chebee7i

Hi, it would be nice if we could specify a maximum length for the queue. If you push onto a full queue, it pops from the end, never increasing in...

Trouble building conda package.

1

**Describe the bug** `requirements.txt` lists `setuptools` as a requirement. When I prepared a conda `meta.yml` to build this package, it was failing, complaining about being able to find `setuptools_scm`. In...

How to specify that headers must be regenerated.

1

I've have a GitHub app which uses a jwt that can expire. Ultimately, I generate an installation access token that can also expire and I am trying to use `ghapi`...

More than marginals?

1

Is it possible to obtain various joint distributions instead of marginals? For example, in the Monty Hall example, I'd want something like: `g.jq(guest_door='A')` to display the joint distribution `P(monty_door, prize_door...

Functionality from pyentropy

1

pyentropy seems to have some useful inference functionality. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2647335/ https://github.com/robince/pyentropy Unsure if it is actively developed anymore.

Pickle-able Distributions

1

Make distributions pickle-able.

DAG based distributions

3

The current method of storing all nonzero outcomes simply won't scale. We'll need to implement a distribution based on Bayesian networks, queries via belief propagation.

Provide a function to sample distributions uniformly from a Bayesian network.

In most cases, we can go back and forth between linear probabilities and log probabilities without loss. However, there are cases when going to linear probabilities results in a distribution...

See if there is any way to speed up `LinearOperations` and `LogOperations`. Code ends up spending up a lot of time here.