Long-rangeforcesandtheEwaldsum.docVIP

Long-Range Forces Truncating the potential When presenting hard-sphere molecular dynamics we introduced the use of periodic boundaries as a means to model a bulk phase without walls. We examined the minimum-image convention, but only in the context of identifying collision pairs. Because the hard-sphere potential is very short-ranged, we were able to avoid the issue of how to treat interactions with atoms that do not lie in the minimum-image shell. But more realistic potentials are of sufficiently long range that interactions with second-nearest and even more distant images cannot be neglected. However, it is not feasible, and fortunately it is not necessary, to sum the interactions an atom has with all these far-away replicas of the atom nearest to it. In fact in many cases it is disadvantageous to include interactions even with some nearest-image atoms. The reason is described in Illustration 1. The yellow square depicts the nearest-image region about the purple atom at its center. Note that the green atom to the right of the central atom is a nearest-image atom, while its (green) image in the box to the left is more distant, and the interaction of the central atom with it is neglected. However, both this second-image atom and the black atom (representing an entirely different atom) in the corner of the nearest-image region are equally distant from the central atom. So if all nearest-image interactions with the central atom were included, there would be an inequity, in that two equally distant atoms would be contributing differently to the total interaction. This situation complicates the application of any adjustment that corrects for all the neglected distant-image interactions. Consequently it is better to neglect all such unequal interactions, even if they are with nearest images. The problem arises only with interactions more distant than half the length of the simulation box, so the interatomic potential should be truncated at this distance, or