probabilistic-programming

NumPyro now has several excellent introductory examples with no direct counterparts in Pyro. Porting one of these to Pyro would be a great way for someone to simultaneously learn more about Bayesian data analysis and make a valuable open source contribution.

If you are reading this and want to give one of them a try, please leave a comment here so that other peo

Poisson Binomial distributions (and the more general Poisson Multinomial) crop up all over the place in the social sciences, particularly ecological inference, and linguistics. Having such a distribution in tensorflow would be very valuable to those communities. Efficient implementations exist in R: https://cran.r-project.org/web/packages/PoissonBinomial/PoissonBinomial.pdf

And descriptions for

Ankit Shah and I are trying to use Gen to support a project and would love the addition of a dirichlet distribution

Dear Numpyro developers,

Please develop Euler Maruyama features in numpyro similar to features found in PyMC.

Thanks alot.

The current example on MDN from Edward tutorials needs small modifications to run on edward2. Documentation covering these modifications will be appreciated.

Hi,

Looks like there is support for lots of common distribution. There are a handful of other distributions which are not presently supported but could (fingers crossed) be easily implemented. Looking at [Stan's Function Reference] I see...

• Beta Binomial
• [Chi-Square](https://mc-stan.org/docs/2

There are a variety of interesting optimisations that can be performed on kernels of the form

k(x, z) = w_1 * k_1(x, z) + w_2 * k_2(x, z) + ... + w_L k_L(x, z)

A naive recursive implementation in terms of the current Sum and Scaled kernels hides opportunities for parallelism in the computation of each term, and the summation over terms.

Notable examples of kernels with th

Rather than trying to rebuild all functionality from Distributions.jl, we're first focusing on reimplementing logdensity (logpdf in Distributions), and delegating most other functions to the current Distributions implementations.

So for example, we have

distproxy(d::Normal{(:μ, :σ)}) = Dists.Normal(d.μ, d.σ)

This makes some functions in Distributions.jl available through

Aeppl is a very expressive PPL written in aesara. Aehmc implements a NUTS sampler identical to blackjax's and it would be interesting for me to have a benchmark. It would also be a nice example for this PPL.

We need to add a static analysis tool that triggers on each PR and provides a report, ideally flake8 style where we can configure its behaviour and have the action fail until the PR respects all imposed rules. The "configure behaviour" bit is important since we might have some standards that are not in line with static analysis preferences (e.g. there's certain bits where we use exec and `ev