1 paper
Yoshiyuki Kabashima
Standard Markov chain Monte Carlo methods update variables one at a time to satisfy detailed balance, which prevents them from fully exploiting massively parallel hardware such as…