Determination of the chemical potential by the particle insertion method and by its inverse

Abstract

An explanation has been found for the disparity between two widely used and apparently equivalent methods for the determination of the chemical potential by molecular simulation. There is a hitherto unrealised error in the implementation and this is sufficient to explain the discrepancies observed; the possibility of correcting for this source of error is considered. There is no reason to seek further for errors in the underlying statistical mechanical theory. The different patterns of random errors in the two methods arising from their very different sampling distributions are not considered here.