Computation of a chemical potential using a residence weight algorithm.

The test particle insertion method and its generalization to biased insertion schemes allows the computation of chemical potentials in fluids. Even though these techniques can be implemented in dense systems, the convergence of the estimated value for the chemical potential must be carefully checked and additional simulations are actually required. We propose to compute the chemical potential using a residence weight algorithm. With this algorithm, it is shown that, for a given amount of computer time, the degree of convergence towards the exact chemical potential correlates with the mean rate for accepting the trial particle insertions or deletions. The residence weight algorithm thus offers a reliable built-in tool for diagnosing the numerical convergence.