Support for discrete variables in Hamiltonian Monte Carlo sampler

[zoj613]

Ive read in the documentation that this library only supports continuous variables. Any thoughts on adding support for loglikelihood functions that have a mixture of continuous and discrete parameters? The following papers seem to be addressing how to build a sampler in such a case:
Discontinuous Hamiltonian Monte Carlo for discrete parameters and discontinuous likelihoods

Continuous Relaxations for DiscreteHamiltonian Monte Carlo

Auxiliary-variable Exact Hamiltonian MonteCarlo Samplers for Binary Distributions

[eigenfoo]

Hi!

I view discrete variables as outside the scope of this project. Users who want to sample discrete parameters should consider using PyMC3 instead which already supports a bunch of discrete random variables. This project is limited to merely exposing the vanilla HMC/NUTS sampler (that was already in PyMC3) for use with an arbitrary Python callable as a logp/dlogp function.

Apologies if this wasn't clear in the README: I've written this up in the littlemcmc docs, but I should definitely point to that in the README. I'll close this issue now, but thanks for raising this!