam data - Description of dead material in simulation
(fraction of lost energy in dead material from
few to 15 ) -
21Calibration of the L1Calo system up to the start
of collisions
By using test-pulses the calibration procedures
can checked and first Calibration constants can
be derived - close to final values Beam-gas
collisions (one-beam running) in mid-Detector
have the same timing as collision events - the
timing setup can derived and checked
Test-Pulses
Test-Beam signals at known energy
ATLAS LAr/Tile Calorimeters
L1Calo Trigger
Trigger-Inputs
Signals from beam- gas collisions
Calorimeter Read-out
L1Calo Read-out
Calibration constants
22Calibration of the L1Calo system up to the start
of collisions
- Important parameters to calibrate
- Timing of input signals and timing inside the
system - Transverse momentum / energy calibration
- Pedestal values
- Pulse shapes
- Saturation values
- Noise sigma
Many other setup parameters needed to ensure
correct data-flow in the system - to be
determined and checked before collisions start
23Calibration of the L1Calo system The first 108
triggers
Calorimeter
Expect about 104 Z0 ? ee- / 105 W ? e? /
- Clean signals with enough statistics to
- Study the energy calibration
- Verify the timing setup and event identification
- Map out the threshold curves
- Study trigger efficiency
Z0
A rapid calibration cycle is needed especially at
the beginning.
24Muon Detector Trigger
- Preliminary r-t calibration of the MDT tubes
- Calibration of the LVL1 muon trigger system
- System timing
- Coincidence roads
- Evaluation of the single muon trigger efficiency
- Measurement of the cavern background level
detected by the muon chambers - Measurement of the muon spectrum and comparison
with expectations
25Muon Detector Trigger
- MDT calibration
- Chamber Alignment
- Level-1 Trigger calibration
- Detector noise
- System synchronization
- Trigger efficiency
26MDT calibration to r/t
dn/ dt
- ultimate accuracy in to 0.4 ns
- needs 104 hits/tube
- O (109) ?-triggers
- (geometry ? included)
to
?(to) (ns)
t(ns)
- before pp data
- average to (few ns)
- cosmics during commissioning (sys shift)
- cosmics in ATLAS (105 x 100 days ? 1.5ns)
0.4 ns
104
n
27MDT calibration to r/t
t(ns)
ultimate accuracy in r/t 10?m needs 2.5104
good ?/chamber ? O (108) ?-triggers
(geometry ? included) temperatureB-field
corrections a lot of computing.
r(mm)
however, cosmics are seen only in ½ ATLAS ? use
pp data to extend the calibration to full
detector, make checks and start final (?) pass.
- before pp data
- average r/t (100?200 ?m)
- cosmics in ATLAS (105/day100days ? ok)
28Testbeam studies of chamber alignment system
reconstructions with nominal geometry show how
chambers are misaligned in terms of sagitta and
relative angles (barrel, 2nd tower)
Mean sagitta - 0.6 mm
29SAGITTA ANALYSIS
- 7 BIL2 translations along Z/Rail (16/08/2003),
magnet 800A - Analysis with fixed nominal geometry
30SAGITTA ANALYSIS
- RUNS WITH BIL2 DISPLACEMENTS, MAGNET 800 A
(BL3.4 Tm) - ANALYSIS WITH ALIGNMENT CORRECTIONS (ASAP)
Magnet 800 A
Magnet 800 A
Mean -0.59 mm
31- Alignment system still under test this year in
H8 ? will verify the performance of the overall
alignment system some preliminary result
- The analysis of cosmic ray data will help to
further check the initial performance of the
optical alignment system. - The pp data will help to check and possibly
improve the detector alignment.
Conclusion on the alignment accuracy absolute
chamber position within 100-200 mm. Ultimate
accuracy 30 mm
32RPC/TGC counting rate
- Intrinsic RPC noise can be measured with LHC off
- typical measured noise is about 3 Hz/cm2.
- Cavern background (neutron, photons,...)
- Incoherent physics background
- RPC counting rate 0.02 Hz/cm2 physics rate
(L1x1031 cm-2 s-1) - too low wrt detector noise check that this noise
doesnt depend on LHC on/off - Coherent physics background count the number
of RPC events with the same strip (or -1) fired
in RPC doublets. - Penetrating background particles specific tools
as pattern recognition programs will be used.
