The program is invoked using
or, for the explicitly correlated variants
Available options are described in later sections of this manual.
An example input file can be found in
In this section we describe the parameters most relevant to the accuracy and performance of the PNO program.
We note that our team has carefully selected the default options through benchmark calculations, and the options only need to be modified for special cases.
Options for PAO/OSV/PNO generation:
- Charge threshold for selection of primary PAO domains when using IBOs or NBOs (default 0.2).
- Domain extension using connectivity. value corresponds to the number of bonds by which the primary domains are extended. The default is IEXT=2.
- Domain extension using distance. value is the radius in from any atom in the primary domain. IEXT and REXT can be combined. If both are given all atoms are included that are selected by one or the other criterion. The default is REXT=5.
- OSV selection threshold based on occupation numbers (default 1.d-9).
- PNO selection threshold based on the energy criterion (default 0.998). If thresh is smaller than 0.1 it refers to the natural occupation numbers, otherwise it refers to the fraction of OSV pair energies to be recovered. It is also possible to specify 3 values for strong+close, weak, and distant pairs as THRPNO=[thrstrong, thrweak, thrdist].
- PNO selection threshold based on occupation numbers (default 1.d-8). If THRPNO is also given, a PNO is selected only when both criteria are fulfilled.
Options for F12 calculations:
- (logical). If true use the local and parallel F12 energy routines (default .true.). Otherwise F12 routines relying on hard disk I/O (and therefore shall be run on a single computer node) will be used. This option shall only be set false if problems (such as insufficient memory available) arise from the parallel F12 routines.
- If BB_F12_PNO=0, both OSV- and PNO-LMP2-F12 energy will be computed; If BB_F12_PNO=1, only PNO-LMP2-F12 energy will be computed; If BB_F12_PNO=2, only OSV-LMP2-F12 correction will be computed and the resulting energy will be set to PNO-LMP2 energy plus OSV-LMP2-F12 correction (default 1).
- LMP2 pair energy threshold for selecting pairs for which F12 corrections are computed. The default value is 1.d-4.
Options for domain approximations and orbital screening:
- Disables local fitting (default is LOCFIT=1).
- Fitting domain extension using connectivity. The default is 3.
- Fitting domain extension using distances in . The default is 7. If both IDFDOM and RDFDOM are given, the DF functions at a center will be included when either criterion is fulfilled.
- Disables local RI (default is LOCRI=1).
- RI domain extension using connectivity. The default is 3.
- RI domain extension using distances in . The default is 7. If both IRIDOM and RRIDOM are given, the RI basis functions at a center will be included when either criterion is fulfilled.
- In the PNO program LMOs are truncated if the square sum of the coefficients at one center is smaller than threshold (default 1.d-6). The remaining LMO coefficients are fitted to the original LMO.
- In the PNO program PAOs at one center are truncated when none of the PAOs at the center has a square sum of coefficients greater than threshold (default 1.d-6). The remaining PAO coefficients are fitted to the original PAOs.
- The MP2 pair energy threshold in determining the LMO domains
used in the F12 strong orthogonality projector (default 1.d-4). This option only applies to the selection of the valence occupied orbitals in the LMO domains.
- ICOREDOM_F12=value, RCOREDOM_F12=value
- The connectivity and distance (in ) criteria in determining the LMO domains
used in the F12 strong orthogonality projector (defaults are 2 and 5.0, respectively). This option only applies to the selection of the core orbitals in the LMO domains.
Options for symmetry:
- Use point group symmetry (default is LOCSYM=0).
Note that the PNO-LMP2-F12 program does not support point group symmetry.