Closed hartytp closed 1 year ago
I'm going to close this for now as there is no immediate action. This is something which the library was designed to accommodate and should be possible without invasive changes to existing code; just needs doing when someone actually wants it!
Most relevant are
PoissonianFit
andBinomialFit
. Probably the best path here is to add aMLEFit
class which inherits fromFitBase
and handles the optimization, calculation of error contours etc. and gets statistics from a method which isNotImplementedError
by default. NBNormalFit
uses ay_err
input. Binomial will neednum_shots
while Poisson distributions don't need anything beyond the x and y data.