33Timing of Coincidence Matrices
- Pulse test and the run with cosmics will allow a
preliminary evaluation of the time alignment of
the RPC signal to be put in space and time
coincidence - With pp collisions can be done with single
muons - Assume 106 muons pTgt10 GeV in the barrel
system - gt With this statistic we expect about 600
events per Coincidence Matrix (in total 2x1700
CMs), likely sufficient for a first time aligment
of each of them. - The trigger system does not need a very accurate
time alignment when running at low luminosity. - Coincidence Windows
- Not a problem. Rely on simulation knowing the
collision point position along z with about 1cm
accuracy.
34Trigger efficiency
- Can be done with the study of di-muon systems
(J/Y, Z,...) triggered with an inclusive muon
selection analyze the muon with no trigger as a
function of h, f and pT.
35The offline data flow
Rate, Hz Size (KB) Total (TB), for 108 triggers
Raw Data 100 1600 160
ESD (Event Summary Data) 500 50
AOD (Analysis Object Data) 100 10
TAG (Event DataBase) 10 1
36The Computing System
Pb/sec
Event Builder
10 GB/sec
Event Filter1.5MSI2k
- Some data for calibration and monitoring to
institutes - Calibrations flow back
450 MB/sec
Tier 0
T0 5MSI2k
HPSS
300MB/s/T1 /expt
Tier 1
UK Regional Centre (RAL)
US Regional Centre
French Regional Centre
Italian Regional Centre
HPSS
?622Mb/s
Tier 2
Tier2 Centre 200kSI2k
Tier2 Centre 200kSI2k
Tier2 Centre 200kSI2k
?622Mb/s
Each Tier 2 has 25 physicists working on one or
more channels Each Tier 2 should have the full
AOD, TAG relevant Physics Group summary
data Tier 2 do bulk of simulation
Lancaster 0.25TIPS
Sheffield
Manchester
Liverpool
Physics data cache
100 - 1000 MB/s
Desktop
Workstations
37Operation of Tier-0
- The Tier-0 facility at CERN will have to
- Hold a copy of all raw data to tape
- Copy all raw data to T-1s for later reprocessing
- Keep calibration data on disk
- Store master conditions database here
- Run first-pass calibration/alignment and
reconstruction - ESD copy retained on tape
- Distribute ESDs to N external Tier-1s
- (2/N to each one of N Tier-1s assume N10 for
now) - Express lines
- Need for calibration line to reduce latency
- Might also serve as physics monitoring line
- Physics hotline more contentious
- Streaming of ESD and AOD
- Always envisaged to stream AOD into 10 exclusive
streams, collections being built from unions of
streams - Streaming of ESD is required to reduce file access
38Operation of Tier-1s and Tier-2s
- Tiers defined by capacity, role and level of
service - No assumption of single site (esp. T2), but must
present as a single entity in human/response
terms - We envisage likely 10 Tier-1s for ATLAS. Each
one will - Keep on disk 1/5 of the ESDs and a full AODs
and TAGs - Keep on tape 1/10 of Raw Data, reprocess and
retain ESD produced - Keep on disk 1/5 of currently simulated ESDs and
on tape 1/10 of previous versions - Provide facilities for physics group controlled
ESD analysis - Calibrations
- Support role for defined set of Tier-2s
- We estimate 4 Tier-2s (various sizes, slower
response) for each Tier-1. Each one will - Keep on disk a full copy of TAG and roughly one
full AOD copy per four T2s - Keep on disk a small selected sample of ESDs
- Provide facilities (CPU and disk space) for user
analysis and user simulation (25 users/Tier-2) - Run central simulation
39Which physics with this data sample?
Process Events/s Events in 10 pb-1 Total statistics collected at previous experiments by 2007
W ? en 0.07 7x104 104 LEP / 106 Tevatron
Z ? ee 0.011 104 106 LEP
tt?WbWb?mn 0.008 8x103 104 Tevatron
H, m130 GeV 100
Gluino-gluino , m1 TeV 1-10
40Physics
- Study properties of minimum bias events and check
with existing data at lower energies. - Study the muon spectrum check inclusive LVL1
muon trigger rates - Check J/Psi, Ypsilon, W and Z to leptons
preliminary estimates of production cross
sections of W and Z to leptons, with an expected
error of 15-20 (10 lumi10-15 efficiency) - Inclusive jet cross section measurement
- Top signal visible?
41Conclusions
- The first pp events have many invaluable tasks
- Subdetectors
- Calibrations, almost always not final, but
improved wrt test-beam and cosmics run - From subdetectors to ATLAS
- Trigger commissioning efficiency
- Subdetector integration and event building
- Offline commissioning
- From detector to results
- Some very preliminary physics analysis W, Z
cross-sections, jet spectrum,