A Reply to the Comment by M. Fuchs et al.Received 17 July 2002DOI:https://doi.org/10.1103/PhysRevLett.90.189702©2003 American Physical Society

The fruitfulness of traditional many-body Green's function theory for calculating the total energy of real systems is demonstrated using the random phase approximation in the Luttinger-Ward formulation. As the first application to a real system, the total energy of H2 is calculated as a function of nuclear separation and compared with the configuration interaction and the local density approximation results. While the local density result is in large error for large separations, the present approach gives satisfactory agreement with the configuration interaction results. The method is promising as an alternative to the quantum Monte Carlo technique.

Total energies of solids are calculated by an ab initio method based on the Green's-function theory. Green's function is constructed from one-body wave functions and eigenvalues obtained in the local-density approximation (LDA) to density-functional theory, and the correlation energy is estimated within the random-phase approximation. The scheme is applied to Na and Si. In both cases, the equilibrium lattice constants are in reasonable agreement with experiments. The role of the exchange-correlation energy in the total-energy curve is discussed in detail in comparison with the LDA.

Ab initio calculations of excited state properties of solids have become feasible with a steady increase in computing power. A suitable method for studying excited-state properties of extended systems is the Green function method which requires knowledge of the self-energy operator. A simple and fruitfull approximation to the self-energy beyond the Hartree-Fock approximation that takes into account screening is the GW approximation (GWA). It has been found to be successful in describing quasiparticle energies in a wide range of systems. Despite its success, there are some theoretical difficulties. The GWA has been found to be inadequate for describing satellite structures and self-consistent GW calculations tend to worsen the good agreement with experiment. Recent development beyond the GWA to improve the satellite description as well as the self-consistency issue will be discussed. As an application of the GWA we consider YH(3). Metal hydrides show a reversible metal-insulator transition in the visible range, making them attractive for optical switch. Density-functional calculations give incorrectly a metallic state unless we assume a complicated structure but GW calculations suggests that YH(3) is a normal insulator.