Input¶
The GOUDA engine is selected and configured inside an Engine GOUDA block:
Engine GOUDA
... GOUDA keywords ...
EndEngine
General input syntax and AMS driver keywords are documented in the AMS manual.
Method¶
GOUDA calculations require one to specify the method (i.e. Hartree-Fock or DFT functional), and the level of relativistic effects to be included.
XCFunctional selects Hartree–Fock or a density functional. GOUDA uses Hartree–Fock by default. LocalXCBackend normally selects the native cuEST implementation when one is available and uses LibXC otherwise.
XCFunctional string
XCFunctional- Type:
String
- Default value:
HF
- Description:
Exchange-correlation functional. Native cuEST 0.2 names are HF, B3LYP1, B3LYP5, B97, BLYP, M06-L, PBE, PBE0, R2SCAN, SVWN5, B97M-V, LC-wPBE, wB97X, wB97X-V, wB97M-V, LC-wPBEh, CAM-B3LYP, HSE06, M06, and M06-2X. Other AMS LibXC functional names are accepted when their density ingredients and derivatives are supported by the cuEST LibXC integration.
LocalXCBackend [Auto | CuEST | LibXC]
LocalXCBackend- Type:
Multiple Choice
- Default value:
Auto
- Options:
[Auto, CuEST, LibXC]
- Description:
Implementation used for the local XC energy and potential. Auto uses cuEST for native functionals and LibXC otherwise. Exact exchange and nonlocal VV10 contributions continue to use cuEST.
Scalar-relativistic calculations use the scalar ZORA Hamiltonian. Scalar ZORA and effective core potentials cannot be combined in the same calculation.
Relativity
Level [None | Scalar]
End
Relativity- Type:
Block
- Description:
Relativistic Hamiltonian settings. Scalar selects scalar ZORA using one-center neutral-atom Dirac potentials and Gaussian finite nuclei. Explicit scalar ZORA and ECP basis data are mutually exclusive.
Level- Type:
Multiple Choice
- Default value:
None
- Options:
[None, Scalar]
- GUI name:
Relativity (ZORA)
- Description:
Relativistic treatment. None is nonrelativistic; Scalar enables scalar ZORA.
Basis sets¶
The Basis block selects the primary orbital basis, density-fitting basis, and basis used for the superposition-of-atomic-densities (SAD) initial guess. The defaults are def2-tzvp, def2-universal-jkfit, and minao-1, respectively.
Selections at the root of the block apply to all atoms. PerAtomType overrides the root selection for an element, and PerRegion has the highest precedence. Named basis resources are read from $AMSRESOURCES/GOUDA. You can also provide explicit files in Gaussian basis set (GBS) format.
Basis
File string
FitFile string
FitType string
GuessFile string
GuessType string
PerAtomType
File string
FitFile string
FitType string
GuessFile string
GuessType string
Symbol string
Type string
End
PerRegion
File string
FitFile string
FitType string
GuessFile string
GuessType string
Region string
Type string
End
Type string
End
Basis- Type:
Block
- Description:
Primary, density-fitting, and SAD-guess basis sets. Root settings apply to every atom; PerAtomType overrides root settings for an element; PerRegion has highest precedence. Type and File, FitType and FitFile, or GuessType and GuessFile may not both be specified at the same level.
File- Type:
String
- GUI name:
Basis file
- Description:
Explicit primary GBS file. Relative paths are resolved under $AMSRESOURCES/GOUDA. Mutually exclusive with Type.
FitFile- Type:
String
- Description:
Explicit density-fitting GBS file. Relative paths are resolved under $AMSRESOURCES/GOUDA. Mutually exclusive with FitType.
FitType- Type:
String
- Default value:
def2-universal-jkfit
- Description:
Density-fitting basis resource name under $AMSRESOURCES/GOUDA. The reserved value AutoAux generates an uncontracted spherical auxiliary basis from the resolved primary basis on each atom.
