oomph::BlackBoxFDNewtonSolver Namespace Reference

Namespace for black-box FD Newton solver. More...

Typedefs
typedef void(*	ResidualFctPt) (const Vector< double > &parameters, const Vector< double > &unknowns, Vector< double > &residuals)
typedef void(*	JacobianFctPt) (const Vector< double > &, const Vector< double > &, DenseDoubleMatrix &jacobian)

Functions
void	black_box_fd_newton_solve (ResidualFctPt residual_fct, const Vector< double > &params, Vector< double > &unknowns, JacobianFctPt jacobian_fct)
	Black-box FD Newton solver: Calling sequence for residual function is.
void	line_search (const Vector< double > &x_old, const double half_residual_squared_old, const Vector< double > &gradient, ResidualFctPt residual_fct, const Vector< double > &params, Vector< double > &newton_dir, Vector< double > &x, double &half_residual_squared, const double &stpmax)
	Line search helper for globally convergent Newton method.

Variables
unsigned	Max_iter = 20
	Max. # of Newton iterations.
unsigned	N_iter_taken = 0
	Number of Newton iterations taken in most recent invocation.
bool	Doc_Progress = false
	Flag to indicate if progress of Newton iteration is to be documented (defaults to false)
double	FD_step = 1.0e-8
	FD step.
double	Tol = 1.0e-8
	Tolerance.
bool	Use_step_length_control = false
	Use steplength control do make globally convergent (default false)

Detailed Description

Namespace for black-box FD Newton solver.

Typedef Documentation

JacobianFctPt

typedef void(* oomph::BlackBoxFDNewtonSolver::JacobianFctPt) (const Vector< double > &, const Vector< double > &, DenseDoubleMatrix &jacobian)

Definition at line 60 of file black_box_newton_solver.h.

ResidualFctPt

typedef void(* oomph::BlackBoxFDNewtonSolver::ResidualFctPt) (const Vector< double > &parameters, const Vector< double > &unknowns, Vector< double > &residuals)

Definition at line 55 of file black_box_newton_solver.h.

Function Documentation

black_box_fd_newton_solve()

void oomph::BlackBoxFDNewtonSolver::black_box_fd_newton_solve	(	ResidualFctPt	residual_fct,
		const Vector< double > &	params,
		Vector< double > &	unknowns,
		JacobianFctPt	jacobian_fct
	)

Black-box FD Newton solver: Calling sequence for residual function is.

residual_fct(parameters,unknowns,residuals) 

where all arguments are double Vectors. unknowns.size() = residuals.size()

Reset number of Newton iterations taken in most recent invocation

Definition at line 67 of file black_box_newton_solver.cc.

References Doc_Progress, FD_step, i, line_search(), Max_iter, N_iter_taken, oomph::oomph_info, oomph::FiniteElement::size(), oomph::DoubleMatrixBase::solve(), oomph::Matrix< T, MATRIX_TYPE >::sparse_indexed_output(), oomph::OomphInfo::stream_pt(), Tol, and Use_step_length_control.

line_search()

void oomph::BlackBoxFDNewtonSolver::line_search	(	const Vector< double > &	x_old,
		const double	half_residual_squared_old,
		const Vector< double > &	gradient,
		ResidualFctPt	residual_fct,
		const Vector< double > &	params,
		Vector< double > &	newton_dir,
		Vector< double > &	x,
		double &	half_residual_squared,
		const double &	stpmax
	)

Line search helper for globally convergent Newton method.

Definition at line 233 of file black_box_newton_solver.cc.

References i, and oomph::FiniteElement::size().

Referenced by black_box_fd_newton_solve().

Variable Documentation

Doc_Progress

bool oomph::BlackBoxFDNewtonSolver::Doc_Progress = false

Flag to indicate if progress of Newton iteration is to be documented (defaults to false)

Definition at line 51 of file black_box_newton_solver.cc.

Referenced by black_box_fd_newton_solve().

FD_step

double oomph::BlackBoxFDNewtonSolver::FD_step = 1.0e-8

FD step.

Definition at line 54 of file black_box_newton_solver.cc.

Referenced by black_box_fd_newton_solve().

Max_iter

unsigned oomph::BlackBoxFDNewtonSolver::Max_iter = 20

Max. # of Newton iterations.

Definition at line 44 of file black_box_newton_solver.cc.

Referenced by black_box_fd_newton_solve().

N_iter_taken

unsigned oomph::BlackBoxFDNewtonSolver::N_iter_taken = 0

Number of Newton iterations taken in most recent invocation.

Definition at line 47 of file black_box_newton_solver.cc.

Referenced by black_box_fd_newton_solve().

Tol

double oomph::BlackBoxFDNewtonSolver::Tol = 1.0e-8

Tolerance.

Definition at line 57 of file black_box_newton_solver.cc.

Referenced by black_box_fd_newton_solve().

Use_step_length_control

bool oomph::BlackBoxFDNewtonSolver::Use_step_length_control = false

Use steplength control do make globally convergent (default false)

Definition at line 60 of file black_box_newton_solver.cc.

Referenced by black_box_fd_newton_solve().