Closed mjo22 closed 1 year ago
We should not tie ourselves to any one framework or add something like a PPL as a dependency. In order to use a PPL, one should, say, sample from a numpyro Distribution and use this to build a model.
There might at some point be an argument for building numpyro support by subclassing Distributions.
cryojax
should integrate with the JAX bayesian inference ecosystem. For example, it would be great to have support for probabilistic programming libraries such as pymc and numpyro.For example, tinygp adds
numpyro
support by subclassing aDistribution
: https://github.com/dfm/tinygp/blob/main/src/tinygp/numpyro_support.py.Also,
pymc
has instructions for wrapping JAX functions: https://www.pymc.io/projects/examples/en/latest/case_studies/wrapping_jax_function.html