Feras Saad

One complexity is that some State methods alter the representation of the dataset, i.e. `incorporate_dim, unincorporate_dim, incorporate_row, unincorporate_row`, and others do not.

After exploring this issue further, I have the following conclusions: - **Sending** the dataset from Engine process to each State subprocess happens on every invocation of a public method in...

``` -rw-rw-r-- 1 fs fs 15M May 9 07:37 20160504-001756-satellites-5-20.engine -rw-rw-r-- 1 fs fs 4.3M May 25 15:54 20160504-001756-satellites-5-20.engine.2 ``` The top file is using the original scheme where each...

But you can avoid all the duplication by propertly implementing a true multinomial(k,n) and all others (bernoulli, binomial, and categorical) degrade to special cases: - bernoulli: multinomial(2,1) - binomial(n): multinomial(2,n)...

It is problematic that the mean of the Cauchy distribution is undefined (and in fact, all higher order moments)? Are there other distribution SPs with this characteristic?

When @Axch and I discussed these examples in the L&W book before the PI meeting, the conclusion was we could not really tell a compelling story for Venstan with toy...

RE: > highlights the limitations of Stan's claim to "handle discrete variables". In the thread, Carpenter clearly says that Stan can in principle handle the discretes, but it will not...

How about https://github.com/probcomp/Venturecxx/issues/525

I think it is a good idea, with relatively little effort we can characterize the sensitivity of CRP to the prior on alpha (in terms of mixing time/accuracy), which is...

Also coincides with a previous use of "generator" in bayeslite (now known as an analysis schema), which users, even those familiar with probabilistic modeling and inference, found as a confusing...