Kernel-based MCMC post-processing algorithms

Kernax is a small package that implements kernel-based post-processing and subsampling algorithms for MCMC output. It currently provides three algorithms:

• The vanilla Stein thinning algorithm, proposed by M. Riabiz et al. in Optimal thinning of MCMC output
• The regularized Stein thinning algorithm, proposed by C. Bénard et al. in Kernel Stein Discrepancy thinning: a theoretical perspective of pathologies and a practical fix with regularization.
• A greedy maximum mean discrepancy (MMD) subsampling algorithm (see, e.g., Optimal quantisation of probability measures using maximum mean discrepancy).