In this section we list some advanced options to the PNO program.
These options exist for technical or historical reasons and we do not recommend modifying the default values in general.
The following options are available for the orbital localization:
- Localization method. method can be IBO (intrinsic bond orbitals, default), PM (Pipek-Mezey), or NBO (natural bond orbitals). IBO is strongly recommended, as it is most efficient and stable. The PNO-LMP2-F12 method only supports IBO presently.
- Projector type for generation of intrinsic atomic orbitals. value can be 1 or 2 (default).
- Exponent used in the PM-like localization functional. value can be 2 or 4 (default).
Options for PAO domain selection:
- Boughton-Pulay completeness criterion for selection of primary PAO domains when using PM.
Options for OSV generation:
- Amplitudes to used to generate OSVs. PAO (default) means to use semi-canonical
amplitudes in the PAO domains; CAN means semi-canonical in the full virtual space; OPT means fully optimized
LMP2 amplitudes in the full virtual space.
Options for PNO generation:
- Amplitudes to used to generate PNOs.
OSV (default) means to use semi-canonical (non-iterative) OSV amplitudes; PAO means semi-canonical (non-iterative) PAO
amplitudes; if (OPT) is appended the amplitudes are iteratively optimized (can be expensive with OSV(OPT) and very
expensive with PAO(OPT)!
- (logical). If true, use PNO domain selection criterion also for diagonal pairs. Otherwise OSV domains
are used for diagonal pairs. If PNO_DIAG=true the threshold THROSV only affects the distant pair multipole treatment.
The PNO program devides basis functions to blocks for integral screening.
The following options are available for defining the block sizes.
A smaller block size encourages more efficient integral screening and improves the parallel efficient, while a larger block size improves the performance of matrix operations and reduces the bookkeeping cost.
In addition, the following options control the integral screening thresholds:
- Target blocking size in the AO basis (default 32).
- Target blocking size in the DF basis for the F12 calculations (default 32). The option does not affect PNO-LMP2 calculations.
- Target blocking size in the RI basis for the F12 calculations (default 128).
- Block screening threshold (default 1.d-5).
- Block screening radius (default 4).
The following options controls some technical details of the PNO program.
- (logical) If true use the blocked density fitting routines when possible (default).
- Whether a local cache is used in combination with globalarrays (GA). Default value is 1, i.e., loading from GA is buffered.