Vertex findingand
B-Taggingfor the
ATLAS Inner Detector
A.H. WildauerUniversität Innsbruck
CERN ATLAS Computing Group
on behalf of the ATLAS collaboration
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Introduction
The development of the vertexing and B-Tagging software followed these basic principles:
• Requirements:
- integrated in the ATLAS Athena framework
- compatible with the ATLAS Event Data Model
• Define and use standardized interfaces
• It has to be modular and expandable for future needs
e.g.: implementation of new vertex fitting methods
adding new tag algorithms to the B-Tagging
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
ATLAS Event Data Model
ESDEvent Summary Data
AODAnalysis Object Data
Analysis
raw data~1.6 MB
~500 kB
~100 kB
• amount of data per year: ~ 1PB
• cost of distribution and storage
• standardized Event Data Model
for online and offline is necessary
• e.g. common EDM for tracking(see talk by E. Moyse)
• Vertexing and B-Tagging should work on ESD and AOD
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Design of Vertexing Package
A client feeds a fit method with tracks andgets back a vertex candidate
It can choose the fit method and other parameters via a Python steering file
VertexFit package implements differentfitting methods through inheritance
The base class takes care of commonframework issues
Modularity: derived fit classes only haveto implement the fit method
tracks vertex
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
• Currently implemented are 2 fit methods: “full” and “fast”
• Mathematical basis: the “Billoir” method (Fast vertex fitting with a local parameterization of tracks
Nuclear Instruments and Methods in Physics Research A311 (1992) 139-150)
• Local parameterization of track parameters q at a fixed point, e.g. the perigee, as a function of V and p
• Minimize the 2:
• Result: vertex and track parameters
Vertex Fitting Methods
),( where2ipVFmeas
iqiqiqiWiTiq
ipiEViDoip
oVFipVF ),(),(
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Primary Vertex Finder
• applied standard track selection on pt, # of hits, aso.
• 2nd vertex fit after 2 track selection
Reconstruction of the primary vertex in WH(120)uu
Example application using the VertexFit package
preliminary
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Primary Vertex Finder with pile-up
• look for primary vertices in windows of z
• the primary vertex is defined as the one with the highest pt
Reconstruction of the primary vertex in
H(130) ZZ* 4l with pile-up
Pile-up: about 23 vertices will cover the primary vertex of an interesting physics event
another example
preliminary
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
B-TaggingUse Cases and Considerations
• Ingredients for B-Tagging are e.g.
1) Signed impact significance
2) Secondary vertex information (mass, multiplicity, …)
3) Particle identification (lepton tag)
• And are provided by the
1) Track/TrackParticle and primary vertex (ESD/AOD)
2) vertex finder using the VertexFit package
3) particle objects in the AOD
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
B-Tagging Package Design
The Algorithm retrieves relevant info from StoreGate and calls several tag Tools
The Tools perform the tagging and add their results to the BJet object
The Algorithm combines their information into a single btag likelihood and returns the resulting BJet
Tools, order, cuts aso. can be set via Python steering files
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Design of the BJet Object
The BJet object stores all information of the btag
It inherits from several classesto obtain the functionality necessary for an EDM object
In addition to the kinematics ofthe jet, it contains the overall likelihood of the btag
It has a vector of IBInfo like objects which are used tostore tag specific information
1… *
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Various discriminating variables used for B-Tagging:
• Signed impact significance
• 2D distance between 2nd and primary vertex
• vertex mass
B-Tagging: Results
b-Jetsu-Jets
preliminary
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Likelihood
Combined likelihood and efficiency with
• 2D impact parameter
• secondary vertex tools
b-Jetsu-Jets
preliminary
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
Conclusion
Vertexing and B-Tagging software is presented which:
• is implemented in the ATLAS Athena framework
• is compatible with the ATLAS Event Data Model
and therefore works on ESD and AOD
• is modular and expandable to suit future needs
• first prototype implementations have been done and look encouraging
• future work involves lots of tuning of parameters
28-Sep-2004 CHEP 2004 Interlaken Wildauer Andreas
SPARE
Secondary Vertex Finding
• Inside the secondary vertex tag tool, two strategies for finding secondary vertices in a jet are in use:
Build-Up
• fit all 2 track verticesin a jet
• take lowest 2 one• add tracks 1 by 1 and refit• take track if 2 < 4
Tear-Down
• use all tracks in a jet tofit a vertex
• remove tracks with 2 > 4• refit the vertex• take track if 2 < 4
• the vertex fitter Tool is called from within the SecVtxTag
Top Related