MONTE-CARLO SIMULATIONS OF 3D SELF-AVOIDING WALKS

We present simulation results for 3D Self-avoiding walks of moderate length (N less-than-or-equal-to 2000), using a recently introduced recursive sampling method. We show that this method is indeed maximally efficient for determining the connectivity constant mu. Combining our data with previous results from exact enumerations and from simulations of very long walks using the pivot algorithm, we obtain very precise estimates of mu and of the critical exponents gamma and nu. We also point out significant deviations from uniformity in a very popular random number generator which was considered safe until now.