Does ignoring unhit sites change the likelihood?
Test 1 - Normal reconstruction, v3.03Notes:
- Taking away unhit sites improves the likelihood at a given energy.
- Given the same set of sites, the likelihood is worse at higher energies
even though the LDF fit appears better (and chisquare is better)
=> Something else is causing the likelihood to behave oddly.
Comparing the dummy reconstruction using 2.42 with 3.03: the "fit" time is different between these versions. Everything else is the same, because the data is being pulled from the "fitshower" object (Brant's) not "fitshower_Shawn". However, the fit values are calculated in ShowerFitFunction, and in Shawn's version, Takeda's time function is replaced with the Mode of the TDF. This appears to move the times systematically earlier.
Test V. 242