LVL1 Selection. B Physics processor. TrigDiMuon. Tag. Probe. Sergio Grancagnolo (University of Salento & INFN Lecce - Italy) on behalf of the ATLAS Muon Trigger group. ACAT 2008 Advanced Computing and Analysis Techniques in Physics Research.
Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.
B Physics processor
Sergio Grancagnolo (University of Salento & INFN Lecce - Italy)
on behalf of the ATLAS Muon Trigger group
Advanced Computing and Analysis Techniques in Physics Research
The ATLAS Muon Trigger vertical slice at LHC startup
The ATLAS trigger system is designed to keep high efficiency for interesting events, while rejecting standard model physics low pT events, with a suppression factor of the order of 107, reaching the ~200 Hz data storage capability of the DAQ system.
Level 1 Trigger
Level 2 Trigger
Core algorithm is muFast, that confirms/rejects LVL1 result and refine muon pT evaluation, using MDT precision measurements.
LVL1 selects active detector regions (Region of Interest, RoI), in each event. Coincidence windows are defined on the allowed geometrical roads with their center corresponding to the infinite momentum track. For the single beam run in september a special trigger configuration was set, allowing for tracks not pointing to the interaction region as for normal running.
High pT muons are important for many known processes, that can be used for monitoring and calibration (Zmm) and for several new phenomena predicted at the LHC energy (Higgs, SUSY), therefore the muon trigger system is of primary importance.
The Muon Spectrometer (MS) is the detector dedicated to the identification of muons. It consists of RPC and TGC trigger chambers and MDT and CSC precision chambers.
The Muon Vertical Slice consists of three main trigger steps, one hardware, level 1 (LVL1) and two software, level 2 (LVL2) and event filter (EF). Last two compose the High Level Trigger (HLT).
Total latency time
e vs. pT
Minimum bias trigger
muFast efficiency with respect to LVL1 for ideal and noisy detector conditions.
To refine the muFast pT, muCombuses results from LVL2 Inner Detector algorithms, allowing to sharpen the threshold at low pT.
Muon EF ID algs
LVL1 trigger uses RPC (up) in the barrel (|h|<1) and TGC (down) in the endcaps (1<|h|<2.4)
~ 200 Hz
Decision on each event is based on reduced-granularity detector data for interesting region at LVL1, full-granularity and precision, but only for same LVL1 regions, at LVL2, and full event data, as in offline, at EF.
e vs. pT
Low pT rates: muons from pions and kaons
muComb efficiency with respect to LVL1 for ideal and noisy detector conditions.
Using information from calorimeter, muTilelooks for energy deposit compatible with energy loss from muons.
One of the main sources of muons at low pT are in flight decays of light mesons.
A track from such decays appears with a kink, and the c2 of the fit is worse. All possible kinematical parameters and statistical techniques must be used in order to reject such tracks.
The observed trigger events are mainly cosmics.
LVL1 Efficiency is 83% Low-pT, 79% High-pT. Rates (lumi): low pT ~11 kHz (1033cm-2s-1), high pT ~2 kHz (1034cm-2s-1)
Differential cross-section for
production of muons from pions and kaons
Reconstruction of cosmic events with muTile.
Efficiency from Zmm sample
Different kinds of physic events need to share available bandwidth, that is limited. Flexible trigger menus allow to avoid saturation from few processes, and guarantee the possibility of organizing the analysis depending from luminosity conditions.
When the rates are too high, prescale factors (PS) can be applied to low pT thresholds at LVL1. The HLT capabilities are studied using the pass through (PT) mode: events are flagged without rejection.
Express streams, not prescaled, at fixed thresholds, are used for calibration pourposes. Below a possible startup menu @1031 luminosity.
Method to evaluate performance, using one m reconstructed in both Inner Detector (ID) and Muon Spectrometer (MS) as a tag, and the other requiring only the ID as a probe.
This method can be applied to extract MS trigger efficiency directly from data.
The Muon Event Filter consists of four algorithms: SegmentFinder, TrackBuilder, Extrapolator and Combiner.
They are wrappers for the offline reconstruction tools. EF processing normally starts from muFast result but, for debug purpose, can use LVL1 RoI directly. Segments are made first, using MDT precision hits. Tracks are made from segments, adding information from other muon detectors.
pT > 20 GeV
|η| < 2.4
single beam data
pT > TPcut
|Mμμ - MZ| < 10 GeV
Δφ > 0.3
Beam coming from z<0
TPcut = 20 GeV
trigger measurement over threshold
TPcut = 1 GeV
for integrated trigger efficiency
Number of segments reconstructed by the SegmentFinder
Tracks reconstructed by the TrackBuilder
Allows threshold trigger efficiency measurements with good precision when compared to MC generator.
The extrapolation to the interaction point uses a parametrization of the energy loss in the calorimeter, for faster computation. Information from the inner detector EF algorithms is then added to make a combined track. The parameters of the tracks are finally obtained after refitting the hits actually used.
A trigger chain example
= Trigger Element
Trigger software works with objects called Trigger Elements (TE). Feature Extraction Algorithms (FEX) are activated by input TE produced by previous trigger levels. FEX are able to access the detector data and compute physical quantities, Features, that are then attached to the output TE. Selection is done in Hypothesis Algorithms, that can validate or reject TE that do not satisfy trigger requirements.
= Feature Extraction Algorithm
= Hypothesis Algorithm
pT, h, f
In-flight decays of pions and kaons are the main source of LVL1 trigger rate at low pT. One goal of the muon HLT is to reject such fake muons while having high selection efficiency on prompt muons. Hypothetical chains like the one on the left can be used for this purpose.
Extrapolator and Combiner pT resolution for ideal and noisy detector conditions.
Vtx close to IP
m production point