MathLib::LinAlg Namespace Reference

Detailed Description

Some general linear algebra functionality.

By using the provided functions linear algebra capabilities can be used for different matrix and vector types in a way that is agnostic towards the specific type used.

For documentation, refer to that of the templated method. All specializations or overload must behave in the same way.

Functions
void	setLocalAccessibleVector (PETScVector const &x)
void	set (PETScVector &x, double const a)
void	copy (PETScVector const &x, PETScVector &y)
void	scale (PETScVector &x, double const a)
void	aypx (PETScVector &y, double const a, PETScVector const &x)
void	axpy (PETScVector &y, double const a, PETScVector const &x)
void	axpby (PETScVector &y, double const a, double const b, PETScVector const &x)
template<>
void	componentwiseDivide (PETScVector &w, PETScVector const &x, PETScVector const &y)
template<>
double	norm1 (PETScVector const &x)
template<>
double	norm2 (PETScVector const &x)
template<>
double	normMax (PETScVector const &x)
void	copy (PETScMatrix const &A, PETScMatrix &B)
void	scale (PETScMatrix &A, double const a)
void	aypx (PETScMatrix &Y, double const a, PETScMatrix const &X)
void	axpy (PETScMatrix &Y, double const a, PETScMatrix const &X)
void	matMult (PETScMatrix const &A, PETScVector const &x, PETScVector &y)
void	matMultAdd (PETScMatrix const &A, PETScVector const &v1, PETScVector const &v2, PETScVector &v3)
void	finalizeAssembly (PETScMatrix &A)
void	finalizeAssembly (PETScVector &x)
template<typename MatrixOrVector >
void	copy (MatrixOrVector const &x, MatrixOrVector &y)
	Copies x to y. More...
template<typename MatrixOrVector >
void	scale (MatrixOrVector &x, double const a)
	Scales x by a. More...
template<typename MatrixOrVector >
void	aypx (MatrixOrVector &y, double const a, MatrixOrVector const &x)
	Computes $y = a \cdot y + x$. More...
template<typename MatrixOrVector >
void	axpy (MatrixOrVector &y, double const a, MatrixOrVector const &x)
	Computes $y = a \cdot x + y$. More...
template<typename MatrixOrVector >
void	axpby (MatrixOrVector &y, double const a, double const b, MatrixOrVector const &x)
	Computes $y = a \cdot x + b \cdot y$. More...
template<typename MatrixOrVector >
void	componentwiseDivide (MatrixOrVector &w, MatrixOrVector const &x, MatrixOrVector const &y)
	Computes $w = x/y$ componentwise. More...
template<typename MatrixOrVector >
double	norm1 (MatrixOrVector const &x)
	Computes the Manhattan norm of x. More...
template<typename MatrixOrVector >
double	norm2 (MatrixOrVector const &x)
	Computes the Euclidean norm of x. More...
template<typename MatrixOrVector >
double	normMax (MatrixOrVector const &x)
	Computes the maximum norm of x. More...
template<typename MatrixOrVector >
double	norm (MatrixOrVector const &x, MathLib::VecNormType type)
template<typename Matrix >
void	finalizeAssembly (Matrix &)
template<typename Matrix , typename Vector >
void	matMult (Matrix const &A, Vector const &x, Vector &y)
template<typename Matrix , typename Vector >
void	matMultAdd (Matrix const &A, Vector const &v1, Vector const &v2, Vector &v3)

Function Documentation

axpby() [1/2]

template<typename MatrixOrVector >

void MathLib::LinAlg::axpby	(	MatrixOrVector &	y,
		double const	a,
		double const	b,
		MatrixOrVector const &	x
	)

Computes $y = a \cdot x + b \cdot y$.

Definition at line 65 of file LinAlg.h.

{
    y = a*x + b*y;
}

axpby() [2/2]

void MathLib::LinAlg::axpby	(	PETScVector &	y,
		double const	a,
		double const	b,
		PETScVector const &	x
	)

Definition at line 64 of file LinAlg.cpp.

