Cluster Monte Carlo algorithm for the quantum rotor model

We propose a highly efficient "worm"-like cluster Monte Carlo algorithm for the quantum rotor model in the link-current representation. We explicitly prove detailed balance for the algorithm even in the presence of disorder. For the pure quantum rotor model with mu=0, the algorithm yields high- precision estimates for the critical point K-c=0.333 05(5) and the correlation length exponent nu=0.670(3). For the disordered case, mu=1/2+/-1/2, we find nu=1.15(10).