Key points are not available for this paper at this time.
Recent publications by Slater and, particularly, by Hedin and Lundqvist suggest that Hohenberg-Kohn-Sham theory used in conjunction with the local-density approximation and accurate atomic or electron-gas calculations at last make it possible to evaluate the ground-state properties of real solids without appeal to experimental data. Self-consistent Korringa-Kohn-Rostoker calculations of the Fermi surface of copper show that these two local assumptions lead to neck areas about 20% too large. However, a local exchange-correlation potential can be found which gives a copper Fermi surface in excellent agreement with experiment. We suspect that the 20% error in the ab initio methods is due mainly to nonlocal exchange-correlation effects.
Janak et al. (Fri,) studied this question.