Sparse matrix partitioned in an unknown and a prescribed part. More...

#include <GooseFEM/MatrixPartitioned.h>

Inheritance diagram for GooseFEM::MatrixPartitioned:

Public Member Functions
	MatrixPartitioned (const array_type::tensor< size_t, 2 > &conn, const array_type::tensor< size_t, 2 > &dofs, const array_type::tensor< size_t, 1 > &iip)
	Constructor.

const Eigen::SparseMatrix< double > &	data_uu () const
	Pointer to data.

const Eigen::SparseMatrix< double > &	data_up () const
	Pointer to data.

const Eigen::SparseMatrix< double > &	data_pu () const
	Pointer to data.

const Eigen::SparseMatrix< double > &	data_pp () const
	Pointer to data.

void	set (const array_type::tensor< size_t, 1 > &rows, const array_type::tensor< size_t, 1 > &cols, const array_type::tensor< double, 2 > &matrix)
	Overwrite matrix.

void	add (const array_type::tensor< size_t, 1 > &rows, const array_type::tensor< size_t, 1 > &cols, const array_type::tensor< double, 2 > &matrix)
	Add matrix.

Public Member Functions inherited from GooseFEM::MatrixPartitionedBase< MatrixPartitioned >
size_t	nnu () const
	Number of unknown DOFs.

size_t	nnp () const
	Number of prescribed DOFs.

const array_type::tensor< size_t, 1 > &	iiu () const
	Unknown DOFs.

const array_type::tensor< size_t, 1 > &	iip () const
	Prescribed DOFs.

array_type::tensor< double, 2 >	Reaction (const array_type::tensor< double, 2 > &x, const array_type::tensor< double, 2 > &b) const
	Right-hand-size for corresponding to the prescribed DOFs:

array_type::tensor< double, 1 >	Reaction (const array_type::tensor< double, 1 > &x, const array_type::tensor< double, 1 > &b) const
	Same as Reaction(const array_type::tensor<double, 2>&, const array_type::tensor<double, 2>&), but of "dofval" input and output.

void	reaction (const array_type::tensor< double, 2 > &x, array_type::tensor< double, 2 > &b) const
	Same as Reaction(const array_type::tensor<double, 2>&, const array_type::tensor<double, 2>&), but inserting in-place.

void	reaction (const array_type::tensor< double, 1 > &x, array_type::tensor< double, 1 > &b) const
	Same as Reaction(const array_type::tensor<double, 1>&, const array_type::tensor<double, 1>&), but inserting in-place.

array_type::tensor< double, 1 >	Reaction_p (const array_type::tensor< double, 1 > &x_u, const array_type::tensor< double, 1 > &x_p) const
	Same as Reaction(const array_type::tensor<double, 1>&, const array_type::tensor<double, 1>&), but with partitioned input and output.

void	reaction_p (const array_type::tensor< double, 1 > &x_u, const array_type::tensor< double, 1 > &x_p, array_type::tensor< double, 1 > &b_p) const
	Same as Reaction_p(const array_type::tensor<double, 1>&, const array_type::tensor<double, 1>&), but writing to preallocated output.

Public Member Functions inherited from GooseFEM::MatrixBase< D >
size_t	nelem () const
	Number of elements.

size_t	nne () const
	Number of nodes per element.

size_t	nnode () const
	Number of nodes.

size_t	ndim () const
	Number of dimensions.

size_t	ndof () const
	Number of DOFs.

const array_type::tensor< size_t, 2 > &	dofs () const
	DOFs per node.

const array_type::tensor< size_t, 2 > &	conn () const
	Connectivity.

std::array< size_t, 1 >	shape_dofval () const
	Shape of "dofval".

std::array< size_t, 2 >	shape_nodevec () const
	Shape of "nodevec".

std::array< size_t, 3 >	shape_elemmat () const
	Shape of "elemmat".

template<class T >
void	assemble (const T &elemmat)
	Assemble from "elemmat".

array_type::tensor< double, 2 >	Todense () const
	Copy as dense matrix.

