Modeling experimental data in a Monte Carlo simulation

A method is presented for modeling the structure of disordered media consistent with a set of experimental observations, such as scattering data. The data are incorporated into a conventional semigrand canonical Monte Carlo simulation by introducing a generalized, polydisperse composition space. This approach improves upon previous reverse Monte Carlo procedures in that thermodynamic consistency is retained. By way of example, the structure of a Lennard-Jones fluid is derived solely from radial distribution data.