Tony Morris

@dibblego @dcsobral point(x).get = x, always. The simplest possible distribution. Don't see how Distribution[X] can be Identity[X], though

@dave_griffith @dcsobral I might be missing it. case class Distribution[A](a: get). Where is the difference?

@dibblego @dcsobral Many distributions not of that form. size(Distribution[X]) >> size(X). More similar to Hom[X] than Identity[X]

@dave_griffith @dcsobral These are operations on Distribution. The structure is the same right?

@dibblego @dcsobral Except "get" is in general nondeterministic, and thus should really be modeled differently (which break comonad).

@dave_griffith @dcsobral Which laws break? https://github.com/tonymorris/probability-monad/blob/master/src/main/scala/probability-monad/Distribution.scala#L441 …

@dibblego @dave_griffith I don't think that pure makes sense.

@dcsobral @dibblego pure fits all laws. Might just be odd because they are deterministic random variables, which confuses some.