aemcmc

aemcmc copied to clipboard

Add closed-form posterior for Gamma-Exponential observation model

3

Related to #48 This PR provides a closed-form posterior solution to an Exponential observation model with a Gamma prior. Some part was directly inspired from other rewrites and tests, but...

larryshamalama

Update versions of pre-commit hooks to latest version.

github-actions[bot]

Extend exact posteriors to condition on multiple observations

7

## Description of your problem or feature request It is my understanding that current exact posteriors, e.g. `gamma_poisson_conjugateo`, can only condition on a single observation $Y \sim \text{Poisson}$ rather than...

larryshamalama

Add more conjugates

5

Trying to add more conjugates. will be addition with current commit Uniform -Pareto. @brandonwillard @rlouf Please let me know which of the conjugates should be more prioritized.

xjing76

We can use the same approach we're currently using for deriving Gibbs samplers to instead identify and use proximal envelopes for many non-standard and discrete prior and observed distributions (see...

brandonwillard

Inspired by https://github.com/aesara-devs/aeppl/pull/221. Making sure the documentation for the transformations is up to date will quickly get cumbersome and can easily be automated with custom directives.

rlouf

We need to make sure that our `kanren`-based rewrites (e.g. conjugates) update the RNG `SharedVariable.default_update`s; otherwise, we'll get implicit updates that include the original non-conjugated graphs in the compiled results.

brandonwillard

rlouf

rlouf

rlouf