Trigger slices - status and plans

Document Sample
Trigger slices - status and plans Powered By Docstoc
					Trigger slices - status and plans
              Artur Kalinowski
    (Physics Department, University of Regina)
 on behalf of the trigger community

                22 May 2008

  • Introduction
  • Slices overview
  • Conclusions

Trigger slices - status & plans      1      TDAQ Week, 22.05.08

 The Main Trigger task is reduction of the initial event 40MHz BC rate to ∼200Hz
 recording rate
 Also the Trigger:

  • selects events for physics studies, alignment, calibration and efficiency measurements
  • assigns events to streams for data recording :
      – physics streams to facilitate data access for analysis

      – express stream for rapid feedback on alignment, calibration and efficiency

 Those vital functions must be achieved fast & within the resources of the LVL2 and EF
 farms, with robust online (real-time) monitoring ensuring good Data Quality and providing
 tools for fast issues tracking
 Trigger work is organised in vertical slices : groups of selection chains identifying an object used for
            electron, photon, muon, jet, tau, missing ET , B-jet, B-physics + Cosmic Ray
 triggering i.e.
 Triggers and Level-1

Trigger slices - status & plans                         2                                        TDAQ Week, 22.05.08
                                  Event selection chain

Trigger slices - status & plans             3             TDAQ Week, 22.05.08
                                  Readiness for day “0”

  The data taking period starts quite soon. It is crucial now to:

  • have robust selections for early data: including different luminosities and detector and
     beam conditions.

  • have the cuts optimised w.r.t. trigger rate, efficiency and execution time
  • be ready with techniques to measure trigger efficiency from data
  • understand biases introduced via trigger cuts (as far as possible using data).

                   We are switching working mods from ”prepare” to ”ready”.

Trigger slices - status & plans                4                                TDAQ Week, 22.05.08
                                  Commissioning the slices

  Several ways of Commissioning the Trigger for first physics:

  • Offline Running:
     – run LVL1 trigger emulation + real High Level Trigger algorithms as part of
        Reconstruction in Athena
     – develop and optimise selection chains and menus.
     – measure trigger rates using bulk samples of min bias and di-jet events.
     – measure trigger efficiencies for signal samples
     – perform Trigger aware physics analysis

  • Tests on the pre-series and dedicated Trigger/DAQ machines
     – special versions of athena emulate the environment for algorithms running online
        in LVL2 (athenaMT) and the Event Filter (athenaPT)

Trigger slices - status & plans              5                                TDAQ Week, 22.05.08
                                  Commissioning the slices

  • Cosmic runs:
     – exercises many components of the final system incl. readout chain and DAQ
     – test integration of trigger with different detectors, separately and combined
     – exercises software with real data (real noise, real errors etc.)
     – using trigger algorithms either specific to, or adapted for, cosmic running

  • Technical runs:
     – simulated data downloaded to the Read Out Buffers
     – tests the full algorithms and menus running on the online farms.

Trigger slices - status & plans                6                                TDAQ Week, 22.05.08
               Events & Progress since ATLAS Week in February

  • Milestone week M7 is ongoing - use of the full trigger chain (LVL1 + HLT) online with
     real cosmic data

  • Trigger Robustness Workshop in March - studies on assessing trigger response to
     problems such as data corruption, detector issues, beam-related issues, etc. and
     developing a strategy to deal with these problems.

