62.4.1 Atom-based method

In the atom-based method for truncating the active subsystem basis set, only basis functions that are centered on atoms near the active subsystem are included in the active subsystem basis set. This method is described more detail in:

T. A. Barnes, J. D. Goodpaster, F. R. Manby, and T. F. Miller III, J. Chem. Phys. 139, 024103 (2013)

This method can be invoked using the BATOMS card:

`BATOMS,atom1,atom2,atom3...`where

`atom1`is the name of an atom from the geometry specification, and the atoms are listed in the order in which they appear in the geometry specification. The calculation on the active subsystem is performed using only basis functions centered on atoms listed by the`BATOMS`card. The minimum input to run using the`BATOMS`card is:`{ks}``{embed,proj;atoms,atom1;batoms,atom1}``{hf}``{ccsd(t)}`

Alternatively to using the `BATOMS` card, the user can instead specify the `BASECUT` option:

`BASECUT=value``{ks}``{embed,proj,basecut=value;atoms,atom1}``{hf}``{ccsd(t)}`where

`value`is a distance in Angstroms. The calculation on the active subsystem is performed using only basis functions centered on atoms within a distance`BASECUT`of an atom specified by the`ATOMS`card.

When employing basis set truncation, the embedding potential produced by the MOs in the frozen subsystem is only partially calculated using the numerically exact projection-based embedding method described in section 62.
For MOs that have negligible Mulliken population on the basis functions that are employed in the calculation on the active subsystem, the non-additive kinetic energy contribution to embedding potential is calculated using the approximate Thomas-Fermi (TF) functional.
Specifically, the TF functional is used for any MO in the frozen subsystem for which the sum of the absolute Mulliken populations on the atoms included by the `BATOMS` card or `BASECUT` option is less than the option `TAU` (default = 0.1).
For closed-shell MOs, `TAU` should typically be in the range 0.0-2.0.
Setting `TAU` to a value that is too close to zero can cause numerical artifacts that prevent convergence of the energy with respect to the value of the `MU` option (see subsection 62.1); as a result, it is recommended to perform all calculations with multiple values of `MU` to ensure that these numerical artifacts are not present.

molpro@molpro.net 2019-03-25