Trigger

switch to my:
L2 Tau page A level 2 global tau trigger.

skip to:

back to: Bryan's homepage

Introduction

The Level 1, 2, and 3 triggers are all configurable. What type and combination of sub-triggers should be used at each of the three trigger levels? To zeroth order, just use the same types of triggers that worked at Run I; however, there are many new subdetectors, the trigger processors are themselves new, and don't forget the magnetic field. In order to intelligently decide which triggers to use, one must have a trigger emulator so one can play with the trigger and understand the physics impact of each of the proposed sub-triggers. In addition, it is also nice to have a trigger simulator to make sure your emulator makes sense.

In 1999, a trigger simulation group was formed. Dave Toback was the coordinator. He quit because the "Report of the Trigger Emulation Design Group", D0Note 3726, criticized his method. Essentially, what Toback did was to collect together the "chunks" of data various people had worked on simulating (tsim_xxx). As was pointed out, one can not easily convert the Toback ntuple into an emulation.

My first task will be to understand how we can trigger at level 2 on Taus. I will use the (soon to exist) trigger emulator to do my study. It may very well be that this work has, to some extent, already been done, but it is in the form of little pieces here and there.

Here are a few important trigger related links:

A few more trigger links:

back to the top

Tau Triggers

The following information was extracted from the Run II Trigger List.

There are 10 overall categories of trigger, of which 4 have explicit tau triggeres:

The 10 version 2.51 trigger list triggers which mention the word "tau" are:
Category Title L1 Terms L2 Terms Scaling
lepton+jets tau-jet CJT(2,3) + CJT(1,7) + P1C_jt L2TA(1,10,T) + L2JT(1,15) prescaled
missing Et tau jetc ttkCJT(1,3) + CJR(1,C) + CME(20) + P1C_jt L2TA(1,15,T) + L2ME(20,s) prescaled
missing Et tau jetc qt CJT(1,3) + CJR(1,C) + CME(20) + CEM(1,LO) + CER(1,LO,C) + TTA(1) + P1C_sh L2TA(1,15,T) + L2ME(20,s) prescaled
missing Et tau met CJT(1,3) + CJR(1,C) + CME(20) L2ME(20,s) + L2JT(1,15,C) monitor
mixed lepton ele tau a CEM(1,LO) + TEL(5,h) + TTK(2,5) + CJT(2,3) + CJR(1,C) + P1C_el + P1C_jt L2TA(1,5,T) + L2EM(1,5,fpi) unprescaled
mixed lepton ditau 5c CEM(1,LO) + TIS(5) + TTK(2,5) + CJT(2,3) + CJR(2,C) + P1C_jt + P1A_shLO L2TA(1,5,T) + L2JT(1,10) unprescaled
mixed lepton gam tau CEM(1,5) + TIS(5) + CJT(2,3) + CJR(1,C) + P1C_jt + P1A_shHI L2TA(1,5,T) + L2EM(1,7,s,fpi) unprescaled
mixed lepton mu tau MUO(1,4,X,T) + CEM(1,LO) + CER(1,LO,C) + TIS(1,5) + P1C_tr L2TA(1,5,T) + L2MU(1,4,X,T) unprescaled
track-minibiasll tel TEL(10,l) + TLO STT prescaled
track-minibiasttk2 tht TTK(2,5) + TIS(1,5) + THT(lo) L2TA(1,10,T) + STT unprescaled

Note, 8 of the 10 listed use the level 2 tau tool, which at my current level of understanding, does not yet exist.

It is know yet known to what extend the tau trigger sub-set overlaps with the rest of the triggers. Just to make the point clear, here are some examples of event sets:

Trigger segmentation: in the calorimeter, there are 32 trigger towers in phi and 40 in eta.

back to the top

  • Trigger Rates

    Trigger Rates

    Level 1 will accept at about 10kHz. Level 2 will accept at about 1kHz. The Tau triggers will get about 50Hz of L2 bandwidth to send to Level 3. (Level 3 will accept at about 50Hz.)

    NOTE: was not yet able to explicitly use the level 1 info (waiting for tsim_l1cal to be made public), so the following plots were made assuming that level 1 sends 100% of the information to level 2, which may not be such a bad assumption since the file used was QCD pt 10GeV. Further, Hirosky has a level 1 simulation built into "L2Cal".

    The L2Tau "tool" must be able to cut the QCD rate by X, at nominal Et cut Y, as seen in the following plot. Note the step decrease in the Z->TauTau and W->TauNu efficiency as the leading JET Et cut is increased. Obviously, with Et cut alone to meet the bandwidth requirement, the signal efficiencies can be quite low, see the lower half of the plot. Interestingly, some of the new physics trigger efficiency drops less rapidly since the Taus are harder.

    The L2Tau "tool" must be able to cut the QCD rate by X, at nominal MEt cut Y, as seen in the following plot:

    In the following plot, one can see that the L2 MEt * Jet Et missing Et monitor trigger will be firing at about 30Hz.

    Here is what a "L2Tau tool" would do to the rate if it is based on the L2 jet width only:

    To do:

    back to the top

    Simulation + Emulation

    The following simple block diagram is helpful for understanding the issues involved in simulating and emulating the trigger (all "d0-parlance" has been abstracted from the plot). The "data","trigger data", and "trigger decision" are all part of the triggered event which is written to tape. One must check that the "trigger decision", "trigger data", and "data" are mutually compatible, i.e. self-consistent.

    Only the "trigger data" to "trigger decision" can be checked via emulation of the "trigger processor" (EMULATION = description in terms of buffers and address rules). One must simulate (SIMULATION = algorithms) the "data" to "trigger data" conversion.

    In addition, one should also simulate the "trigger data" to "trigger decision" and compare it with the emulation because the emulation will be very complicated (hence full of bugs) and need cross-checking.

    Obviously, unless the state of the "detector" and "trigger processor" are recorded, no simulation or emulation is possible.

    back to the top

    Misc.

    For my trigger study, I will look at 2 luminosity scenarios:

    At 1.8TeV CoM, here are a few approximate cross-sections:
    Process Cross Section Event Rate at 5x1031cm-2sec-1 (if 100% eff.)
    QCD Pt=2GeV 39700µb 2.0MHz
    QCD Pt=5GeV 7130µb 360KHz
    QCD Pt=10GeV 684µb 34KHz
    QCD Pt=20GeV 36.4µb 1.8KHz
    QCD Pt=40GeV 1.93µb 97Hz
    W * BR(TauNu) 2.1nb 0.10Hz (6 per minute, if running 100% of time)
    Z * BR(TauTau) 0.194nb 0.0097Hz (35 per hour, if running 100% of time)
    Soft SUSY Tau 5pb 0.00025Hz (21 per day, if running 100% of time)
    HM=120 * BR(TauTau) 64fb 0.0000032Hz (8 per month, if running 100% of time)
    Hard SUSY Tau 20fb 0.0000010Hz (2.5 per month, if running 100% of time)

    back to the top

    Send comments and suggestion to bryan.smith@cern.ch

    Last modified: April 25, 2000