Reply to Lindgren and Salomonson

Abstract

In the accompanying Comment [Phys. Rev. A {\bf 67}, 056501 (2003)], I. Lindgren and S. Salomonson claim to prove for the Kohn-Sham kinetic energy functional of ground state electron density that a Fr\"echet functional derivative exists, equivalent to a multiplicative local potential function. If true, this result would imply an exact Thomas-Fermi theory for ground states of noninteracting electrons. However, such a theory is not consistent with the exclusion principle for more than one electron of each spin. The simplest counterexample is the lowest triplet state of a noninteracting two-electron atom. If only the total electron density were normalized, as in Thomas-Fermi theory, the lowest state would collapse into a doubly-occupied $1s$ spin-orbital. Two independent parameters $\epsilon_{1s}$ and $\epsilon_{2s}$ are required to maintain independent subshell normalization. The argument presented by these authors is discussed in the light of this unphysical implication.