This project is a code implementation of a Gibbs sampler for mixed support distributions, where th component samplers are Langevin samplers. The sampler is demonstrated on a simple Gaussian Mixture Model (GMM) parameter inference. For the continuous parameters. For the continuous variables, we use the Stochastic Gradient Hamiltonian Monte Carlo (SGHMC) and for the discrete variables, the Discrete Metropolis Adjusted Langevin sampler (DMALA) is used.
Although the sampler is demonstrated on a simple GMM, it can be easily extended to more complex models such as spike and slab models for variable selection.
