RELIABILITY OF ELECTRONIC ENERGY CALCULATIONS BY A MODIFIED LOCAL-ENERGY METHOD

Abstract

This statistical study of the performance of a modified local-energy method using random selection shows that there is little advantage in using large numbers of electron positions, the quality of the wave functions being a much more significant factor. A relationship is given between the quality of the wave function and the resulting accuracy. Use of as few as 25 sets of electron positions is suggested.A method of avoiding the increase in the calculation time with the size of a system is given and was found to be very accurate in a simple situation.