A Note on Metropolis–Hastings Kernels for General State Spaces I: Necessary and Sufficient Conditions for a Metropolis–Hastings Kernel to Satisfy Detailed BalanceResearch Paper
The Metropolis–Hastings algorithm (Metropolis et al. 1953; Hastings 1970) is the basic construction of Markov chain Monte Carlo. It turns a target probability distribution , known only up to a constant, into a Markov chain that has as its invariant distribution. Bayesian computation depends on it, and a sampler is usually designed by proving one…