Closed TTitscher closed 3 years ago
In the current implementation I (I think) introduced a bug in the loglike_contribution that was like
+ np.sum(np.square(error / sigma ** 2))
so sigma**2 within a np.square.
I reformulated the definition to use the precision and provided a test against scipy.stats.norm.logpdf(terms).
scipy.stats.norm.logpdf(terms)
In the current implementation I (I think) introduced a bug in the loglike_contribution that was like
so sigma**2 within a np.square.
I reformulated the definition to use the precision and provided a test against
scipy.stats.norm.logpdf(terms)
.