{
    // TODO check sizes
    VecAXPBY(y.getRawVector(), a, b, x.getRawVector());
}

References MathLib::PETScVector::getRawVector().

Referenced by NumLib::TimeDiscretization::getXdot().

axpy() [1/3]

template<typename MatrixOrVector >

void MathLib::LinAlg::axpy	(	MatrixOrVector &	y,
		double const	a,
		MatrixOrVector const &	x
	)

Computes $y = a \cdot x + y$.

Definition at line 58 of file LinAlg.h.

{
    y += a*x;
}

axpy() [2/3]

void MathLib::LinAlg::axpy	(	PETScMatrix &	Y,
		double const	a,
		PETScMatrix const &	X
	)

Definition at line 131 of file LinAlg.cpp.

{
    // TODO check sizes
    // TODO sparsity pattern, currently they are assumed to be different (slow)
    MatAXPY(Y.getRawMatrix(), a, X.getRawMatrix(), DIFFERENT_NONZERO_PATTERN);
}

References MathLib::PETScMatrix::getRawMatrix().

axpy() [3/3]

void MathLib::LinAlg::axpy	(	PETScVector &	y,
		double const	a,
		PETScVector const &	x
	)

Definition at line 57 of file LinAlg.cpp.

{
    // TODO check sizes
    VecAXPY(y.getRawVector(), a, x.getRawVector());
}

References MathLib::PETScVector::getRawVector().

Referenced by NumLib::NonlinearSolver< NonlinearSolverTag::Picard >::calculateNonEquilibriumInitialResiduum(), NumLib::TimeDiscretization::computeRelativeChangeFromPreviousTimestep(), NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeResidual(), ProcessLib::computeResiduum(), NumLib::NonlinearSolver< NonlinearSolverTag::Newton >::solve(), NumLib::NonlinearSolver< NonlinearSolverTag::Picard >::solve(), and ProcessLib::TimeLoop::solveCoupledEquationSystemsByStaggeredScheme().

aypx() [1/3]

template<typename MatrixOrVector >

void MathLib::LinAlg::aypx	(	MatrixOrVector &	y,
		double const	a,
		MatrixOrVector const &	x
	)

Computes $y = a \cdot y + x$.

Definition at line 51 of file LinAlg.h.

{
    y = a*y + x;
}

aypx() [2/3]

void MathLib::LinAlg::aypx	(	PETScMatrix &	Y,
		double const	a,
		PETScMatrix const &	X
	)

Definition at line 123 of file LinAlg.cpp.

{
    // TODO check sizes
    // TODO sparsity pattern, currently they are assumed to be different (slow)
    MatAYPX(Y.getRawMatrix(), a, X.getRawMatrix(), DIFFERENT_NONZERO_PATTERN);
}

References MathLib::PETScMatrix::getRawMatrix().

aypx() [3/3]

void MathLib::LinAlg::aypx	(	PETScVector &	y,
		double const	a,
		PETScVector const &	x
	)

Definition at line 50 of file LinAlg.cpp.

{
    // TODO check sizes
    VecAYPX(y.getRawVector(), a, x.getRawVector());
}

References MathLib::PETScVector::getRawVector().

Referenced by NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeA(), and ProcessLib::ComponentTransport::ComponentTransportProcess::solveReactionEquation().

componentwiseDivide() [1/2]

template<typename MatrixOrVector >

void MathLib::LinAlg::componentwiseDivide	(	MatrixOrVector &	w,
		MatrixOrVector const &	x,
		MatrixOrVector const &	y
	)

Computes $w = x/y$ componentwise.

componentwiseDivide() [2/2]

template<>

void MathLib::LinAlg::componentwiseDivide	(	PETScVector &	w,
		PETScVector const &	x,
		PETScVector const &	y
	)

Definition at line 73 of file LinAlg.cpp.

{
    VecPointwiseDivide(w.getRawVector(), x.getRawVector(), y.getRawVector());
}

References MathLib::PETScVector::getRawVector().

