Options

### Frequency-Dependent Calculations

#### RdFreq

Perform frequency-dependent (dynamic) CPHF, reading in the incident light frequency for the electromagnetic field perturbation. The desired frequency must be provided in the input stream. The default units for this value are Hartrees. Other units may be specified by including a suffix, one of cm (cm^{-1}) and nm (wavelength). This option is relevant for Freq and Polar jobs. It is the default for Freq=ROA.

#### InputFreq

Read in perturbation frequencies rather than take them from the checkpoint file when doing Geom=AllCheck.

#### Static

Automatically include the static perturbations when doing dynamic ones. This is the default except for Polar=OptRot and Freq=ROA. NoStatic says not to perform static perturbations in combination with dynamic via RdFreq.

### Specifying the Integration Grid

#### Grid=grid

Specify the integration grid for the CPHF portion of the calculation. The syntax is the same as for the Int=Grid option. The argument to this option may be a grid keyword (Fine, UltraFine, and so on) or a specific grid.

The default grid is UltraFine. In this case, the default grid for the CPHF is SG1. When a specific grid is specified to the Int=Grid option, then that grid is also used for the CPHF. Finally, be aware that Fine is used in the CPHF as the default integration grid for a few DFT jobs including Polar=OptRot, Freq=Anharmonic and Freq=NNROA.

See the discussion of Int=Grid for full details on grid specification.

#### OneStep

DFT nuclear 2^{nd} derivatives (ground- or excited-state) should use a grid for CPHF and CPTD one step smaller than the rest of the calculation.

#### TwoStep

DFT nuclear 2^{nd} derivatives (ground- or excited-state) should use a grid for CPHF and CPTD two steps smaller than the rest of the calculation. This is the default.

#### TauOneStep

DFT 2^{nd} derivatives (ground- or excited-state) should use a grid for CPHF and CPTD one step smaller than the rest of the calculation for tau-functionals but two steps smaller for GGAs.

#### PSCFOneStep

TDDFT 2^{nd} derivatives should use a grid for CPHF and CPKS one step smaller than the rest of the calculation, but ground-state frequencies continue to default to 2 steps.

#### PSCFTauOneStep

TDDFT 2^{nd} derivatives should use a grid for CPHF and CPKS one step smaller than the rest of the calculation when using tau functionals, but excited-state frequencies with GGAs and all ground-state frequencies continue to default to 2 steps.

The step options are primarily useful in Default.Route

### Procedure-Related Options

#### Conver=N

Set the CPHF convergence criterion to 10^{-N}. N>=10 defaults to CPHF=Separate for the ground-state CPHF/CPKS. N>=9 defaults to CPHF=Separate for CPCIS/CPTD.

#### RecursiveDIIS

Solve reduced equations using recursive DIIS. This is the default when the number of right-hand sides is at least twice the dimension of the reduced matrix and the dimension of the reduced matrix is large (occurs only for ONIOM(MO:MM) using electronic embedding), or the limit set by MaxInv is exceeded. Otherwise, the default is NoRecursiveDIIS, which says to invert the reduced A-matrix.

#### MaxInv=N

Specifies the largest reduced space for in-core inversion during simultaneous solution (up to dimension N). Larger reduced problems are solved by a second level of DIIS. The default is 5000.

#### Simultaneous

Use one expansion space for all variables. This is faster than using separate spaces, but is slightly less accurate. This is the default except when multiple frequencies are specified with RdFreq (see above).

#### Separate

Use a separate expansion space for each variable in the CPHF (the opposite of Simultaneous). This is the default and only choice when multiple frequencies are specified with RdFreq.

#### AO

Solve CPHF in the atomic orbital basis [Stevens63, Osamura81, Osamura82, Pulay83]. This is the default.

#### MO

Solve in the molecular orbital basis.

#### Canonical

Canonical CPHF, the default.

#### MOD

Use MOD orbital derivatives for SAC-CI gradients (which uses configuration selection).

