ConjGradient.h File Reference

Preconditioned conjugate gradient solver (solves Ax = b using the conjugate gradient method with one of a selection of preconditioners) More...

#include <openvdb/Exceptions.h>
#include <openvdb/Types.h>
#include <openvdb/util/logging.h>
#include <openvdb/util/Assert.h>
#include <openvdb/util/NullInterrupter.h>
#include "Math.h"
#include <tbb/parallel_for.h>
#include <tbb/parallel_reduce.h>
#include <algorithm>
#include <cmath>
#include <limits>
#include <sstream>
#include <string>

Go to the source code of this file.

Classes
class	Vector< T >
	Lightweight, variable-length vector. More...
class	SparseStencilMatrix< ValueType_, STENCIL_SIZE >
	Sparse, square matrix representing a 3D stencil operator of size STENCIL_SIZE. More...
class	Preconditioner< T >
	Base class for conjugate gradient preconditioners. More...
class	JacobiPreconditioner< MatrixType >
	Diagonal preconditioner. More...
class	IncompleteCholeskyPreconditioner< MatrixType >
	Preconditioner using incomplete Cholesky factorization. More...
struct	State
	Information about the state of a conjugate gradient solution. More...
class	Vector< T >
	Lightweight, variable-length vector. More...
class	SparseStencilMatrix< ValueType_, STENCIL_SIZE >
	Sparse, square matrix representing a 3D stencil operator of size STENCIL_SIZE. More...
class	SparseStencilMatrix< ValueType_, STENCIL_SIZE >::ConstValueIter
	Iterator over the stored values in a row of this matrix. More...
class	SparseStencilMatrix< ValueType_, STENCIL_SIZE >::ConstRow
	Read-only accessor to a row of this matrix. More...
class	SparseStencilMatrix< ValueType_, STENCIL_SIZE >::RowEditor
	Read/write accessor to a row of this matrix. More...
class	Preconditioner< T >
	Base class for conjugate gradient preconditioners. More...
struct	CopyOp< T >
struct	FillOp< T >
struct	LinearOp< T >
struct	Vector< T >::DeterministicDotProductOp
struct	Vector< T >::InfNormOp
struct	Vector< T >::IsFiniteOp
struct	SparseStencilMatrix< ValueType_, STENCIL_SIZE >::MatrixCopyOp
struct	SparseStencilMatrix< ValueType_, STENCIL_SIZE >::IsFiniteOp
class	JacobiPreconditioner< MatrixType >
	Diagonal preconditioner. More...
class	IncompleteCholeskyPreconditioner< MatrixType >
	Preconditioner using incomplete Cholesky factorization. More...

Namespaces
	openvdb
	openvdb::v12_0
	openvdb::v12_0::math
	openvdb::v12_0::math::pcg
	openvdb::v12_0::math::pcg::internal

Typedefs
using	SizeType = Index32
using	SizeRange = tbb::blocked_range< SizeType >
using	VectorS = Vector< float >
using	VectorD = Vector< double >

Functions
template<typename ValueType >
State	terminationDefaults ()
	Return default termination conditions for a conjugate gradient solver. More...
template<typename PositiveDefMatrix >
State	solve (const PositiveDefMatrix &A, const Vector< typename PositiveDefMatrix::ValueType > &b, Vector< typename PositiveDefMatrix::ValueType > &x, Preconditioner< typename PositiveDefMatrix::ValueType > &preconditioner, const State &termination=terminationDefaults< typename PositiveDefMatrix::ValueType >())
	Solve Ax = b via the preconditioned conjugate gradient method. More...
template<typename PositiveDefMatrix , typename Interrupter >
State	solve (const PositiveDefMatrix &A, const Vector< typename PositiveDefMatrix::ValueType > &b, Vector< typename PositiveDefMatrix::ValueType > &x, Preconditioner< typename PositiveDefMatrix::ValueType > &preconditioner, Interrupter &interrupter, const State &termination=terminationDefaults< typename PositiveDefMatrix::ValueType >())
	Solve Ax = b via the preconditioned conjugate gradient method. More...
std::ostream &	operator<< (std::ostream &os, const State &state)
template<typename T >
void	axpy (const T &a, const T *xVec, const T *yVec, T *resultVec, SizeType size)
	Compute ax + y. More...
template<typename T >
void	axpy (const T &a, const Vector< T > &xVec, const Vector< T > &yVec, Vector< T > &result)
	Compute ax + y. More...
template<typename MatrixOperator , typename VecValueType >
void	computeResidual (const MatrixOperator &A, const VecValueType *x, const VecValueType *b, VecValueType *r)
	Compute r = b − Ax. More...
template<typename MatrixOperator , typename T >
void	computeResidual (const MatrixOperator &A, const Vector< T > &x, const Vector< T > &b, Vector< T > &r)
	Compute r = b − Ax. More...

Detailed Description

Preconditioned conjugate gradient solver (solves Ax = b using the conjugate gradient method with one of a selection of preconditioners)

Authors

D.J. Hill, Peter Cucka