Multigrid Monte Carlo methods

This paper is intended to be a tutorial on multigrid Monte Carlo techniques, illustrated with two examples. Path-integral quantum Monte Carlo is seen to take only a finite amount of computer time even as the paths are discretized on infinitesimally small scales. A method for eliminating critical slowing down completely — even for models with discrete degrees of freedom, as in Potts models, or discrete excitations, such as isolated vortices in the XY model — is presented.