template<class T >
void	todense (T &ret) const
	Copy to dense matrix.

array_type::tensor< double, 2 >	Dot (const array_type::tensor< double, 2 > &x) const
	Dot-product \( b_i = A_{ij} x_j \).

array_type::tensor< double, 1 >	Dot (const array_type::tensor< double, 1 > &x) const
	Dot-product \( b_i = A_{ij} x_j \).

void	dot (const array_type::tensor< double, 2 > &x, array_type::tensor< double, 2 > &b) const
	Dot-product \( b_i = A_{ij} x_j \).

void	dot (const array_type::tensor< double, 1 > &x, array_type::tensor< double, 1 > &b) const
	Dot-product \( b_i = A_{ij} x_j \).

Protected Attributes
Eigen::SparseMatrix< double >	m_Auu
	The matrix.

Eigen::SparseMatrix< double >	m_Aup
	The matrix.

Eigen::SparseMatrix< double >	m_Apu
	The matrix.

Eigen::SparseMatrix< double >	m_App
	The matrix.

std::vector< Eigen::Triplet< double > >	m_Tuu
	Matrix entries.

std::vector< Eigen::Triplet< double > >	m_Tup
	Matrix entries.

std::vector< Eigen::Triplet< double > >	m_Tpu
	Matrix entries.

std::vector< Eigen::Triplet< double > >	m_Tpp
	Matrix entries.

array_type::tensor< size_t, 2 >	m_part
	Renumbered DOFs per node, such that:

array_type::tensor< size_t, 1 >	m_part1d
	Map real DOF to DOF in partitioned system.

Protected Attributes inherited from GooseFEM::MatrixPartitionedBase< MatrixPartitioned >
array_type::tensor< size_t, 1 >	m_iiu
	See iiu()

array_type::tensor< size_t, 1 >	m_iip
	See iip()

size_t	m_nnu
	See nnu.

size_t	m_nnp
	See nnp.

Protected Attributes inherited from GooseFEM::MatrixBase< D >
array_type::tensor< size_t, 2 >	m_conn
	Connectivity [nelem, nne].

array_type::tensor< size_t, 2 >	m_dofs
	DOF-numbers per node [nnode, ndim].

size_t	m_nelem
	See nelem().

size_t	m_nne
	See nne().

size_t	m_nnode
	See nnode().

size_t	m_ndim
	See ndim().

size_t	m_ndof
	See ndof().

bool	m_changed = true
	Signal changes to data.

Friends
template<class >
class	MatrixPartitionedSolver
	Class to solve the system (allowing single factorisation for multiple right-hand-sides).

Additional Inherited Members
Public Types inherited from GooseFEM::MatrixPartitionedBase< MatrixPartitioned >
using	derived_type
	Underlying type.

Public Types inherited from GooseFEM::MatrixBase< D >
using	derived_type = D
	Underlying type.

Detailed Description

Sparse matrix partitioned in an unknown and a prescribed part.

In particular: \( \begin{bmatrix} A_{uu} & A_{up} \\ A_{pu} & A_{pp} \end{bmatrix} \)

See VectorPartitioned() for bookkeeping definitions.

Definition at line 34 of file MatrixPartitioned.h.

Constructor & Destructor Documentation

◆ MatrixPartitioned()

GooseFEM::MatrixPartitioned::MatrixPartitioned	(	const array_type::tensor< size_t, 2 > &	conn,
		const array_type::tensor< size_t, 2 > &	dofs,
		const array_type::tensor< size_t, 1 > &	iip )

inline

Constructor.

Parameters

conn	connectivity [nelem, nne].
dofs	DOFs per node [nnode, ndim].
iip	prescribed DOFs [nnp].

Definition at line 87 of file MatrixPartitioned.h.

Member Function Documentation

◆ add()

void GooseFEM::MatrixPartitioned::add	(	const array_type::tensor< size_t, 1 > &	rows,
		const array_type::tensor< size_t, 1 > &	cols,
		const array_type::tensor< double, 2 > &	matrix )

inline

Add matrix.

