Vertex Finder Options and
Past Findings Summary

Dmitri Smirnov

September 27, 2017

  • Motivated by concerns expressed after QA of recently reconstructed Run 17 pp 510 GeV data
    • Increased number of negatively charged tracks with higher \( P_T \)
      link 1, link 2 by Jinlong Zhang
    • Comparison is for
      Run 17 pp500 st_W stream reconstructed with P17g
      Run 13 pp500 st_WB stream reconstructed with P16g
    • Quite different set of options: 
      DbV20170915 pp2017a StiCA btof mtd mtdCalib PicoVtxDefault
fmsDat fmsPoint fpsDat BEmcChkStat OSpaceZ2 OGridLeakFull -hitfilt

DbV20140905 pp2013b StiCA mtd btof VFPPVnoCTB beamline StiHftP
BEmcChkStat -evout Corr4 OSpaceZ2 OGridLeak3D -hitfilt
    • Need to see logs to compare unrolled list of all options
    • Change from beamline to beamline3D option is the primary suspect?
  • Also, picoDst vertex selection options

Beam Line in Vertex Reconstruction

⟨ w/o beamline ⟩ beamline beamline3D
3D fit naive 1D 1D fit 3D fit
PPV ✗ ✔ ✗
MinuitVF ✔ ✔
  • Prior to SL16k

Beam Line in Vertex Reconstruction

⟨ w/o beamline ⟩ beamline beamline3D
3D fit naive 1D 1D fit 3D fit
PPV ✔ ✔ ✔ ✔
MinuitVF ✔ ✔ ✔
  • SL17e and thereafter

Highlights of beamline vs beamline3D Algorithms

  • New functionality in STAR vertex finders does not affect past behavior, i.e. the previously existed options are backward compatible
  • In new vertex fits ( \( \ge \text{SL17e} \) ) the beam line is treated the same way in both PPV and MinuitVF
  • In PPV beamline option forces the vertex to be on the beam line ignoring the track error matrices except along the \(z\) axis
  • With beamline3D option the full track and the beam line error matrices are used in the fit
  • No additional cuts introduced for primary track candidate selection

How New Features Were Tested

  • During development performed tests with custom W and Hijing embedding samples. Tested low level performance (number of tracks, vertex position, errors, ...) rather than physics signals
    See Collab. Meeting, May 19, 2017
  • Independent tests
    • Only known (to me) test for beamline3D was done by S. Radhakrishnan and X. Dong. They used Run 16 dAu 200 GeV data.
      See S&C Meeting, April 26, 2017
    • Tests with beamline3D showed expected features including
      • Sharper vertex peak in \( x-y \) plane due to vertex attraction to beam line
      • Similar distributions without
      • Wider DCA distribution with beamline
    • Not beamline3D related but L. Adamczyk, L. Fulek, and R. Sikora (LFS/UPC) performed embedding studies with PPV to test recent tunes of beamline option

Run 16 dAu 200 GeV

  • By S. Radhakrishnan and X. Dong
  • As of today, no independent tests for beamline3D in pp data

Vertex Options for picoDst Production

  • There are three vertex options:
    PicoVtxMode::Default
    Select the first vertex in the muDst vertex collection. Usually the highest rank vertex
    PicoVtxMode::Vpd
    Select a vertex matching the VPD one, i.e. select the first available muDst vertex within a window around the VPD vertex
    PicoVtxMode::VpdOrDefault
    If a vertex matching the VPD one is not found pick the first one from the muDst collection
  • Somewhat confusing but PicoVtxMode::Default does not mean the option will be selected by default!
    Might be a good idea to rename the option to First, Best, MaxRank, or ...
    ...and use it by default when no other option is provided