  • Data Quality Workshop in May - dedicated session devoted to HLT and combined
     performance DQ (see talk by Martin Zur Nedden later this session)
  • April Technical Run - full 1031 , 1032 menus used online with simulated tt and Mixed
     data downloaded to Read Out Buffers (RoBs), new tdaq release 1.9.0 used online

  • ongoing algorithm work increasing focus on final preparations for data
  • significant progress on trigger menus (see talks by Andrew Brandt and William Bell
     later this session)

Trigger slices - status & plans              7                               TDAQ Week, 22.05.08
                Electron and Photon slices: algo improvements I

  • Better calibration/cluster corrections available at L2:
     – cluster corrections and gap correction as done in offline
     – results in slight increase in efficiency by ∼0.5% in total

  • Improvements for non-isolated electrons from J/Ψ → ee
     – problem: both electrons end up in L1 RoI and twice in the event the higher
        energetic one is reconstructed
     – problem solved by restricting the L2 window for the cluster search
        Currently being validated and window size being optimised
     – alternatively sliding window algorithm has been implemented at L2
        1st prototype available, needs testing now

Trigger slices - status & plans                8                             TDAQ Week, 22.05.08
               Electron and Photon slices: algo improvements II

  • New backtracking option at EF:
     – allows conversion reconstruction at EF for photons
     – available as back-up trigger for electrons at start-up

  • New L2 TRT standalone algorithm (TRTSegFinder) available
     – first performance studies under way
     – this is another back-up trigger for electrons at start-up

Trigger slices - status & plans                9                   TDAQ Week, 22.05.08
                             Back Tracking in the Event Filter

               Method                Efficiency
                                                 • Forward Tracking (FT) reconstructs tracks in the silicon
          Forward Tracking             99.9        and then extends to the TRT,
        Back Tracking default          93.2      • Back Tracking (BT) starts in the TRT and eventually
   Back Tracking with relaxed cuts     94.8        extends the track to the silicon:

   Back Tracking with Si extension     91.1         – it is meant to deal with photon conversions,
                                                    – it is foreseen as a back-up solution if the silicon fails.

                                                 • BT reco is available at the EF since 14.0.0 in the fol-
                                                   lowing slices:
                                                    – photons(conversions);
                                                    – electrons (back-up),

                                                 • first results for 200 GeV single muons show we can rely
                                                   on it;

                                                 • timing measurement is well within the budget.

Trigger slices - status & plans                   10                                          TDAQ Week, 22.05.08
             Electron slice: performance highlights for electrons

    Level       e22i w/o PU        e22i + PU    • Effect of pileup on e-selection
      L1       94.11±0.06%        93.10±0.21%      – Drop of      1-2% in eff.      w.r.t.      offline seen
   L2 Calo     91.54±0.07%        90.39±0.24%         in Z → ee sample in case      L = 1033 cm−2 s−1
                                                      pileup present
    L2 ID      87.50±0.09%        85.78±0.29%
                                                   – Shows trigger selections a bit more sensitive than
      EF       86.85±0.09%        85.00±0.30%
                                                   – Studies ongoing for eff. overall

                                                • Effect of misalignment
                                                   – Drop of overall eff. of ∼10% on single electrons with
                                                      ET = 25 GeV observed
                                                   – Drop only affects EF tracking and track-cluster
                                                      matching part

                                                • Emphasis on studies now on readiness for data taking
                                                   – 1-day e/γ workshop (trigger+offline) on June 10

Trigger slices - status & plans                  11                                          TDAQ Week, 22.05.08
                                  Jet slice

                                          • forward jet reconstruction at L2 and EF imple-

                                          • initial L2 forward jet calibration available for
                                            physics studies.

                                          • L2 calibration constants can now be stored/re-
                                            trieved as pool files from COOL db

                                          • ability to use topo clusters in EF implemented
                                          • ongoing work to define DQ monitoring his-
                                            tograms for shifters, provide documentation, etc.

                                          • optimization of jet trigger menu for all trigger lev-
                                            els see talk by Andrew Brandt later in this session

Trigger slices - status & plans      12                                         TDAQ Week, 22.05.08
                                  Jet slice

                                    Investigation of the possibility to calibrate L1 Jets at L2:

                                      • would provide very fast hypothesis testing
                                      • could allow the implementation of more complex L2
                                          jet algorithm.

                                      • significant improvement of trigger response as com-
                                          pared to L1 can be achieved.