Related Keywords

### Frequency-Dependent Calculations

#### RdFreq

Perform frequency-dependent (dynamic) CPHF, reading in the incident light frequency for the electromagnetic field perturbation. The desired frequency must be provided in the input stream. The default units for this value are Hartrees. Other units may be specified by including a suffix, one of cm (cm^{-1}) and nm (wavelength). This option is relevant for Freq and Polar jobs. It is the default for Freq=ROA.

#### InputFreq

Read in perturbation frequencies rather than take them from the checkpoint file when doing Geom=AllCheck.

#### Static

Automatically include the static perturbations when doing dynamic ones. This is the default except for Polar=OptRot and Freq=ROA. NoStatic says not to perform static perturbations in combination with dynamic via RdFreq.

### Specifying the Integration Grid

#### Grid=grid

Specify the integration grid for the CPHF portion of the calculation. The syntax is the same as for the Int=Grid option. The argument to this option may be a grid keyword (Fine, UltraFine, and so on) or a specific grid.

The default grid is UltraFine. In this case, the default grid for the CPHF is SG1. When a specific grid is specified to the Int=Grid option, then that grid is also used for the CPHF. Finally, be aware that Fine is used in the CPHF as the default integration grid for a few DFT jobs including Polar=OptRot, Freq=Anharmonic and Freq=NNROA.

See the discussion of Int=Grid for full details on grid specification.

#### OneStep

DFT nuclear 2^{nd} derivatives (ground- or excited-state) should use a grid for CPHF and CPTD one step smaller than the rest of the calculation.

#### TwoStep

DFT nuclear 2^{nd} derivatives (ground- or excited-state) should use a grid for CPHF and CPTD two steps smaller than the rest of the calculation. This is the default.

#### TauOneStep

DFT 2^{nd} derivatives (ground- or excited-state) should use a grid for CPHF and CPTD one step smaller than the rest of the calculation for tau-functionals but two steps smaller for GGAs.

#### PSCFOneStep

TDDFT 2^{nd} derivatives should use a grid for CPHF and CPKS one step smaller than the rest of the calculation, but ground-state frequencies continue to default to 2 steps.

#### PSCFTauOneStep

TDDFT 2^{nd} derivatives should use a grid for CPHF and CPKS one step smaller than the rest of the calculation when using tau functionals, but excited-state frequencies with GGAs and all ground-state frequencies continue to default to 2 steps.

The step options are primarily useful in Default.Route

### Procedure-Related Options

#### Conver=N

Set the CPHF convergence criterion to 10^{-N}. N>=10 defaults to CPHF=Separate for the ground-state CPHF/CPKS. N>=9 defaults to CPHF=Separate for CPCIS/CPTD.

#### RecursiveDIIS

Solve reduced equations using recursive DIIS. This is the default when the number of right-hand sides is at least twice the dimension of the reduced matrix and the dimension of the reduced matrix is large (occurs only for ONIOM(MO:MM) using electronic embedding), or the limit set by MaxInv is exceeded. Otherwise, the default is NoRecursiveDIIS, which says to invert the reduced A-matrix.

#### MaxInv=N

Specifies the largest reduced space for in-core inversion during simultaneous solution (up to dimension N). Larger reduced problems are solved by a second level of DIIS. The default is 5000.

#### Simultaneous

Use one expansion space for all variables. This is faster than using separate spaces, but is slightly less accurate. This is the default except when multiple frequencies are specified with RdFreq (see above).

#### Separate

Use a separate expansion space for each variable in the CPHF (the opposite of Simultaneous). This is the default and only choice when multiple frequencies are specified with RdFreq.

#### AO

Solve CPHF in the atomic orbital basis [Stevens63, Osamura81, Osamura82, Pulay83]. This is the default.

#### MO

Solve in the molecular orbital basis.

#### Canonical

Canonical CPHF, the default.

#### MOD

Use MOD orbital derivatives for SAC-CI gradients (which uses configuration selection).

Last updated on: 05 January 2017. [G16 Rev. C.01]