The atomic pair distribution function (PDF) represents the histogram of interatomic distances between the atoms comprising a material. Itis obtained as the Fourier transform of total scattering measurements made using X-rays, neutrons or electrons. Models for the PDF based onsimulation of the atomic positions in nanoparticles are a topic of recent interest. Such PDF models may have stochastic elements to account for defects in the crystal structure, and for size and shape variations in the particles. Fitting the model PDF to measured dataunder the criterion of least squares is a global optimization problem. In this talk I discuss the utility of the differential evolution algorithm, a heuristic search method in many respects similar to agenetic algorithm, for fitting PDF models to measured data. Differential evolution will be described in some detail, along with the publicly available software package DEoptim implementing the method in the R language and environment for statistical computing.