Displaced Vertex analysis trigger strategies for Run2

Triggers for Run2

The list of triggers implemented in DC14 Run 2 MC can be found in TriggerMEnuDC13Run2.

DV signatures

Triggers used at Run1

Efficiencies, rates,...

Potential interesting triggers

Trackless-Jet trigger

Calorimeter Ratio

Muon RoI cluster trigger

b-jet triggers

B-jet triggers exploits the presence of a b-jet characterized by:
  • Relatively large lifetime of B-hadrons (around 1.5 ps, i.e several millimeters before decaying): produce secondary vertices.
  • Secondary vertices with high mass associated (B-hadrons masses > 5 GeV)

The b-jet triggers use several L2 and EF algorithms, being the main package Trigger/TrigHypothesis/TrigBjetHypo (next info extracted from doxygen page). The package contains the HLT b-tagging selection based on the likelihood ratio method and on a track-based chi2 probability method. It is running both at LVL2 and EF and it basically contains a feature extraction algorithm to compute the b-tagging weight of all the different taggers and an hypothesis algorithm which implements the selection.

The track-based chi2 probability tagger (CHI2) computes the probability for a jet to originate from the primary vertex using the signed transverse impact parameter significance of tracks pointing to the jet. The likelihood taggers implemented are based on the signed impact parameter significance of tracks and on the different properties of secondary vertices; in particular:

  • significance of longitudinal impact parameter (IP1D);
  • significance of transverese impact parameter (IP2D);
  • 2D combination of the track-based methods (IP3D);
  • invariant mass of tracks linked to the secondary vertex (MVTX);
  • energy fraction of the secondary vertex (EVTX);
  • number of tracks lined to the secondary vertex (NVTX);
  • 3D combination of the vertex-based methods (SVTX);
  • combination of the two likelihoods based on tracks and secondary vertex informations (COMB).
Hypothesis algorithm can perform the selection using these different taggers: CHI2, IP2D, IP3D or COMB. Also, in order to optimize b-tagging performance keeping the reconstruction efficiency as high as possible, in the feature extraction algorithm a dedicated track selection has been added. The main classes are:
Class Purpuse
TrigBjetHypo Derived HLT::HypoAlgo class for the Bjet trigger, it contains the algorithm to make decision using the TrigBjetFex class
TrigBjetFex FEX (Feature extraction) class, i.e. where it is performed the relevant calculations (extracting the relevant quantities from the input Trigger Element), based on track impact parameters and secondary vertices. Also it creates Bjet object and attaches it to TE.
TrigBjetTagger Base class for the computation of the different likelihood taggers

The triggers implemented in the DC14 MC samples can be found in TriggerMEnuDC13Run2. In particular, the main physics triggers for b-jets are

Trigger L1 Description
j75_bmedium_3j75 4J20 1 b-jet in 4 jets trigger
j175_bmedium_j60_bmedium J100 2 b-jet trigger with asymmetric thresholds. Still some technical problems.
j175_bmedium J100 Single b-jet trigger.
j300_bloose J100 Single b-jet trigger.
2j55_bmedium_2j55 4J20 2 b-jets in 4 jets trigger.
2j45_bmedium_3j45 5J15.ETA24 2 b-jets in 5 jets trigger.
Note that:
  • All chains by default use the same b-tagger as in Run1 (IP3D+SV1) ---> Investigate what it means.
  • All chains by default use anti-Kt R=0.4 jets running on topo clusters ---> Investigate what it means.

Any of those chains could be used to relax the lepton and/or jets kinematic thresholds (also in L1), keeping affordable rates by using the vertex information.

Proof of Concept
First test: trying to find b-jets trigger information in signal samples for MC12: RPV signal with heavy/light neutralino. The samples used was processed with the displaced vertex software to obtain n-tuples (info extracted from ATLAS-COM-PHYS-2014-370).

The table is showing the fraction of the total events that each trigger chain accepted (#events trigger accepted/#total event).

Trigger chain Neutralino --> mu+jets (DS: 114006) M_n=498GeV Neutralino --> mu+jets (DS:202874), M_N=108GeV Neutralino--> e+jets (DS:177572) M_n=108GeV Neutralino--> e+jets (DS:174280) M_n=498GeV
EF_b45_mediumEF_j110_j45_xe60_tclcw -- 17.510% 7.716% 4.768%
EF_b55_mediumEF_j110_j55_xe60_tclcw -- 17.134% 7.576% 4.616%
EF_b80_loose_j80_a4tchad_xe55_tclcw -- 18.814% 9.559% 4.640%
EF_b80_loose_j80_a4tchad_xe60_tclcw -- 18.528% 8.571% 3.908%
EF_b80_loose_j80_a4tchad_xe70_tclcw -- 18.112% 7.336% 2.970%
EF_b80_loose_j80_a4tchad_xe75_tclcw -- 17.626% 6.576% 2.441%
EF_b110_loose_j110_a4tchad_xe55_tclcw -- 17.684% 9.184% 4.149%
EF_b110_loose_j110_a4tchad_xe60_tclcw -- 17.432% 8.226% 3.508%
EF_e24vh_medium1_b35_mediumEF_j35_a4tchad -- 0.042% 2.277% 4.775%
EF_mu24_tight_b35_mediumEF_j35_a4tchad 7.763% 3.502% 0.769% 0.421%
Run-1 Anaylisis
EF_mu50_MSonly_barrel_tight 65.198% 32.218% 1.752% --
EF_g120_loose -- -- 50.941% 83.863%
EF_2g40_loose -- -- 28.429% 79.026%

