For intermolecular perturbation theories in which it is assumed that t
he unperturbed wave function of the composite system is a product of t
he unperturbed wave functions of its components, and which satisfy one
general constraint, we derive two renormalized interaction energy exp
ressions which are more accurate than the perturbation expansions, whe
n all are evaluated to comparable order. This is accomplished by focus
ing on the parameter lambda in terms of which the perturbation expansi
ons are derived rather than on the potential of interaction between co
mponents. In the derivation of each renormalized energy formula, we di
scard zeroth- through infinite-order terms which do not contribute to
the interaction energy when the interaction is turned on fully, i.e.,
when lambda = 1. The first renormalized interaction energy when lambda
= 1 is identical in form to the interaction energy in the symmetrized
Rayleigh-Schrodinger (SRS) theory, but not in interpretation. The wav
e function appearing in the renormalized energy cannot generally be th
at assumed in the SRS theory, and the renormalized energy to zeroth or
der in lambda is not zero. The latter is not surprising because we dis
carded a zeroth-order term in the derivation. The second renormalized
interaction energy formula is derived from the first by using the same
set of assumptions and arguments that were used in deriving the first
. We expect it to be more accurate than the first, which is expected t
o be more accurate than the sum of the perturbation energies, all eval
uated to comparable order. These expectations are supported by the res
ults of calculations on LiH using two perturbation theories, the polar
ization approximation and the Amos-Musher theory. The first-order wave
functions for both were calculated in the configuration interaction (
CI) approximation; then the interaction energies were calculated by su
mming the perturbation energies through third order and by evaluating
the renormalized energy expressions. The perturbation results are comp
ared to interaction energies calculated by full CI with the same basis
set. As important as the formulas is the light our analysis throws on
the meaning of order in intermolecular perturbation theory. (C) 1996
John Wiley & Sons, Inc.