Ring Imaging Cherenkov Counter
Track based particle identification algorithm
Iterative algorithm for the analysis of the Cherenkov rings 


Identify reconstructed tracks as one of the particle : electron, muon, pion, kaon or proton.
Usage in the analysis:

For the  time being, there is more than one method used  for particle identification in the RICH detector.  Therefore  RICH likelihoods are copied to the parallel RTRI table and the entries in the table RTRA are  reset to -1.  Therefore one should  use one of the special functions, which are implemented to copy the corresponding fields in RTRI back to RTRA or to combine them in the unified likelihood. The later method suffers from the unknown efficiency vs. misidentification relation.

The  following function should be used to copy likelihoods calculated wit one method back to the appropriate places.

int copyRTRItoRTRA(int algorithm);                 // algorithm = 1 for RITER or =0 for RISE

This function should be called after the reconstruction in the


  • Soft cut lcut=0.1
  • Hard cut lcut=0.9
      Apply the following selection:
  • identify kaon : lrk>lcut
  • identify proton: lrp>lcut
  • Separate light particles ( electron, pion, muon) from protons and kaons: lre+lrmu+lrpi > lcut
    Make sure that HITB table for the RICH is filled:
  • Note that if you omit HITB in the last command, there will be also RICH standalone package RISE running, which is not needed by RITER, but it consumes a lot of CPU-time!!!

    Make sure to include appropriate kuip commands for the track reconstruction in your ARTE kumac and that refit is on, e.g:

  • etc.

    To run the iterative analysis during the reconstruction include the following kuip commands:

  • /RITER/DWIN 1000


    Particle identification module RITER needs as an input:

  • RICH detector hits ( HITB table )
  • track parameters in front of the RICH detector ( RPNT table )


  • Likelihood functions are calculated for the valid charged tracks (rtra->fit=0) with hits in vertex detector system (rtra->hsi>0) and in the main tracker (rtra->hou+hin > 0)
  • Likelihoods are normalized in the way that sum of the likelihoods for electron, muon, pion, kaon, proton and background hypothesis is one. For the multielectron hypothesis  the  normalization factor is the same.



    During the reconstruction, the following field in the table of the reconstructed tracks RTRA are filled.
    rtra->flr & 0xff RITER particle identification flag  : If greater than 0, a track   is within geometrical acceptance of the RICH 
    rtra->lre likelihood of electron hypothesis
    rtra->lrmu likelihood of muon hypothesis
    rtra->lrpi likelihood of pion hypothesis
    rtra->lrk likelihood of kaon hypothesis
    rtra->lrp likelihood of proton hypothesis
    rtra->lrme likelihood for multi electron hypothesis
    background = 1- (lre+lrmu+lrpi+lrk+lrp) likelihood for hypothesis that the distribution of photons around track is consistent with backrgound


    The iterative method is based on the assumption that most of the RICH detector hits, which represent a background in the Cherenkova angle histogram, originate from other tracks within the saame event. An iterative procedure is used to clear such histograms.  A weight is assigned  to each Cherenkov photon hit - track pair. After the iterations, the photon hit has the highest weight  for the track, where represents the signal and lowest where it is part of the background.

    The weigted photon hits are used for the  calculation of  likelihood functions for different hypotheses.

    The method is described in more detail in

  • An iterative method for the analysis of the Cherenkov rings in the HERA-B RICH ,
  • An iterative method for the analysis of the Cherenkov rings ,
  • Identification of pions, kaons and protons in the HERA-B spectrometer , Doctoral thesis, Ljubljana 2001


    Invariant mass distribution for decay of Phi to two Kaons. Different RICH likelihood cuts are applied in the selection of track pairs

  • Results (PID matrix) of 1. reprocessing
  • Results (PID matrix) of 2. reprocessing
    Related links:

  • PiD using RICH in the Single Lepton data sample
  • $HBROOT/ANALYSIS/SingleLepton/anl_richid.C , Single lepton RICH PID analysis routine
  • Ring Search FAQ
  • Hera-B RICH homepage
  • Hera-B software page
    7.4.2001 Rok Pestotnik