Difference between revisions of "Photon Reconstruction in b1pi events 02/10/2012"
(→Reconstructed photons) |
(→Cuts) |
||
Line 22: | Line 22: | ||
* A timing cut fabs(t_shower-t_flight) < 1 ns can be applied to remove many of the extra "photons". (Red curve below) | * A timing cut fabs(t_shower-t_flight) < 1 ns can be applied to remove many of the extra "photons". (Red curve below) | ||
* I don't think low energy showers are well understood at the moment, so remove clusters with E<60 MeV. (Green) | * I don't think low energy showers are well understood at the moment, so remove clusters with E<60 MeV. (Green) | ||
− | * Also, cut out a problem area at forward angles and lower energies (E< | + | * Also, cut out a problem area at forward angles and lower energies (E<120 MeV && z>300 cm) (Blue) |
More work needed to optimize cuts. | More work needed to optimize cuts. |
Revision as of 18:44, 13 February 2012
BCAL
Thrown photons
Look at the distribution of decay photons from the pi0.
For comparison with reconstructed data, useful to look at z rather than theta. This is the z that the photon would reach the inner radius of the BCAL, given its momentum and vertex.
Reconstructed photons
Spectrum of reconstructed "photons": exclude clusters that can be matched to a charged track.
Where does each of these peaks come from?
- Peaks at upstream and downstream ends of detector from noise that cause garbage timing info?
- Other three sharp peaks? Look like FDC structure, but spacing is not right?
Cuts
- A timing cut fabs(t_shower-t_flight) < 1 ns can be applied to remove many of the extra "photons". (Red curve below)
- I don't think low energy showers are well understood at the moment, so remove clusters with E<60 MeV. (Green)
- Also, cut out a problem area at forward angles and lower energies (E<120 MeV && z>300 cm) (Blue)
More work needed to optimize cuts.
With all cuts:
Number of "photons" decreases from 60,000 to 17,000.
Two gamma invariant Mass
Look at pairs of 2 BCAL photons and 1 BCAL+1 FCAL photon, with the cuts described above. Using truth vertex information.
Compare KLOE algorithm to GlueX algorithm.
Black is KLOE, red is GlueX. Fits are to gaussian + straight line.
GlueX has less background, the fit also indicates that it has a taller peak, but I don't know how much this fit can really be trusted.