LS_SCF

References: VandeVondele2012

Specifies the parameters of the linear scaling SCF routines [Edit on GitHub]

Subsections

• CHEBYSHEV
• CURVY_STEPS
• PAO
• PEXSI
• RHO_MIXING

Keywords

• CHECK_S_INV

• DYNAMIC_THRESHOLD

• EPS_DIIS

• EPS_FILTER

• EPS_LANCZOS

• EPS_SCF

• EXTRAPOLATION_ORDER

• FIXED_MU

• INI_DIIS

• LS_DIIS

• MATRIX_CLUSTER_TYPE

• MAX_DIIS

• MAX_ITER_LANCZOS

• MAX_SCF

• MIXING_FRACTION

• MU

• NMIXING

• NON_MONOTONIC

• PERFORM_MU_SCAN

• PURIFICATION_METHOD

• REPORT_ALL_SPARSITIES

• RESTART_READ

• RESTART_WRITE

• SIGN_METHOD

• SIGN_ORDER

• SIGN_SYMMETRIC

• SUBMATRIX_SIGN_METHOD

• S_INVERSION

• S_PRECONDITIONER

• S_SQRT_METHOD

• S_SQRT_ORDER

Keyword descriptions

CHECK_S_INV: logical = F

Lone keyword: T

Usage: CHECK_S_INV

Perform an accuracy check on the inverse/sqrt of the s matrix. [Edit on GitHub]

DYNAMIC_THRESHOLD: logical = F

Lone keyword: T

Usage: DYNAMIC_THRESHOLD .TRUE.

Should the threshold for the purification be chosen dynamically [Edit on GitHub]

EPS_DIIS: real = 1.00000000E-001

Usage: EPS_DIIS 1.e-1

Threshold on the convergence to start using DIIS [Edit on GitHub]

EPS_FILTER: real = 1.00000000E-006

Usage: EPS_FILTER 1.0E-7

Mentions: ⭐PAO-ML

Threshold used for filtering matrix operations. [Edit on GitHub]

EPS_LANCZOS: real = 1.00000000E-003

Usage: EPS_LANCZOS 1.0E-4

Threshold used for lanczos estimates. [Edit on GitHub]

EPS_SCF: real = 1.00000000E-007

Usage: EPS_SCF 1.e-6

Mentions: ⭐PAO-ML

Target accuracy for the SCF convergence in terms of change of the total energy per electron. [Edit on GitHub]

EXTRAPOLATION_ORDER: integer = 4

Usage: EXTRAPOLATION_ORDER 3

Number of previous matrices used for the ASPC extrapolation of the initial guess. 0 implies that an atomic guess is used at each step. low (1-2) will result in a drift of the constant of motion during MD. high (>5) might be somewhat unstable, leading to more SCF iterations. [Edit on GitHub]

FIXED_MU: logical = F

Lone keyword: T

Usage: FIXED_MU .TRUE.

Should the calculation be performed at fixed chemical potential, or should it be found fixing the number of electrons [Edit on GitHub]

INI_DIIS: integer = 2

Usage: INI_DIIS 2

Iteration cycle to start DIIS Kohn-Sham matrix update [Edit on GitHub]

LS_DIIS: logical = F

Lone keyword: T

Usage: LS_DIIS

Perform DIIS within linear scaling SCF [Edit on GitHub]

MATRIX_CLUSTER_TYPE: enum = ATOMIC

Usage: MATRIX_CLUSTER_TYPE MOLECULAR

Valid values:

• ATOMIC Using atomic blocks

• MOLECULAR Using molecular blocks.

Specify how atomic blocks should be clustered in the used matrices, in order to improve flop rate, and possibly speedup the matrix multiply. Note that the atomic s_preconditioner can not be used. Furthermore, since screening is on matrix blocks, slightly more accurate results can be expected with molecular. [Edit on GitHub]

MAX_DIIS: integer = 4

Usage: MAX_DIIS 4

Size of LS_DIIS buffer [Edit on GitHub]

MAX_ITER_LANCZOS: integer = 128

Usage: MAX_ITER_LANCZOS

Maximum number of lanczos iterations. [Edit on GitHub]

MAX_SCF: integer = 20

Usage: MAX_SCF 200

Maximum number of SCF iteration to be performed for one optimization [Edit on GitHub]

MIXING_FRACTION: real = 4.50000000E-001

Usage: MIXING_FRACTION 0.4

Mixing density matrices uses the specified fraction in the SCF procedure. [Edit on GitHub]

MU: real = -1.00000000E-001

Usage: MU 0.0

Value (or initial guess) for the chemical potential, i.e. some suitable energy between HOMO and LUMO energy. [Edit on GitHub]

NMIXING: integer = 2

Usage: NMIXING 2

Minimal number of density mixing before start DIIS [Edit on GitHub]

NON_MONOTONIC: logical = T

Lone keyword: T

Usage: NON_MONOTONIC .TRUE.