                                      • for higher energy threshold, trigger response of L1
                                          calibrated jets is comparable to standard L2 jets.

                                          For further details, please see the talk by Cibran
                                      Santamarina at Trigger Open Meeting from April, 23
                                    Other ongoing studies:

                                      • performance studies of different calibration/cluster-
                                          ing schemes in EF, possibility of full event recon-
                                          struction in EF, jet trigger rates vs threshold

Trigger slices - status & plans      13                                          TDAQ Week, 22.05.08
                                          Missing ET slice
  The MET trigger slice passed through an important transition in the past few months, for what concerns
 the trigger EDM and the EF Fex. Release 14.2.0 is the first release in which all new functionalities will be

  • New TrigMissingET design:
      – more complete information

      – can include details of each calorimeter sampling and of the muon correction

  • New EF Feature EXtraction (FEX) algorithm:
      – modular structure: sequence of tools

      – FEB tool: fast but poor resolution (not used by default)

      – cell tool: default MET measurement

      – muon tool: correction to calorimeter based measurement

      – event flag tool, for trigger robustness [dummy so far]

      – calibration tool: leaves components uncorrected

Trigger slices - status & plans                       14                                      TDAQ Week, 22.05.08
                                           Missing ET slice
 Given the new structure, an effort is ongoing to adjust monitoring histograms:

  • Online shifters:
      – the total number of histograms still has to be reduced

      – the reference histograms have to be provided

      – histogram explanations, check-lists, etc. have to be provided

  • Data quality:
      – additional histograms that show detailed info, are being introduced

      – focus is on aspects that will be important for commissioning

  • Commissioning and early data taking:
      – trigger robustness addressed by different points of view:
        ∗   channel failures (hardware side)
        ∗   search for fake sources of MET (detector side, physics oriented)
        ∗   the first trigger menus has to be selected
        ∗   the calibration at EF is being refined

Trigger slices - status & plans                         15                        TDAQ Week, 22.05.08
                                  Muon slice: L2
                                    • MuFast: improvement in pattern recognition and tracking on
                                      the full muon spectrometer allows for coherent monitoring of
                                      barrel and endcaps reconstruction, code allows access to
                                      CSC data, BEE chambers data, MDT data of the neighbour
                                      sector (for alignment into the calibration stream)

                                    • finalized implementation of the optimized cuts for π/K re-
                                      jection in MuCombHypo for mu4 and mu6 thresholds

                                    • cut optimization for track and calorimeter isolation (MuIso)
                                      finalized in |η| < 2.4, optimization for low pT muon thresh-
                                      old (e.g. mu15, for tau selection) done for ID and calorime-
                                                  → flat efficiency (∼95%) in eta and rejections
                                      ter isolation
                                      in 87-91% in the pT range

                                    • upgrades in the format for Tile ROD subfragment with muon
                                      tagging information in the low pT muon tagging algorithm in
                                      Tile (ROD based approach TileRODMu)

Trigger slices - status & plans          16                                       TDAQ Week, 22.05.08
                                  Muon slice II
                                    • TrigMuonEF: a new package committed for the new Muon
                                      EF code, hosting the EF wrappers for the offline tools per-
                                      forming the Muon Spectrometer and combined reconstruc-
                                      tion (14.2.0), test/debug features were set-up, in order to val-
                                      idate results, configuration to run it on cosmics is on-going

                                    • old TrigMoore in the usual shape in rel. 14
                                    • Level 2 & EF: major work to migrate L2 (BS→RDO) and
                                      EF (RDO→PRD) converters to the algtools/new IDC mech-
                                      anism, not yet completed

                                    • menus: optimization of menus for 1032 in release 14 to keep
                                      the final rates under control

