jwebb's blog


{
  mEEclusters->setSmdSeedEnergy( 3.0, 0.0 ); // seed and slope [MeV]
  mEEclusters->setSmdMaximumSize(4);
  mEEclusters->setSmdMinimumStrips(5);
  mEEclusters->setSeedEnergy(2.0);
  mEEclusters->setMinimumEnergy(0.5);

  mEEpoints=new StMyPointMaker("mEEpoints",mEEanalysis,mEEclusters);
  mEEpoints->setSplit();
  mEEpoints->setSplitMinimumET(6.5);
}

Cuts:

0) Reconstruct a π⁰ candidate

1) Isolation cut -- ET(π⁰) / ET(0.3) > 0.9

n.b. isolation cut does not include BEMC towers

2) CPV cut -- E(pre1) == 0 w/in R<0.3

3) Analyzer -- E(pre2) > 0 w/in R<0.3

1.0 Method

π⁰candidates are reconstructed with the π⁰finder. We apply the same set of gamma candidate selection cuts to the π⁰ sample. Events which survive the CPV cut are binned in psuedorapidity. We count the number of events reconstructed with 0.1 < M < 0.2 which satisfied the CPV cut and the analyzer.

Figure 0 -- Invariant mass spectrum for all events which satisfy the CPV and analyzing cuts (left). Conversion probability w/in R<0.3 versus digamma invariant mass (right).

Obserrvations:

1) The background efficiency w/in the pi0 mass region (0.1 to 0.2 GeV) is about 91.5%, averaged over all eta, pt, etc...

2) We see a consistent efficiency versus invariant mass.

3) The eta mass region (0.5 to 0.6 GeV) is consistent w/ the pi0 mass region.

4) The yield above the pi0 mass (and not in the eta region) is given by combinatoric backgrounds from correlated photons from the jet, and uncorrelated random backgrounds.

I take this as a confirmation that this is an appropriate surrograte for measuring the jet background in the gamma sample.

Figure 1 -- Conversion probability for background events with an identified π⁰, plotted versus pseudorapidity. The fit is described below. The red(ish) lines indicate conversion probabilities for 1,2,3 and 4 gammas versus pseudorapidity. At right we show the "D" distribution for the events used to measure the efficiency. These events show no significant contamination from neutral hadrons.

The fit in the above figure uses Scott's observation that background conversion probability should be related to the single gamma conversion probability by the expression

(1.0 - ε_bg) = (1.0 - ε_γ)^N where "N" is the number of detectable gammas within the isolation cone. The single photon conversion rate is calculated based on the known

amount of material in the first radiator and the incident angle, i.e. 1.0 - ε_γ= e^{-0.6873 / cos(θ)}.

2.0 Observations and Conclusions

The shape of the measured background conversion rate is consistent with the shape expected given the amount of material in first radiator of the EEMC and a photon multiplicity of 3.35 w/in a jet cone of R<0.3 around the reconstructed π⁰ candidate.
It may be possible to perform a 2-parameter fit, simultaneously fitting ε_γ and "N" from the π⁰ conversions. However, a quick-n-dirty attempt to do this resulted in rather large uncertainties in both the single photon efficiency and the number of gammas w/in the isolation cone. The two parameters were highly correlated w/in the fit. This probably requires better statistics to dig out.
If I play around a bit with step sizes, initial conditions and parameter boundaries, I can convince Minuit to deliver something in line with my expectations for a 2-parameter fit. Specifically N=3.42+/-0.5 and MFP=0.68+/-0.04. The chi2/dof remains at 21/11. However, to get this result I need to constrain the MFP to be w/in +/- 2% of its nominal value (reasonable, since this is the integration tolerance for the 1st radiator). But I also need to constrain the number of photons to be between 3 and 4. If I don't do this, the uncertainties get rather large.
I've reduced the SMD seed threshold to 2.0 MeV, and am rerunning the pi0 finder this evening, to try to get more statistics.