FCAL Reconstruction 04/08/2012

From GlueXWiki
Jump to: navigation, search

Efficiency

There is a tunable parameter in the FCAL code with default:

MIN_CLUSTER_BLOCK_COUNT = 2;

At lower energies (<~500 MeV), this causes photon reconstruction efficiency to be lower because often only one block has energy deposited in it (above the 20 MeV threshold in mcsmear).

In single photon sample:

SinglePhotonFCAL.recon eff.minBlock1+minBlock2.svn8951.png

  • red: MIN_CLUSTER_BLOCK_COUNT = 1
  • black: MIN_CLUSTER_BLOCK_COUNT = 2

In b1pi events:

  • Overall FCAL efficiency increases by ~6%: 78% -> 84%
  • 15% in total "photon" count (after timing cut) (1.18->1.36)

1-block "clusters" will collect less energy than multi-block clusters of same photon energy-->will affect energy calibration, energy resolution

Energy resolution

SinglePhotonFCAL.sigma E err.minBlock1+minBlock2.svn8951.png

  • red: MIN_CLUSTER_BLOCK_COUNT = 1
  • black: MIN_CLUSTER_BLOCK_COUNT = 2

Fit to black curve says sigma(E) = 6.1%/sqrt(E) ⊕ 1.9%

SinglePhotonFCAL.mean E err.minBlock1+minBlock2.svn8951.png

Need to recalibrate energy corrections at low E.

FCAL energy corrections

What's the deal with FCAL energy correction?

  • current code by Mihajlo (DocDB 1093)
  • non-linear correction corrects for attenuation, lack of full containment?

Assume form

Nonlin energy corr eq.png,

where E_c is the energy of clustered blocks and E_gamma is the energy of thrown photon.

Fit MC data for A,B,C,\epsilon .

In reconstruction, get E_corrected from E_clust by solving this equation iteratively.

The simplest option for recalibrating would be to use the current scheme but re-fit for A,B,C,\epsilon using sample with 1 block clusters. But is there a better way?

E_raw/E_thrown plots

Check for theta dependence (currently not accounted for):

SinglePhotonFCAL.ratio 2D theta.svn8951.png

Only apparent at low angles near hole in FCAL.

Dependence on number of blocks in cluster

SinglePhotonFCAL.ratio 2D n blocks.svn8951.png