You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Oct 5, 2018. It is now read-only.
Thanks for the package! It has a lot of really neat features.
My one nitpick is that TransformLogLikelihood is not really that accurate of a name. As I understand it, it accounts for the Jacobian term when doing the change-of-variables. But the Jacobian term is due to the fact that it's a density; likelihoods by themselves are parametrisation invariant (e.g. maximum likelihood estimates won't change).
I would suggest TransformLogDensity (since you don't need a likelihood at all, e.g. you could just be sampling from an arbitrary distribution).
The text was updated successfully, but these errors were encountered:
Thanks for the package! It has a lot of really neat features.
My one nitpick is that
TransformLogLikelihood
is not really that accurate of a name. As I understand it, it accounts for the Jacobian term when doing the change-of-variables. But the Jacobian term is due to the fact that it's a density; likelihoods by themselves are parametrisation invariant (e.g. maximum likelihood estimates won't change).I would suggest
TransformLogDensity
(since you don't need a likelihood at all, e.g. you could just be sampling from an arbitrary distribution).The text was updated successfully, but these errors were encountered: