Minuit vertex finder recent development

1
Minuit vertex finder recent development

• Matthew Cervantes
• Texas AM University

2
Introduction

• I. Importance of primary vertex reconstruction
• - Physical quantities used in analyses anchored
  to vertex location
• - Associating the event vertex to the event that
  fired a trigger
• II. Electronic readout limits of STAR detector
• - TPC (80 micro-seconds) and EMC (10
  nano-seconds) readouts
• - High luminosity and collision rates lead to
  vertex pile-up
• III. STAR uses two vertex finding algorithms
• - Minuit Vertex Finding (MinuitVF)?
• - Pile-Up Proof Vertexer (PPV)?

3
MinuitVF Method

• I. High-quality tracks for vertex reconstruction
• - Vertex candidates found via z-distribution of
  DCA to
• beam-line.
• - Tracks require radial DCA to the beam-line
  less than 2 cm.
• II. Crude peak-search performed w/ these tracks
• - Candidate vertices have at least 5 tracks
  that extrapolate
• back to within 6 cm of z.
• III. Minuit minimization routine and vertex rank
• - Minimize DCA and robust potential used in
  ?2 to reduce
• outlier track contributions.
• - After stable point reached (or max
  iterations) rank is
• assigned.

4
MinuitVF Method (continue)?

• IV. ltdip anglegttrack
• - The angle between a track and
• z-axis.
• - Vertices reconstructed near the
• edge of TPC have fewer hits
• than those from the center
• skewed ltdip anglegttrack
• V. The ranking of a vertex candidate
• - ltdip anglegttrack vs. z-position of vertex
  (see upper right)
• - of tracks matched to energy deposits in the
  BEMC
• - of tracks which cross the central membrane
  of the TPC

5
Detector limits, collisions and Pile-Up

• I. Detector readout and bunch crossings
• - Peak collision rates for dAu 2008 roughly 300
  Hz.
• - TPC up to 20 separate minbias collisions per
  event recorded.
• - A TPC readout less than the collision rate
  will lead to pile- up.
• - Pile-up vertices are proportional to
  instantaneous luminosity.
• - Increased luminosity will only exacerbate the
  pile-up issue.
• II. There are three types of pile-up that occur
• - Pre-crossing pile-up, post-crossing pile-up and
  bucket
• pile-up.
• - Bucket pile-up vertices are due to collisions
  within the same
• bunch crossing. Can not be distinguished/removed
  from
• data.

6
Pre and Post-Crossing Tracks

• The three types of pile-up vertices cartoon
• - fig. c Bucket pile-up is not separable
• - fig. b Pre-Crossing occurs before
  triggered vertex.
• - fig. a Post-Crossing (red pile-up vertex
  occurs after the
• blue triggered vertex).

7
Towards MinuitVF PCT rejection

• I. PCT rejection currently implemented into
  the
• PPV package.
• II. PCT rejection needs to be implemented into
  the
• MinuitVF package and then tested.
• III. Inspection of PPV package and Vertex
• threads have yielded insight into how to
  go
• about doing this.

8
MinuitVF PCT rejection

• I. We have put a piece of code similar to the PPV
  function
• isPostCrossingTrack() into the Minuit VF,
  then ran the
• bfc.C() over a 2008 dAu 200 GeV daq file.
• bfc.C'(1,100,"DbV20080712 B2008a ITTF IAna
  hitfilt ppOpt VFMinuit l3onl emcDY2 fpd ftpc trgd
  
• ZDCvtx NosvtIT NossdIT Corr4 analysis
  OSpaceZ2 OGridLeak3D beamLine debug logger,
• "/star/data05/scratch/mcc/VertexFinder/rosi
  _dAu2008_daqfile/st_physics_9026071_raw_1010017.da
  q")'
• II. MuDst Tree from the daq output then used to
• create an ntuple of interesting quantities
  (see
• Rosi's previous vertex blog) for future
  study.

9
MinuitVF PCT rejected vertex

• I. Comparison of 100 dAu events appears to shift
  the of
• vertices distribution in the expected
  direction (decrease).
• II. An in-depth study of the effects of PCT
  vertex rejection will
• follow as I learn the details of how MinuitVF
  works.