BOND-UPDATING MECHANISM IN CLUSTER MONTE-CARLO CALCULATIONS

We study a cluster Monte Carlo method with an adjustable parameter: th e number of energy levels of a demon mediating the exchange of bond en ergy with the heat bath. The efficiency of the algorithm in the case o f the three-dimensional Ising model is studied as a function of the nu mber of such levels. The optimum is found in the limit of an infinite number of levels, where the method reproduces the Wolff or the Swendse n-Wang algorithm. In this limit the size distribution of flipped clust ers approximates a power law more closely than that for a finite numbe r of energy levels.