GuessFile- Type:
String
- Description:
Explicit SAD-guess GBS file. Relative paths are resolved under $AMSRESOURCES/GOUDA. Mutually exclusive with GuessType.
GuessType- Type:
String
- Default value:
minao-1
- Description:
Basis resource used to construct the SAD initial density.
PerAtomType- Type:
Block
- Recurring:
True
- Description:
Override any primary, fitting, or guess basis selection for all atoms of one element.
File- Type:
String
- GUI name:
Basis file
- Description:
Explicit primary GBS file.
FitFile- Type:
String
- Description:
Explicit density-fitting GBS file.
FitType- Type:
String
- Description:
Density-fitting basis resource name, or AutoAux to generate it from the resolved primary basis.
GuessFile- Type:
String
- Description:
Explicit SAD-guess GBS file.
GuessType- Type:
String
- Description:
SAD-guess basis resource name.
Symbol- Type:
String
- Description:
Element symbol.
Type- Type:
String
- Description:
Primary basis resource name.
PerRegion- Type:
Block
- Recurring:
True
- Description:
Override any primary, fitting, or guess basis selection for atoms in a region. Selected regions may not overlap.
File- Type:
String
- GUI name:
Basis file
- Description:
Explicit primary GBS file.
FitFile- Type:
String
- Description:
Explicit density-fitting GBS file.
FitType- Type:
String
- Description:
Density-fitting basis resource name, or AutoAux to generate it from the resolved primary basis.
GuessFile- Type:
String
- Description:
Explicit SAD-guess GBS file.
GuessType- Type:
String
- Description:
SAD-guess basis resource name.
Region- Type:
String
- Description:
Region name or expression.
Type- Type:
String
- Description:
Primary basis resource name.
Type- Type:
String
- Default value:
def2-tzvp
- Description:
Primary basis resource name under $AMSRESOURCES/GOUDA.
Spin¶
Use an unrestricted calculation for open-shell systems. SpinPolarization is the difference between the numbers of alpha and beta electrons.
Unrestricted Yes/No
Unrestricted- Type:
Bool
- Description:
Perform a spin-unrestricted calculation. False by default, unless SpinPolarization is set to a non-zero value.
SpinPolarization float
SpinPolarization- Type:
Float
- Default value:
0.0
- Description:
Difference between alpha and beta electrons.
SCF, accuracy, and GPU memory¶
The following keywords control SCF convergence, numerical integration, density fitting, and temporary GPU memory. MaxWorkspaceMiB limits temporary device workspace for individual cuEST compute calls; it is not a limit on all GPU memory used by the calculation.
DFFittingAlgorithm [QR | MatrixPower]
DFFittingAlgorithm- Type:
Multiple Choice
- Default value:
QR
- Options:
[QR, MatrixPower]
- GUI name:
DF fitting algorithm
- Description:
Density-fitting algorithm used when building the cuEST DF plan.
DFGradientMemoryPolicy [DeviceCache | HostCache | Full | Blocked]
DFGradientMemoryPolicy- Type:
Multiple Choice
- Default value:
HostCache
- Options:
[DeviceCache, HostCache, Full, Blocked]
- GUI name:
DF gradient memory policy
- Description:
Memory policy for cuEST DF J/K gradient evaluation. HostCache is the default because it uses significantly less GPU memory, at some performance cost. Full and Blocked are accepted as compatibility aliases for DeviceCache and HostCache, respectively.
MaxWorkspaceMiB integer
MaxWorkspaceMiB- Type:
Integer
- Default value:
2000
- GUI name:
Max workspace (MiB)
- Description:
Maximum temporary cuEST device workspace per compute call, in MiB. Applies to XC, DF exchange/gradient, and ECP compute paths.
ThresholdOverlap float
ThresholdOverlap- Type:
Float
- Default value:
1e-12
- Description:
AO-pair screening threshold used to build the AO pair list. AO pairs whose overlap integral is below the threashold will be exluded.
