Comparison of FSGO, Hartree–Fock–Roothaan and pseudopotential calculations for Li2

Abstract

Plots of the pseudopotentials, valence orbitals and their differences for FSGO, large-basis Hartree–Fock–Roothaan and pseudopotential calculations on Li₂ are presented. In agreement with earlier arguments, it is concluded that the success of the FSGO procedure depends upon a balanced description of core and valence orbitals. If the description of the core is improved compared to the simple FSGO single-Gaussian characterization, the use of simple FSGO for the valence density is not well attuned to the (more structured) core, and poor results are obtained. Conversely, a proper FSGO pseudopotential yields highly satisfactory FSGO valence orbitals, but is not successful if applied in a large-basis ab initio calculation.