MIXED_CDFT

References: Holmberg2017, Holmberg2018

Calculate properties involving multiple constrained states. Each repetition of the FORCE_EVAL section defines a new CDFT state that is included in the simulation. The DFT&QS&CDFT section must be active in each FORCE_EVAL and it must be consistently defined. When the keyword MIXED&NGROUPS is set to a value 2 or larger, the CDFT states are solved in parallel, whereas when it is set to 1, the states are solved in serial. During MD, the system can be translated using only two of the CDFT states, which are selected with the keyword FORCE_STATES. The forces are determined by the linear combination F= lambda F1 + (1-lambda) F2. [Edit on GitHub]

Subsections

Keywords

Keyword descriptions

BLOCK_DIAGONALIZE

Type: logical
Default: F
Lone keyword: T
Usage: BLOCK_DIAGONALIZE

Description: Block diagonalize the CDFT Hamiltonian. Control settings should be given in section &BLOCK_DIAGONALIZE. All requested electronic couplings are printed out after block diagonalization. When CDFT-CI and block diagonalization are both requested, the CI calculation is performed using the block diagonalized Hamiltonian.

[Edit on GitHub]

CI

Type: logical
Default: F
Aliases: CONFIGURATION_INTERACTION
Lone keyword: T
Usage: CI

Description: Perform a CDFT configuration interaction calculation (CDFT-CI). The CI vector is expanded in the basis of the CDFT states. Diagonalizes the nonorthogonal diabatic CDFT Hamiltonian. The energies and expansion coefficients of the CDFT-CI states are outputted. Keyword COUPLING must be active to use this feature.

[Edit on GitHub]

COUPLING

Type: integer
Default: -1
Usage: COUPLING

Description: Parameter determining how often the CDFT electronic coupling element is calculated. Use a negative number to disable and 0 means every step. By default, the coupling is calculated by rotating the CDFT states to eigenstates of the weight function matrix when a single constraint is active and the constraint definitions are identical in both CDFT states. Otherwise uses Lowdin orthogonalization. For more than two CDFT states, the couplings are not computed pairwise and the values might deviate from values computed separately for each unique CDFT state pair.

[Edit on GitHub]

DLB

Type: logical
Default: F
Lone keyword: T
Usage: DLB

Description: Controls the activation of dynamic load balancing during a mixed CDFT calculation. Requires Gaussian cavity confinement. Works only in conjunction with keyword PARALLEL_BUILD.

Mentions:Constrained DFT

[Edit on GitHub]

EPS_OCCUPIED

Type: real
Default: 1.00000000E-006
Usage: EPS_OCCUPIED

Description: Threshold for determining which molecular orbitals are considered occupied when fractional and/or empty orbitals are employed. Can and usually should be less than the threshold EPS_FERMI_DIRAC defined in section SCF&SMEAR. Note that the number occupied MOs should be constant in each CDFT state, since the CDFT coupling is only defined between states in the same spin state. Fractionally occupied MOs might exhibit linear dependencies and a singular value decomposition (EPS_SVD) can be used for removing these.

[Edit on GitHub]

EPS_SVD

Type: real
Default: 0.00000000E+000
Usage: EPS_SVD

Description: Determines the matrix inversion solver needed to compute the coupling. Default value implies LU decomposition, while values between 0.0 and 1.0 imply SVD decomposition. For SVD, the value acts as a threshold for screening singular values so that only values above it are included in the matrix pseudoinverse.

[Edit on GitHub]

FORCE_STATES

Type: integer[2]
Default: 1 2
Usage: FORCE_STATES 1 1

Description: Defines the CDFT states used to translate the system.

Mentions:Constrained DFT

[Edit on GitHub]

LAMBDA

Type: real
Usage: lambda

Description: Specify the mixing parameter lambda in the formula.

Mentions:Constrained DFT

[Edit on GitHub]

LOAD_SCALE

Type: real
Default: 2.00000000E+000
Usage: LOAD_SCALE

Description: Control parameter for dynamic load balancing during a mixed CDFT calculation. See code for details. Works only in conjunction with keyword PARALLEL_BUILD.

[Edit on GitHub]

LOWDIN

Type: logical
Default: F
Lone keyword: T
Usage: LOWDIN

Description: Compute the CDFT electronic coupling element using Lowdin orthogonalization. This is the default behavior with multiple constraints and nonidentical constraints. By activating this keyword, this method is also used to compute the coupling when a single constraint is active in addition to the standard method.

Mentions:Constrained DFT

[Edit on GitHub]

METRIC

Type: logical
Default: F
Aliases: COUPLING_METRIC
Lone keyword: T
Usage: METRIC

Description: Compute reliability metric for the CDFT electronic coupling element by diagonalizing the difference density matrix.

References: Mavros2015

[Edit on GitHub]

MORE_WORK

Type: integer
Default: 0
Usage: MORE_WORK

Description: Control parameter for dynamic load balancing during a mixed CDFT calculation. See code for details. Works only in conjunction with keyword PARALLEL_BUILD.

[Edit on GitHub]

NONORTHOGONAL_COUPLING

Type: logical
Default: F
Aliases: NONORTHO_COUPLING
Lone keyword: T
Usage: NONORTHOGONAL_COUPLING

Description: Print out the nonorthogonal diabatic CDFT coupling between states, as it appears in the mixed CDFT Hamiltonian before orthogonalization (coupling calculations) and CDFT-CI. Useful for (re)constructing the Hamiltonian for additional analysis. This is the CDFT interaction energy between states.

[Edit on GitHub]

PARALLEL_BUILD

Type: logical
Default: F
Lone keyword: T
Usage: PARALLEL_BUILD TRUE

Description: Build CDFT weight function and gradients in parallel on all N MPI processors before starting the CDFT SCF calculations of the 2 involved CDFT states in parallel on N/2 processors. Supports only Becke constraints that are identical in both states. Limited to 1 charge constraint per state (different target values). The keyword MIXED&NGROUPS must be set to 2.

Mentions:Constrained DFT

[Edit on GitHub]

SCALE_WITH_OCCUPATION_NUMBERS

Type: logical
Default: T
Lone keyword: T
Usage: SCALE_WITH_OCCUPATION_NUMBERS FALSE

Description: Scale molecular orbitals with occupation numbers before calculating the electronic coupling. Affects only simulations which employ MO smearing. Disabling this keyword in conjunction with a properly selected EPS_OCCUPIED threshold might be useful in systems with a large number of fractionally occupied orbitals.

[Edit on GitHub]

VERY_OVERLOADED

Type: real
Default: 0.00000000E+000
Usage: VERY_OVERLOADED

Description: Control parameter for dynamic load balancing during a mixed CDFT calculation. See code for details. Works only in conjunction with keyword PARALLEL_BUILD.

[Edit on GitHub]

WFN_OVERLAP

Type: logical
Default: F
Lone keyword: T
Usage: WFN_OVERLAP

Description: Compute the CDFT electronic coupling element using the wavefunction overlap method in addition to the standard method defined by the keyword COUPLING. In this method, the unconstrained KS ground state wavefunction (WFN_RESTART_FILE_NAME) is represented as a linear combination of the CDFT states. For more than two CDFT states, the coupling is computed pairwise for every state pair (contrary to other coupling methods).

References: Migliore2009

Mentions:Constrained DFT

[Edit on GitHub]

WFN_RESTART_FILE_NAME

Type: string
Usage: WFN_RESTART_FILE_NAME

Description: Name of the wavefunction restart file that defines the unconstrained KS ground state, which is used to compute the electronic coupling with the wavefunction overlap method. May include a path.

[Edit on GitHub]