                                    • DQM: new twiki page describing online histogram-
                                      s/DQMF     checks     (for   the   moment      only    for    EF)
                                      MuonSliceDataQualityMonitoring        additional      histograms
                                      in MuFast monitoring to perform DQA in Endcaps, work on
                                      offline monitoring just started
Trigger slices - status & plans          17                                         TDAQ Week, 22.05.08
                                   Tau slice
                                  • possibility to use topo clusters at EF
                                  • possibility to run Tau1p3p at EF in addition to TrigTauRec
                                  • many new variables are being investigated:
                                     – hadronic radius of cluster
                                     – variables used in tau1p3p

                                  • trigger simulation for ATLFAST is coming:
                                     – provide pt turn-on efficiency curves in bins of eta for ATL-
                                       FAST I
                                     – run HLT using cells generated by fast MC in ATLFAST II

                                  • Data Quality Monitoring:
                                     – first complete set of histograms with automatic checks
                                     – first approximation of shifter instructions ready on TWiki:

Trigger slices - status & plans          18                                       TDAQ Week, 22.05.08
                                   Tau slice II
                                   • Measurement of trigger efficiency from data –
                                     crucial ingredient in successful operation of trigger:
                                      – first step: estimate of fake rates using di-jets events
                                      – second : reconstruction of Z    → τ τ → lhX channel
                                      – third: reconstruction of taus in tt events

                                   • Focus on Validation and commissioning
                                      – a lot of tools are being developed for validation and on-
                                        line integration of software: using HLT-cut simulation on
                                        ntuple level to validate HLT code - very powerful to spot
                                      – moving tools from performance studies (efficiency
                                        curves, comparison with offline objects) to automatic
                                        nightly checks

                                  Many more progress reported at last week Tau Workshop.

Trigger slices - status & plans           19                                        TDAQ Week, 22.05.08
                                           B physics slice

 Di-muon triggers:

  • low luminosity (< 2 · 1033 cm−2 s−1 )
      – require only 1 L1 muon RoI

      – search for second muon in large RoI (1.5x1.5) around the first muon (or FullScan)

  • high luminosity (> 2 · 1033 cm−2 s−1 ), require 2 L1 muon RoI (pT > 4 − 6 GeV )
      –   B → J/Ψ(µµ)X ;
      – rare decays with di-muon, e.g.   B → µµ, B → K0 µµ

 Hadronic/EM B decays:

  • low luminosity (< 2 · 1033 cm−2 s−1 ):
      – require 1 L1 muon RoI

      – search in large RoI (1.5x1.5) around JET or EM RoI (or FullScan).

      – reconstruct B decay products (Bs   → Ds (φπ)π , B → J/Ψ(ee)µX , B → γµX )

Trigger slices - status & plans                     20                                     TDAQ Week, 22.05.08
                                  B physics slice

                                   Di-muon signatures from single L1 muon:

                                    • already had TrigDiMuon at L2, now extended to EF using
                                      TrigMoore or TrigMuGirl

                                    • as of last week TrigMuGirl works in 14.0.0, new signatures
                                      to use TrigMuGirl currently being implemented

                                   Other improvements:

                                    • added EF algorithm for di-muon signatures starting from 2
                                      L1 muons (J/Ψ → µµ etc.)

                                    • added L2 algorithms for B → µµX (X=K + at the mo-
                                      ment, working on K ∗ )

                                    • new histograms added for online monitoring
                                    • working on improvements to Ds (φπ)π algorithms at EF to
                                      reduce rate (reconstruct B, use vertexing)

Trigger slices - status & plans            21                                      TDAQ Week, 22.05.08
                                                Bjet slice

 Different approaches studied for rel 14 to ensure trigger flexibility and robustness wrt different working

  • likelihood track-based taggers: same FEX at L2/EF, well understood, best performance using 2D
     combination of significances of longitudinal and transverse impact parameters;

  • secondary vertex related taggers: different FEX at L2/EF: L2 based on d0 − ϕ technique as
     developed at CDF while EF based on offline (if secondary vertex at EF is demonstrated to be fast
     enough, it will be port at L2).