The table shows the b-jet triggers found in the D3PD MC signal samples along with the triggers used in Run1 for the DV+muon and DV+electron analysis. The total efficiency using the leptonic triggers is greater than using the b-jets in both cases. Nevertheless, we should expect a dependency of the efficiency with the displaced vertex position, in the case of the b-jet triggers, given that the b-jet triggers are tuned to tag b-jets, so we would expect a bigger efficiency around 1-5 mm from the primary vertex. Next plots confirm this hypothesis: the efficiency of the b-jet triggers is greater where the displaced vertex mimics a b-jet (presence of a secondary vertex at several millimeters from the primary vertex with an associated high-mass), i.e. around 1-5 mm. In the case of the light neutralino samples, the decayed lepton has no access to enough transverse momentum energy most of the time to exceed the trigger threshold, therefore the b-jet trigger in the 'b-jet' region is already more efficient than the leptonic one (muon sample). However, this is accomplished with a trigger tuned to select b-tagged events, consequently extending the ability of these triggers to select secondary vertices with high-multiplicity of tracks and high-mass will allow to increase significantly the acceptance of the signal, especially in light LSP scenarios.

M_{neutralino}= 498 GeV M_{neutralino}=108 GeV
mu+jets
eff 114006 mujets Mn494 Msquark700 GroupedTriggers Drdv.png
MC truth distance of the displaced vertex with respect to the interaction point
eff 202874 mujets Mn108 Msquark1000 GroupedTriggers Drdv.png
MC truth distance of the displaced vertex with respect to the interaction point
e+jets
eff 174280 ejets Mn494 Msquark700 GroupedTriggers Drdv.png
MC truth distance of the displaced vertex with respect to the interaction point
eff 177572 ejets Mn108 Msquark1000 GroupedTriggers Drdv.png
MC truth distance of the displaced vertex with respect to the interaction point

Developing the trigger
Follow DVAdaptedBjetsTrigger for a detailed log of the development of displaced vertex trigger adapted from b-jets.

Bphysics triggers

Trigger Software

Trigger as an user: obtaining trigger decisions, objects and matching information from data/MC

The main page guiding a new user through the trigger is the twiki TriggerUserPages. This page is indented to give information to an analysis user with trigger information needed. The two main tools availables are the TrigDecisionTool (TDM), used to obtain trigger decision information and the TriggerObjectsMatching, which is used to determine the relationship between objects generated by offline reconstruction and by objects generated by the trigger (L1, L2, and EF) algorithms.

In the context of testing the b-jets triggers availables, we perform the exercise (following the last trigger tutorial available at TriggerTutorialForAtlasOfflineSWTutorial) to obtain the trigger decision of a b-jet trigger, and also do some matchings:

  • Step 1: Obtain suitable samples
    We want to obtain the
  • Step 2: Get the trigger decision of the b-jets triggers.

Trigger as a developer: developing a new trigger signature

Instructions to develop a new trigger signature, as well as edit and run the TriggerMenu (TM) can be found in TriggerMenuPackageInstructions.

-- JordiDuarte - 15 Oct 2014

Topic attachments
I Attachment History Action Size Date Who Comment
PNGpng eff_114006_mujets_Mn494_Msquark700_GroupedTriggers_Drdv.png r1 manage 15.7 K 2014-10-15 - 21:57 JordiDuarte Trigger efficiencies for MC signal RPV (neutralino-->mu+jets), M_{neutralino)=494 GeV
PNGpng eff_174280_ejets_Mn494_Msquark700_GroupedTriggers_Drdv.png r1 manage 31.3 K 2014-10-15 - 22:20 JordiDuarte Trigger efficiencies for MC signal RPV
PNGpng eff_177572_ejets_Mn108_Msquark1000_GroupedTriggers_Drdv.png r1 manage 28.9 K 2014-10-15 - 22:20 JordiDuarte Trigger efficiencies for MC signal RPV
PNGpng eff_202874_mujets_Mn108_Msquark1000_GroupedTriggers_Drdv.png r1 manage 28.5 K 2014-10-15 - 22:20 JordiDuarte Trigger efficiencies for MC signal RPV
Edit | Attach | Watch | Print version | History: r10 < r9 < r8 < r7 < r6 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r10 - 2014-10-21 - JordiDuarte
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback