GaussianLikelihood implementation
@gregory.ashton @colm.talbot
I am currently reviewing some code and if I understand it correctly, there is an issue in the implementation logic in GaussianLikelihood
. The sigma
parameter in the __init__
is only set if it is None
if self.sigma is None:
self.parameters['sigma'] = None
I think this should just be
self.parameters['sigma'] = sigma
Also this line in the log_likelihood
looks suspicious:
sigma = self.parameters.get('sigma', self.sigma)
This defaults to self.sigma
if sigma
is not set in self.parameters
. With my proposed implementatioin, this should not happen.
I currently work on some separate branch and fixed it there. If the author of this code can confirm, I will move ahead with it.