  • χ2 probability tagger: jet probability to originate from the primary vertex using the significance of the
     transverse impact parameter.

  • µ + jet trigger: needed to measure HLT and offline b-tagging efficiency

Trigger slices - status & plans                       22                                       TDAQ Week, 22.05.08
                                              Bjet slice

   • IP3D tagger at L2 and EF using signed tracks;        • IP χ2 probability performance at L2;
   • EF selection starts from L2 working point.           • request of at least two reconstructed tracks per

Trigger slices - status & plans                      23                                      TDAQ Week, 22.05.08
                                  Minimum bias slice
                                           Three Min Bias Trigger Scintillators (MBTS) Triggers

                                               • MBTS 1 - one or more MBTS above threshold
                                               • MBTS 2 - two or more MBTS above threshold
                                               • MBTS 1 1 - at least one MBTS above threshold
                                                 either side

                                               • MBTS Lvl2 algorithm added in 14.2.0:
                                                 This will allow time cut on MBTS pulse, which
                                                 could be used to remove beam backgrounds or
                                                 select them.

                                               • the Space Point trigger in 14.2.0 includes param-
                                                 eters to allow different thresholds for different sec-
                                                 tions of the SCT

                                               • performance studies of MBTS Lvl2 and Space
                                                 Point trigger in 14.2.0 are needed next.

Trigger slices - status & plans           24                                          TDAQ Week, 22.05.08
                                  Minimum bias slice

Trigger slices - status & plans           25           TDAQ Week, 22.05.08
                                  Level 1 BCM trigger

                                           BCM (Beam Conditions Monitor) will be used as an in-
                                           put to Level-1. It is a better (cleaner) input than using
                                           a random trigger at Lvl1 for the Minimum Bias trigger.

                                               • Four BCM trigger bits available at L1:
                                                  – L1 BCM AtoC : Particle went from side A to
                                                    side C.
                                                  – L1 BCM CtoA : Particle went from side C to
                                                    side A.
                                                  – L1 BCM Wide : Wide time cut for commission-
                                                    ing if the IP is not at the center.
                                                  – L1 BCM Ideal : Calculate ideal collision from
                                                    internal BCM bits

                                               • BCM LVL1 trigger simulation introduced in 14.2.0

Trigger slices - status & plans           26                                          TDAQ Week, 22.05.08
                                  Cosmic slice

                                        Cosmic slice is the most extensively used slice so far.

                                            • large overlap with collision physics algorithms
                                            • excellent place to test and evaluate the perfor-
                                              mance of various trigger/offline algorithms:
                                               – Estimation of the L2 online tracking perfor-
                                                 mance with real data: The LVL2 tracking ef-
                                                 ficiency is based on matching with TRT/SCT
                                                 Offline tracks

                                        Pending issues:

                                            • migration to realistic conditions database (so far
                                              the offline DB was used instead)

                                            • definition of a stable cosmic ray menu

Trigger slices - status & plans        27                                        TDAQ Week, 22.05.08

 Significant progress with Commissioning Trigger Slices since February ATLAS Week:

  • M7 Cosmic run with Trigger and DAQ - continuing with system validation
  • Technical Run on simulated full ATLAS data downloaded to RoBs, with new tdaq
    release, and with full 1031 , 1032 menus

  • last missing trigger functionality in place : MBTS and Forward Jets
  • workshops addressing runtime functionality: Trigger Robustness and Data Quality
  • the ATHENA release for first data taking was closed yesterday - crucial
     functionality is in

Trigger slices - status & plans             28                               TDAQ Week, 22.05.08

Trigger slices - status & plans     29     TDAQ Week, 22.05.08
                                  Muon slice

Trigger slices - status & plans       30       TDAQ Week, 22.05.08

Trigger slices - status & plans   31    TDAQ Week, 22.05.08
                                  Trigger Overview

Trigger slices - status & plans          32          TDAQ Week, 22.05.08