Should the purification be performed non-monotonically. Relevant for TC2 only. [Edit on GitHub]

PERFORM_MU_SCAN: logical = F

Lone keyword: T

Usage: PERFORM_MU_SCAN

Do a scan of the chemical potential after the SCF [Edit on GitHub]

PURIFICATION_METHOD: enum = SIGN

Usage: PURIFICATION_METHOD TRS4

Valid values:

• SIGN Sign matrix iteration.

• TRS4 Trace resetting 4th order scheme

• TC2 Trace conserving 2nd order scheme

• PEXSI PEXSI method

References: VandeVondele2012, Niklasson2003

Scheme used to purify the Kohn-Sham matrix into the density matrix. [Edit on GitHub]

REPORT_ALL_SPARSITIES: logical = T

Lone keyword: T

Usage: REPORT_ALL_SPARSITIES

Run the sparsity report at the end of the SCF [Edit on GitHub]

RESTART_READ: logical = F

Lone keyword: T

Usage: RESTART_READ

Read the density matrix before the (first) SCF. [Edit on GitHub]

RESTART_WRITE: logical = F

Lone keyword: T

Usage: RESTART_READ

Mentions: ⭐PAO-ML

Write the density matrix at the end of the SCF (currently requires EXTRAPOLATION_ORDER>0). Files might be rather large. [Edit on GitHub]

SIGN_METHOD: enum = NEWTONSCHULZ

Usage: SIGN_METHOD NEWTONSCHULZ

Valid values:

• NEWTONSCHULZ Newton-Schulz iteration.

• PROOT p-th order root iteration

• SUBMATRIX Submatrix method

References: VandeVondele2012, Niklasson2003

Method used for the computation of the sign matrix. [Edit on GitHub]

SIGN_ORDER: integer = 2

Usage: SIGN_ORDER 2

Order of the method used for the computation of the sign matrix. [Edit on GitHub]

SIGN_SYMMETRIC: logical = F

Lone keyword: T

Usage: SIGN_SYMMETRIC .TRUE.

Use symmetric orthogonalization when generating the input for the sign function. [Edit on GitHub]

SUBMATRIX_SIGN_METHOD: enum = NEWTONSCHULZ

Usage: SUBMATRIX_SIGN_METHOD NEWTONSCHULZ

Valid values:

• NEWTONSCHULZ Newton-Schulz iteration.

• DIRECT Direct method calculating all eigenvalues.

• DIRECT_MUADJ Direct method with internal adjustment of mu

• DIRECT_MUADJ_LOWMEM Direct method with internal adjustment of mu, using two passes to save memory

Method used for the computation of the sign matrix of all submatrices. [Edit on GitHub]

S_INVERSION: enum = SIGN_SQRT

Usage: S_PRECONDITIONER MOLECULAR

Valid values:

• SIGN_SQRT Using the inverse sqrt as obtained from sign function iterations.

• HOTELLING Using the Hotellign iteration.

Method used to compute the inverse of S. [Edit on GitHub]

S_PRECONDITIONER: enum = ATOMIC

Usage: S_PRECONDITIONER MOLECULAR

Valid values:

• NONE No preconditioner

• ATOMIC Using atomic blocks

• MOLECULAR Using molecular sub-blocks. Recommended if molecules are defined and not too large.

Preconditions S with some appropriate form. [Edit on GitHub]

S_SQRT_METHOD: enum = NEWTONSCHULZ

Usage: S_SQRT_METHOD NEWTONSCHULZ

Valid values:

• NEWTONSCHULZ Using a Newton-Schulz-like iteration

• PROOT Using the p-th root method.

Method for the caclulation of the sqrt of S. [Edit on GitHub]

S_SQRT_ORDER: integer = 3

Aliases: SIGN_SQRT_ORDER

Usage: S_SQRT_ORDER 3

Order of the iteration method for the calculation of the sqrt of S. [Edit on GitHub]