Scaling Properties of Parallelized Multicanonical Simulations

We implemented a parallel version of the multicanonical algorithm and applied it to a variety of systems with phase transitions of first and second order. The parallelization relies on independent equilibrium simulations that only communicate when the multicanonical weight function is updated. That way, the Markov chains efficiently sample the temporary distributions allowing for good estimations of consecutive weight functions. The systems investigated range from the well known Ising and Potts spin systems to bead-spring polymers. We estimate the speedup with increasing number of parallel processes. Overall, the parallelization is shown to scale quite well. In the case of multicanonical simulations of the $q$-state Potts model ($q\ge6$) and multimagnetic simulations of the Ising model, the optimal performance is limited due to emerging barriers.