Cluster-flipping Monte Carlo algorithm and correlations in "good" random number generators
Selke W., Talapov A.L., Shchur L.N.
Monte Carlo .simulations of the two-dimensional Ising model at criticality are performed using the cluster-flipping Wolff algorithm. Correlations found recently in a widely used "g^od" random number generator are observed to depend strongly and systematically on the size of the Monte Carlo system. Rather long-range correlations seem to cause trouble, in growing clusters of specific sise in a biased way.