Brandon T. Willard
Brandon T. Willard
Aside from constructing complete sampler loops per https://github.com/aesara-devs/aemcmc/issues/80, we need to set some standard HMC/NUTS defaults (e.g. mass matrix adaptation) and clarify the settings/options interface to the sampler construction process.
After #45, we need to add a scale-mixture-expansion step to the sampler-generating process. We can start by implementing the Polya-gamma expansion for negative-binomials and Bernoulli random variables. This will work...
We need to implement more conjugate relationships and posteriors.
In #86, I updated the step size dtypes to int64, but I imagine it would be better to perform that conversion somewhere within AeHMC; however, I don't know the best...
We can simplify our IR by using rewrites to derive the exact distributions of location-scale/affine transformations of basic random variables. For example, the following produces an unnecessarily complicated `MeasurableElemwiseTransform` instead...
This PR adds a custom `SphinxDirective` that dynamically generates cross-reference links to `RandomVariable`s in Aesara that are supported by `logprob` (i.e. that have a registered dispatch). See [here](https://aeppl--221.org.readthedocs.build/en/221/api/distributions.html) for the...
The result should be a geometric random variable.
Currently, `Scan` log-probability support only handles cases in which the `MeasurableVariable` is created inside the body/step function of the `Scan`, and not when the body/step function simply references a `MeasurableVariable`...
This PR adds support for general power transforms and replaces #184. It uses a positive support in cases of ambiguity. In its current form, it appears to work as a...