ThresholdCanonical float
ThresholdCanonical- Type:
Float
- Default value:
1e-06
- Description:
Canonical orthogonalization threshold. Eigenvectors of the overlap matrix that have corresponding eigenvalue below the threshold will be discarded.
DIISMaxNvector integer
DIISMaxNvector- Type:
Integer
- Default value:
10
- GUI name:
DIIS max N vector
- Description:
Maximum number of DIIS vectors.
Iterations integer
Iterations- Type:
Integer
- Default value:
100
- Description:
Maximum number of SCF iterations.
Convergence float
Convergence- Type:
Float
- Default value:
1e-06
- Description:
Orbital gradient convergence threshold.
XCGridLevel integer
XCGridLevel- Type:
Integer
- Default value:
3
- GUI name:
XC grid Level
- Description:
Grid level for local XC integration.
NLCGridLevel integer
NLCGridLevel- Type:
Integer
- Default value:
1
- GUI name:
NLC grid level
- Description:
Grid level for nonlocal correlation integration.
Solvation¶
The PCM block enables the polarizable continuum model. Select a solvent from the built-in solvent table or specify its dielectric constant. These two choices are mutually exclusive.
PCM
Epsilon float
Radii # Non-standard block. See details.
...
End
Solvent string
End
PCM- Type:
Block
- Description:
Polarizable continuum model settings. If this block is absent, PCM is disabled.
Epsilon- Type:
Float
- Description:
Dielectric constant of the continuum. Specify either Epsilon or Solvent, but not both. Other PCM defaults currently come from hardcoded cuEST defaults.
Radii- Type:
Non-standard block
- Description:
Optional per-element PCM cavity radii overrides. Each line should contain an element symbol and a radius in Angstrom, for example: ‘N 1.8300’. These values replace the hardcoded defaults for the specified elements only.
Solvent- Type:
String
- Description:
Solvent name or formula from the built-in ADF COSMO solvent table. Specify either Solvent or Epsilon, but not both.
Excitations¶
The Excitations block requests singlet, triplet, or both types of excitations after the ground-state SCF calculation. GOUDA supports TDA and, for pure functionals, full TDDFT. See General for current response-calculation restrictions.
Excitations
Iterations integer
Lowest integer
Multiplicity [Singlet | Triplet | Both]
Tolerance float
Type [TDA | TDDFT]
Vectors integer
End
Excitations- Type:
Block
- Description:
Calculate the lowest singlet-singlet and/or singlet-triplet excitation energies after the ground-state SCF calculation. The current implementation is restricted to closed-shell, integer-occupation, gas-phase calculations with HF, SVWN5, PBE, or PBE0.
Iterations- Type:
Integer
- Default value:
60
- Description:
Maximum number of Davidson iterations.
Lowest- Type:
Integer
- Default value:
10
- Description:
Number of lowest excitation energies to compute.
Multiplicity- Type:
Multiple Choice
- Default value:
Singlet
- Options:
[Singlet, Triplet, Both]
- Description:
Calculate singlet-singlet excitations, singlet-triplet excitations, or both. Triplet electric-dipole transition moments and oscillator strengths are zero without spin-orbit coupling.
Tolerance- Type:
Float
- Default value:
1e-07
- Description:
Residual-norm convergence threshold for the Davidson solver.
Type- Type:
Multiple Choice
- Default value:
TDA
- Options:
[TDA, TDDFT]
- Description:
Use the Tamm-Dancoff approximation or the full Casida TDDFT equations. Full TDDFT is currently available only for pure functionals.
Vectors- Type:
Integer
- Default value:
30
- Description:
Maximum dimension of the Davidson trial-vector subspace. This must be larger than Lowest.
Complete keyword reference¶
See Keywords for every GOUDA input keyword and its default value.