This MR updates the function used to fit the RICH CK resolution (used at the pit for the refractive index calibration). The aim is to give a slightly better fit to the data by
- Allowing some asymmetry in the signal Normal, by way of allowing the +ve and -ve side resolution terms to be slightly different. Only small asymmetries are required, O(few %).
- Use a hyperbolic background term instead of a cubic polynomial. Gives 'flatter' tails which seems to better fit, most of the time.
Fit has quite stringent checks on the convergence, and falls back to simpler forms if a fit fails, so should always provide a reliable result.
Following PDFs give examples of the old and new fits.
I have submitted this against run2-patches, as I am more looking to run this in Run3. The current fit works fine as far as extracting the scale factor (the normal mean) and as it has been used for 2016, 2017 and 2018 (so far) I have decided to leave things as they are until the end of Run 2.
The fitter is not used as part of the HLT2/offline reconstruction, so should have no impact on any of the QM tests.
Please do propagate it to master though.