html5-img
1 / 7

B-Trigger Working Group Status Report

John Baines. B-Trigger Working Group Status Report. Progress on Algorithms Updates to benchmarking & resource estimates Using Calorimeter information to guide LVL2 B-triggers Summary. This talk: http://hepunx.rl.ac.uk/atlasuk/simulation/level2/meetings/pesa250402/PESA.ppt

shirin
Download Presentation

B-Trigger Working Group Status Report

An Image/Link below is provided (as is) to download presentation 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. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. John Baines B-Trigger Working GroupStatus Report • Progress on Algorithms • Updates to benchmarking & resource estimates • Using Calorimeter information to guide LVL2 B-triggers • Summary This talk: http://hepunx.rl.ac.uk/atlasuk/simulation/level2/meetings/pesa250402/PESA.ppt http://hepunx.rl.ac.uk/atlasuk/simulation/level2/meetings/pesa250402/PESA.ps http://hepunx.rl.ac.uk/atlasuk/simulation/level2/meetings/pesa25040202/PESA.pdf

  2. Progress on Algorithms • IDscan (SCT & Pixels):(Nikos Konstantinidis, Hans Drevermann) • Confirmed that IDscan algorithm:- see Nikos' talk. • Reproduces TP B-trigger performance (10% - 15% better effic. for Ds(fp) trigger) • Gives same performance for reduced layout - 2 instead of 3 barrel Pixel layers • Execution time: • 6 ms scaled to a 4GHz 160 SI95 machine, • 11 ms including Si data preparation • Pixel-scan expected to achieve similar execution time (currently longer) • PixTrig:(Genova) • Currently working to reproduce TP results in Athena framework • Will then modify algorithm to include SCT info. • Plan to look at linear fit in transform space, as used by CDF, as alternative to SiKalman to see if it gives better performance.

  3. Benchmarking & Resource Estimates • Updated resource estimates reported at last ATLAS week • Benchmarking since redone to measure mean, m50 and m95 times for • - full-scan in B->mX events • http://hepunx.rl.ac.uk/atlasuk/simulation/level2/Bphys/bmark.html • - EM RoI Dh x Dp = 0.2 x 0.2 in dijet events with/w.o. pile-up • http://hepunx.rl.ac.uk/atlasuk/simulation/level2/eid/bmark.html • Later used to estimate muon confirmation times. • Parameters and estimates documented in ATLAS note : ATL-COM-DAQ-2002-nnn • Jos has re-run paper model using these parameters as input: • http://www.nikhef.nl/pub/experiments/atlas/daq/paper-para-overview1.pdf

  4. Benchmarking & Resource Estimates Preliminary paper model results (April 2002): Final Farm size Difference ~25 c.p.u Min. size incl. B-physics Min. Farm size needed at startup • Note: • In order to save cpu resources, not using TRT info at LVL2: • => No J/psi(e,e) trigger • Resources required at EF small compared with LVL2 (~few cpu) assuming: • xKalman starting with pixels and guided by Level-2 RoI • (see presentation at previous meeting) • neglecting resources needed for muon reconstruction at EF • Current offline muon algorithms too costly in cpu : • => Only includes muons at LVL2 reconstructed by mFast

  5. Using Calorimeter RoI to guide B-physics Triggers • Preliminary studies of using low ET RoI to define regions to search ID at LVL2(Alan Watson) • Uses fast simulation : • ATLFAST + parameterised calorimeter simulation • Initial results encouraging,see Alan Watson’s talk:http://www.ep.ph.bham.ac.uk/user/atw/bmeet/B-roi-feb02.ppt • EM RoI ET>2 :for J/y(ee) and muon-electron triggers • Mean Multiplicity = 1.1 (B->mX , m pT > 6 GeV) • Effic. to tag both e in J/y(e,e) : 80% (e pT >3 GeV) • Jet RoI (0.8 x 0.8 cluster) ET>5 :for B(pp) and Ds(fp) • Mean Multiplicity = 1.7 (B->mX, m pT > 6 GeV) • Efficiency • B p p • pTp> 4 GeV • RoI ET> 5 GeV • B Dsf • pT Ds, f> 1 GeV • RoI ET> 5 GeV • LVL2 reconstruction inside RoI • potential to save ~factor 10 in execution time c.f. full-scan • but with lower efficiency Actual efficiencies and cpu savings depend on thresholds & multiplicities => need full simulation

  6. Future Work • Calorimeter RoI guidance for B-triggers: Need to confirm results with full simulation. Requires : • Simulation of: • Level-1 EM/Tau trigger algorithm - done (Ed Moyse) • Level-1 Jet trigger algorithm - underway (Ed Moyse) • Level-1 electronics (noise) - who? • Generation of simulated data : • single particles for calibration • B->mX events • signal channels : B->pp, B->J/y(ee), B->Ds(f(KK)p) • min. bias • Addition of pile-up (including pulse history in ECAL). • Calibration of ECAL energies • Level-1 studies - Optimisation of thresholds • B-trigger performance studies • Verification / optimisation of algorithms with latest layout :Requires: • New datasets (including pile-up) • Algorithms able to read new datasets (LVL2 & EF) • B-trigger studies including muon trigger. Requires: • New datasets including muon simulation • Muon algorithms available in new framework (LVL2 & EF)

  7. Summary • Encouraging progress with IDscan algorithm • Converging on a baseline B-trigger strategy minimising resource requirements • Encouraging preliminary results for RoI guided B-trigger at Level-2 • => potential to futher reduce resource requirements, but with reduced efficiency • A lot of work on hold awaiting new data and software : • RoI guided B-physics • inclusion of muon trigger • verification/optimisation of algorithms for latest layout • Timescales tight even with a 6 month delay in the TDR

More Related