Trying the New Shower Function on the Data

Starting from scratch, I grabbed the original calibration file from the Fortran analysis and performed one calibration path using the new shower shape. I've also included the slight tweak on Larisa's Chi2, as discussed here. The result was compared to the older results that went through multiple iterations of calibration path based on some version of the procedure. 

Fig. 1. Reconstructed Position Distribution for 2 cluster events, default shower shape

Fig. 2. Reconstructed Position Distribution for 2 cluster events, new shower shape

Fig. 3. Reconstructed X-Position Distribution for 2 cluster events, BLACK: default, RED: new 

Fig. 4. Reconstructed Y-Position Distribution for 2 cluster events, BLACK: default, RED: new 

For 2 cluster events, we see a clear improvement on X, while on Y we get a structure that is just as noticeable but at least somewhat less coarse. Therefore at least for the purpose of calibration, which only uses 2 cluster events, the new shower function with slightly tweaked Chi2 seems to be workable for the moment. 

Now we look at the 1 cluster events, where the story is not as happy. 

EDIT: I've found a bug in 1 cluster fitting algo. I'm not sure the extent to which this bug confuses the fitting, and I'll have to run reconstruction again to find out.