Geometry optimization of atomic microclusters using inverse-power distance coordinates

Using redundant inverse-power distance coordinates we present a series of optimizations on randomly generated argon clusters containing 12, 24, and 48 argon atoms. Our results clearly show that inverse distance coordinates perform significantly better for optimizing the geometries of atomic clusters than either Cartesian or natural/delocalized internal coordinates, The potential utility of these coordinates for global optimization is discussed. (C) 1996 American Institute of Physics.