solve_arpack_shift_invert – Legolas v. 2.1.1

module procedure solve_arpack_shift_invert module module subroutine solve_arpack_shift_invert(arpack_cfg, matrix_A, matrix_B, settings, omega, vr)

Implementation of the ARPACK shift-invert solver

Note

In applying shift-invert we made the transformation $C = inv[B]*A$ and solved the standard eigenvalue problem $CX = \nu X$ instead since B isn't always Hermitian (e.g. if we include Hall). According to the ARPACK documentation, section 3.2.2, this implies that we must manually transform the eigenvalues $\nu_j$ from $C$ to the eigenvalues $\omega_j$ from the original system. This uses the relation $\omega_j = \sigma + \frac{1}{\nu_j}$

Arguments

Type	Intent		Name
type(arpack_t),	intent(in)	::	arpack_cfg	arpack configuration
type(matrix_t),	intent(in)	::	matrix_A	matrix A
type(matrix_t),	intent(in)	::	matrix_B	matrix B
type(settings_t),	intent(in)	::	settings	settings object
complex(kind=dp),	intent(out)	::	omega(:)	array with eigenvalues
complex(kind=dp),	intent(out)	::	vr(:,:)	array with right eigenvectors

None

solve_arpack_shift_invert Module Procedure

Contents

module procedure solve_arpack_shift_invert module module subroutine solve_arpack_shift_invert(arpack_cfg, matrix_A, matrix_B, settings, omega, vr)

Note

Arguments

Contents

solve_arpack_shift_invert Module Procedure

Contents

module procedure solve_arpack_shift_invert module module subroutine solve_arpack_shift_invert(arpack_cfg, matrix_A, matrix_B, settings, omega, vr) Interface →

Note

Arguments

Contents

module procedure solve_arpack_shift_invert module module subroutine solve_arpack_shift_invert(arpack_cfg, matrix_A, matrix_B, settings, omega, vr)