There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
<p class="first" id="d9096594e156">In the present article, we show how to formulate
the partially contracted n-electron
valence second-order perturbation theory (NEVPT2) energies in the atomic and active
molecular orbital basis by employing the Laplace transformation of orbital-energy
denominators (OEDs). As atomic-orbital (AO) basis functions are inherently localized
and the number of active orbitals is comparatively small, our formulation is particularly
suited for a linearly scaling NEVPT2 implementation. In our formulation, there are
two kinds of NEVPT2 energy contributions, which differ in the number of active orbitals
in the two-electron integrals involved. Those involving integrals with either no or
a single active orbital can be formulated completely in the AO basis as single-reference
second-order Møller–Plesset perturbation theory and benefit from sparse active pseudo-density
matrices—particularly if the active molecular orbitals are localized only in parts
of a molecule. Conversely, energy contributions involving integrals with either two
or three active orbitals can be obtained from Coulomb and exchange matrices generalized
for pairs of active orbitals. Moreover, we demonstrate that Laplace-transformed partially
contracted NEVPT2 is nothing less than time-dependent NEVPT2 [A. Y. Sokolov and G.
K.-L. Chan, J. Chem. Phys.
<b>144</b>, 064102 (2016)] iff the all-active intermediates are computed with the
internal-contraction
approximation. Furthermore, we show that for multi-reference perturbation theories
it is particularly challenging to find optimal parameters of the numerical Laplace
transformation as the fit range may vary among the 8 different OEDs by many orders
of magnitude. Selecting the number of quadrature points for each OED separately according
to an accuracy-based criterion allows us to control the errors in the NEVPT2 energies
reliably.
</p>