IE Particle Identification

Analysis Algorithm

IE particle identification is accomplished by pointing PWC tracks at the Inner Electromagnetic Calorimeter and by matching to already reconstructed clusters. The analysis driving routine is iedoid.sf, which is called by ierecon.sf.

For each pwc track which matches one and only one reconstructed cluster, we store the following variables (in the common block iepartid.inc ):

The cluster dimension is the number of blocks the cluster is made of (from 1 to a maximum of 9 blocks), the cluster isolation is the number of pwc tracks that can be matched to that cluster, the cluster fusion id keeps track of the cluster structure and its possible fusion with other nearby clusters. Variables in red are stored in DST format.

Electrons

IE Identification

A particle is identified to be electron-consistent if the following criteria are satisfied (see Figure 1 ):

  • ieclfus <100 i.e. the cluster energy is not stored mainly in the corner blocks (plots (b),(c))
  • iecldm >= 4 , i.e. the matching cluster is composed of at least 4 blocks (showering requirement) (plots (d),(e))
  • ieclis == 1 i.e. no other pwc tracks can be associated with the cluster (plots (f),(g))
  • 0.8 < E/p < 1.25 (E is the cluster energy)
For electron-consistent particles (those in plot (f)), we set the variable ieid=10 .

Additional Cerenkov Identification

It is also possible to far improve the electron identification by combining the IE analysis with Cerenkov analysis ( Figure 2 ). At present, a stand-alone Cerenkov algorithm (see iecrid.sf.html) is run which sets three increasing levels of electron consistency based on the Cerenkov counters response. Correspondingly, the IE identification variable is given values ieid=10 + 1, 2, 3 , where 13 represents the highest likelihood of electron consistency. This Cerenkov-based analysis is likely to change in the future (when the CITADEL algorithm is released).

Misid

The electron misidentification probability was studied using a sample of pions from Ks --> pi+pi- decays. Ks satisfying the following requirements were selected:

  • Ks type = 4,8,9 (better S/N)
  • goodks>=1 (skim condition)
  • cos(phi)<0.999 (e+e- rejection, phi=angle between Ks prongs)
Furthermore, Ks prongs were required to satisfy the following criteria:
  • The pwc track is 5-chambers and hits the IE
  • If the Ks is of type 4, the track has istatp=2,3
Figure 3 shows the Ks sample used for the Misid analysis, divided into momentum bins:
  • Bin 1: track momentum 0 --> 10 GeV
  • Bins 2-8: track momentum 10 --> 15 GeV, ..., 40 --> 45 GeV
  • Bin 9: track momentum > 45 GeV
(Note: each time a Ks prong -with a certain momentum- satisfies the analysis cuts, the parent Ks mass is entered in the corresponding histogram of Figure 3; consequently, each Ks mass may appear twice in the page). Each distribution was fitted with a Gaussian shape for the signal plus a second degree polynomial for the background. A region of +/- 3 sigma around the fitted Ks peak was defined as the signal region . The sideband region was defined by requiring the same combined integral under the background parametrization as for the signal region, starting +/- 5 sigma away from the Gaussian center (see Figure 4 ).

For each momentum bin, the pion/electron misidentification probability was computed as:

M(p) = [N(sig,in) - N(sb,in)]/{ [N(sig,in)+N(sig,out)] - [N(sb,in)+N(sb,out)] }
where:
  • N(sig,in) = number of signal region events defined as electrons
  • N(sig,out)= number of signal region events not defined as electrons
  • N(sb,in) = number of sideband region events defined as electrons
  • N(sb,out)= number of sideband region events not defined as electrons
The four samples N(sig,in), N(sig,out), N(sb,in), N(sb,out) are statistically independent, so the corresponding errors combine incoherently to give the error on M(p).

We considered the following sets of analysis cuts:

For each set of analysis cuts, we plotted:

  • The E/p distribution of the surviving sample integrated over all momenta
  • The misid probability vs momentum, and the global misid probability integrated over all momenta (straight line in the graph).
  • The relative efficiency of the cut vs momentum, and the total relative efficiency for the all sample (straight line in the plot). The efficiency is measured relative to adopted basic definition of electron identification.
  • The signal to noise of the surviving sample vs momentum.

In conclusion, the average electron misidentification probability for the basic set of analysis requirements (ieid>=10) is 2.9%. This number goes down to 2.0% if conservative Cerenkov requirements are imposed (ieid>=11); at the same time the relative efficiency of electron identification drops by 10%. Stricter Cerenkov requirements (ieid>=12) cause a phenomenal drop in misidentification probability, but also a big drop in identification efficiency at momenta p>=20. The strictest Cerenkov requirement (ieid>=13) reduces the misid to practically zero for p<=20, but annihilates the sample above that momentum threshold.

Muons

IE Identification

A particle is identified to be muon consistent if the following criteria are satisfied:

  • iecldm <=3 , i.e. the cluster is composed of at most three blocks with energy above threshold (anti-showering requirement).
  • 0.4 < E < 1. GeV (1 "mip" requirement; E is the cluster energy).
For muon consistent particles, we set the variable ieid=20 . In Figure 5 we show the cluster energy distribution under several analysis requirements. Some cleaning of the muon sample can be achived by requiring iecldm <=2 . The isolation requirement and the stricter condition iecldm <=1 don't help much.

Misid

The muon misidentification probability was studied as a function of momentum, using the same Ks sample and techniques which were applied to electron misid.

In Figure 6 we show the energy distribution for tracks belonging to the Ks signal region . As it can be noticed, many pions produce a 1 mip energy peak, and therefore are peaked up as muons by the IE muon identification algorithm.

In Figure 7 we show the pion/muon misidentification probability as a function of momentum, for different analysis cuts. The average misid for the standard muon id requirements (cldm<=3, i.e. ieid>=20) is about 27%.

Summary

To identify particles with the IE:

  • Electrons : require 10 <= ieid(pwc #) < 20 ; optionally, ieid==11,12,13.
  • Muons : require ieid(pwc #) >= 20 ; optionally, require iecldm<=2.

Author: Luca Cinquini ( cinquini@pizero.colorado.edu)

Last update: 24 October 1997