LOCALIZE

Use one of the available methods to define the localization and possibly to optimize it to a minimum or a maximum. [Edit on GitHub]

Subsections

Keywords

Keyword descriptions

SECTION_PARAMETERS: logical = F

Lone keyword: T

Usage: &LOCALIZE T

controls the activation of the MOS localization procedure [Edit on GitHub]

CG_PO: logical = T

Lone keyword: T

Usage: CG_PO

Use conjugate gradient in conjunction with METHOD GAPO. If FALSE, steepest descent is used instead. [Edit on GitHub]

CPO_GUESS: enum = ATOMIC

Usage: CPO_GUESS (ATOMIC|RESTART|RANDOM)

Valid values:

  • ATOMIC

  • RESTART

  • RANDOM

Initial guess for coefficients if METHOD GAPO is used [Edit on GitHub]

CPO_GUESS_SPACE: enum = WAN

Usage: CPO_GUESS_SPACE (WAN|ALL)

Valid values:

  • WAN

  • ALL

Orbital space from which initial guess for coefficients is determined if METHOD GAPO and CPO_GUESS ATOMIC are employed [Edit on GitHub]

CRAZY_SCALE: real = 1.00000000E+000

Usage: CRAZY_SCALE 0.9

scale angles [Edit on GitHub]

CRAZY_USE_DIAG: logical = F

Lone keyword: T

Usage: CRAZY_USE_DIAG

Use diagonalization (slow) or pade based calculation of matrix exponentials. [Edit on GitHub]

ENERGY_RANGE: real[2] = 0.00000000E+000 0.00000000E+000 [eV]

Usage: _ ENERGY_RANGE lower_bound {real}, higher_bound {real}_

Select the orbitals to be localized within the given energy range. This type of selection cannot be added on top of the selection through a LIST. It reads to reals that are lower and higher boundaries of the energy range. [Edit on GitHub]

EPS_LOCALIZATION: real = 1.00000000E-004

Usage: EPS_LOCALIZATION 1.0E-2

Tolerance used in the convergence criterion of the localization methods. [Edit on GitHub]

EPS_OCCUPATION: real = 1.00000000E-008

Usage: EPS_OCCUPATION 1.E-5

Tolerance in the occupation number to select only fully occupied orbitals for the rotation [Edit on GitHub]

JACOBI_FALLBACK: logical = T

Lone keyword: T

Usage: JACOBI_FALLBACK

Use Jacobi method in case no convergence was achieved by using the crazy rotations method. [Edit on GitHub]

JACOBI_REFINEMENT: logical = F

Lone keyword: T

Usage: JACOBI_REFINEMENT

Use Jacobi method to refine the localisation obtained by SCDM [Edit on GitHub]

LIST: integer

Keyword can be repeated.

Usage: LIST 1 2

Indexes of the occupied wfn to be localized This keyword can be repeated several times (useful if you have to specify many indexes). [Edit on GitHub]

LIST_UNOCCUPIED: integer

Keyword can be repeated.

Usage: LIST 1 2

Indexes of the unoccupied states to be localized, up to now only valid in combination with GPW. This keyword has to be present if unoccupied states should be localized. This keyword can be repeated several times (useful if you have to specify many indexes). [Edit on GitHub]

LOCHOMO_RESTART_FILE_NAME: string

Usage: LOCHOMO_RESTART_FILE_NAME

File name where to read the MOS from which to restart the localization procedure for occupied states [Edit on GitHub]

LOCLUMO_RESTART_FILE_NAME: string

Usage: LOCLUMO_RESTART_FILE_NAME

File name where to read the MOS from which to restart the localization procedure for unoccupied states [Edit on GitHub]

LOCMIXD_RESTART_FILE_NAME: string

Usage: LOCMIXD_RESTART_FILE_NAME

File name where to read the MOS from which to restart the localization procedure for MIXED states [Edit on GitHub]

MAX_CRAZY_ANGLE: real = 2.00000000E-001 [rad]

Usage: MAX_CRAZY_ANGLE 0.1

Largest allowed angle for the crazy rotations algorithm (smaller is slower but more stable). [Edit on GitHub]

MAX_ITER: integer = 10000

Usage: MAX_ITER 2000

Maximum number of iterations used for localization methods [Edit on GitHub]

METHOD: enum = JACOBI

Usage: METHOD (JACOBI|CRAZY|DIRECT|GAPO|L1SD|SCDM|NONE)

Valid values:

  • NONE No localization is applied

  • JACOBI Using 2 x 2 rotations of the orbitals, slow but robust

  • CRAZY A new fast method is applied, might be slightly less robust than jacobi, but usually much faster

  • GAPO Gradient ascent for partially occupied wannier functions

  • L1SD Steepest descent minimization of an approximate l1 norm

  • DIRECT Using a direct minimisation approacha

  • SCDM Use QR factorization

Method of optimization if any [Edit on GitHub]

MIN_OR_MAX: enum = SPREADMIN

Usage: MIN_OR_MAX (SPREADMIN|SPREADMAX)

Valid values:

  • SPREADMIN

  • SPREADMAX

Requires the maximization of the spread of the wfn [Edit on GitHub]

NEXTRA: integer = 0

Usage: NEXTRA 5

Number of orbitals above fully occupied MOs to be localized, up to now only valid in combination with GPW. This keyword has to be present for STATES MIXED option. Otherwise, only the fully occupied MOs are localized. [Edit on GitHub]

OPERATOR: enum = BERRY

Usage: OPERATOR (BERRY|BOYS|PIPEK)

Valid values:

  • BERRY

  • BOYS

  • PIPEK

Type of opertator which defines the spread functional [Edit on GitHub]

OUT_ITER_EACH: integer = 100

Usage: OUT_ITER_EACH 100

Every how many iterations of the localization algorithm (Jacobi) the tolerance value is printed out [Edit on GitHub]

RESTART: logical = F

Lone keyword: T

Usage: RESTART

Restart the localization from a set of orbitals read from a localization restart file. [Edit on GitHub]

STATES: enum = OCCUPIED

Usage: STATES (HOMO|LUMO|MIXED|ALL)

Valid values:

  • OCCUPIED

  • UNOCCUPIED

  • MIXED

  • ALL

Which states to localize, LUMO up to now only available in GPW [Edit on GitHub]

USE_HISTORY: logical = F

Lone keyword: T

Usage: USE_HISTORY

Generate an improved initial guess based on a history of results, which is useful during MD. Will only work if the number of states to be localized remains constant. [Edit on GitHub]