Parameters

rows	Row numbers [m].
cols	Column numbers [n].
matrix	Data entries `matrix(i, j)` for `rows(i), cols(j)` [m, n].

Definition at line 271 of file MatrixPartitioned.h.

◆ data_pp()

const Eigen::SparseMatrix< double > & GooseFEM::MatrixPartitioned::data_pp ( ) const

inline

Pointer to data.

Definition at line 150 of file MatrixPartitioned.h.

◆ data_pu()

const Eigen::SparseMatrix< double > & GooseFEM::MatrixPartitioned::data_pu ( ) const

inline

Pointer to data.

Definition at line 142 of file MatrixPartitioned.h.

◆ data_up()

const Eigen::SparseMatrix< double > & GooseFEM::MatrixPartitioned::data_up ( ) const

inline

Pointer to data.

Definition at line 134 of file MatrixPartitioned.h.

◆ data_uu()

const Eigen::SparseMatrix< double > & GooseFEM::MatrixPartitioned::data_uu ( ) const

inline

Pointer to data.

Definition at line 126 of file MatrixPartitioned.h.

◆ set()

void GooseFEM::MatrixPartitioned::set	(	const array_type::tensor< size_t, 1 > &	rows,
		const array_type::tensor< size_t, 1 > &	cols,
		const array_type::tensor< double, 2 > &	matrix )

inline

Overwrite matrix.

Parameters

rows	Row numbers [m].
cols	Column numbers [n].
matrix	Data entries `matrix(i, j)` for `rows(i), cols(j)` [m, n].

Definition at line 221 of file MatrixPartitioned.h.

Friends And Related Symbol Documentation

◆ MatrixPartitionedSolver

template<class >

friend class MatrixPartitionedSolver

array_type::tensor<size_t, 2> GooseFEM::MatrixPartitioned::m_part

protected

Renumbered DOFs per node, such that:

 iiu = arange(nnu)
 iip = nnu + arange(nnp)

making is much simpler to slice.

Definition at line 58 of file MatrixPartitioned.h.

◆ m_part1d

array_type::tensor<size_t, 1> GooseFEM::MatrixPartitioned::m_part1d

protected

Map real DOF to DOF in partitioned system.

The partitioned system is defined as:

 iiu = arange(nnu)
 iip = nnu + arange(nnp)

Similar to m_part but for a 1d sequential list of DOFs.

Definition at line 69 of file MatrixPartitioned.h.

◆ m_Tpp

std::vector<Eigen::Triplet<double> > GooseFEM::MatrixPartitioned::m_Tpp

protected

Matrix entries.

Definition at line 48 of file MatrixPartitioned.h.

◆ m_Tpu

std::vector<Eigen::Triplet<double> > GooseFEM::MatrixPartitioned::m_Tpu

protected

Matrix entries.

Definition at line 47 of file MatrixPartitioned.h.

◆ m_Tup

std::vector<Eigen::Triplet<double> > GooseFEM::MatrixPartitioned::m_Tup

protected

Matrix entries.

Definition at line 46 of file MatrixPartitioned.h.

◆ m_Tuu

std::vector<Eigen::Triplet<double> > GooseFEM::MatrixPartitioned::m_Tuu

protected

Matrix entries.

Definition at line 45 of file MatrixPartitioned.h.

The documentation for this class was generated from the following file:

GooseFEM/MatrixPartitioned.h

Public Member Functions

Protected Attributes

Friends

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ MatrixPartitioned()

Member Function Documentation

◆ add()

◆ data_pp()

◆ data_pu()

◆ data_up()

◆ data_uu()

◆ set()

Friends And Related Symbol Documentation

◆ MatrixPartitionedSolver

Member Data Documentation

◆ m_App

◆ m_Apu

◆ m_Aup

◆ m_Auu

◆ m_part

◆ m_part1d

◆ m_Tpp

◆ m_Tpu

◆ m_Tup

◆ m_Tuu