Wolff algorithm

Summary

The Wolff algorithm is a single-cluster Monte Carlo method for spin models such as the Ising model. It grows one cluster of aligned spins from a random seed, using the same bond probability as the Swendsen–Wang algorithm, and flips the whole cluster. Every proposed flip is accepted, and large clusters are flipped more often because a random seed is more likely to land in them. Near the critical point it reduces critical slowing down even more than Swendsen–Wang in many models.