Further Study on GSTAR Cerenkov Shower

In parallel with Steve, I've been looking into the Cerenkov shower in GSTAR. There are still some unresolved differences between what I'm doing and what Steve's doing. Most importantly, I am not entirely sure how the glass-mylar interface is being handled yet. That is, I haven't found out if there is an air gap between the glass and mylar in the geometry file.

Unless otherwise noted, all results used absorption, reflection, and cathode efficiency in the NET paper.

Fig. 1. Number of Cerenkov Photons vs. cos(theta) where theta is measured from Z-axis. 

BLACK: Total generated, BLUE: Those that reached the photo-cathode, GREEN: Actually "measured" after passing photo-cathode efficiency

Fig. 2. Number of Cerenkov Photons vs. Z-position of the generation

BLACK: Total generated, BLUE: Those that reached the photo-cathode, GREEN: Actually "measured" after passing photo-cathode efficiency

Fig. 3. cos(theta) vs. Z-vertex

In order to produce an incident angle dependent model of the shower, I've run 4 separate jobs, each containing Cerenkov photons that were generated in the specific Z-slice.  

Fig. 4. 4 Z-slices of the Cerenkov shower

As expected, the shower is very sharp at the beginning, and gets broader towards the end.

I've also re-measured the shower shape in data to be safe.

Fig. 5. Shower Shape in DATA re-measured with the most recent calibration

RED: GSTAR shower shape using energy loss with incident angle, GREEN: GSTAR shower shape using Cerenkov with incident angle 

The shower shape measurement in the data has changed slightly from the last measurement, which was about 15 iterations ago. The major difference is that the central bin fraction has come down by ~1%. I haven't found out the source of this difference yet. It is still narrower than the energy loss based shower, but wider than what we get from GSTAR using the "normal" absorption.

There are two evidences that suggests that we are underestimating the absorption. 

1. The shower in the data is somewhat wider, which can come about if absorption is higher.

2. Cerenkov shower introduces an actual energy dependent change in energy measurement. This seems to correlate with the level of absorption, and with the normal values, I seem to get energy dependence that is potentially smaller than what we see in the data. 

Fig. 6. The effect of putting 0.3eV shift in absorption function to emulate PgG radiation damage

The shift in the shower shape does go in the right direction, but at this point it is not clear if this is consistent with Steve's findings.

The following plot shows the energy dependence of gain when using the normal absorption for pi0s.

Fig. 7. E_sum / E_gen vs. E_gen, normal absorption, pi0s.

Compare this result to the case when we put in 0.3eV and 0.5eV shift in the absorption.

Fig. 8. Energy dependence with normal, 0.3eV shift, 0.5eV shift, for pi0s.

BLACK: normal absorption, RED: 0.3eV shift, BLUE: 0.5eV

We know that this effect is coming NOT from the change in number of generated photons. 

Fig. 9. N_gen (normalized) vs. Energy for single photons

So it makes sense that the change in absorption would change the energy dependence. However, the change I had to put in to see significant change, which is 0.5eV shift, is quite big, and I'm not sure if this result is consistent with Steve's.