Referenced by NumLib::LocalLinearLeastSquaresExtrapolator::extrapolate().

copy() [1/3]

template<typename MatrixOrVector >

void MathLib::LinAlg::copy	(	MatrixOrVector const &	x,
		MatrixOrVector &	y
	)

Copies x to y.

Definition at line 37 of file LinAlg.h.

{
    y = x;
}

copy() [2/3]

void MathLib::LinAlg::copy	(	PETScMatrix const &	A,
		PETScMatrix &	B
	)

Definition at line 111 of file LinAlg.cpp.

{
    B = A;
}

copy() [3/3]

void MathLib::LinAlg::copy	(	PETScVector const &	x,
		PETScVector &	y
	)

Definition at line 37 of file LinAlg.cpp.

{
    if (!y.getRawVector())
        y.shallowCopy(x);
    VecCopy(x.getRawVector(), y.getRawVector());
}

References MathLib::PETScVector::getRawVector(), and MathLib::PETScVector::shallowCopy().

Referenced by GMSHPrefsDialog::GMSHPrefsDialog(), FileIO::Gocad::GocadSGridReader::GocadSGridReader(), GeoLib::Raster::Raster(), FileIO::Gocad::GocadSGridReader::addFaceSetQuad(), FileIO::Gocad::GocadSGridReader::addGocadPropertiesToMesh(), addIntegrationPointData(), addIntegrationPointMetaData(), MeshLib::MeshLayerMapper::addLayerToMesh(), MeshLib::addLayerToMesh(), MeshLib::addPropertyToMesh(), FileIO::SwmmInterface::addResultsToMesh(), MeshGeoToolsLib::appendLinesAlongPolylines(), ProcessLib::LIE::PostProcessTool::calculateTotalDisplacement(), NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeA(), NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeJacobian(), NumLib::TimeDiscretization::computeRelativeChangeFromPreviousTimestep(), ProcessLib::TimeLoop::computeTimeStepping(), MeshLib::VtkMeshConverter::convertTypedArray(), ProcessLib::LIE::PostProcessTool::copyPropertyValues(), MeshLib::createQuadraticOrderMesh(), ChemistryLib::createReactionRates(), BaseLib::excludeObjectCopy(), ProcessLib::ComponentTransport::ComponentTransportProcess::extrapolateIntegrationPointValuesToNodes(), flipRaster(), MeshLib::getBaseNodes(), ProcessLib::PhaseFieldIrreversibleDamageOracleBoundaryCondition::getEssentialBCValues(), ProcessLib::LIE::getFractureMatrixDataInMesh(), LayeredMeshGenerator::getMesh(), NumLib::SimpleMatrixVectorProvider::getVector(), NumLib::BackwardEuler::getWeightedOldX(), MaterialLib::Solids::MFront::MFront< DisplacementDim >::integrateStress(), main(), BaseLib::operator<<(), operator<<(), FileIO::Gocad::operator<<(), ChemistryLib::PhreeqcIOData::anonymous_namespace{PhreeqcIO.cpp}::operator<<(), ApplicationUtils::NodeWiseMeshPartitioner::partitionOtherMesh(), ProcessLib::HeatTransportBHE::HeatTransportBHEProcess::postTimestepConcreteProcess(), MeshLib::PropertyVector< T * >::print(), ApplicationUtils::NodeWiseMeshPartitioner::processPartition(), MathLib::TemplatePoint< T, DIM >::read(), FileIO::Gocad::GocadSGridReader::readElementPropertiesBinary(), FileIO::GMSInterface::readGMS3DMMesh(), FileIO::SwmmInterface::readSwmmInputToLineMesh(), FileIO::TetGenInterface::readTetGenMesh(), GeoLib::MinimalBoundingSphere::recurseCalculation(), ProcessLib::TimeLoop::setCoupledSolutions(), ProcessLib::Process::setInitialConditions(), DiagramList::setList(), ProcessLib::Deformation::solidMaterialInternalToSecondaryVariables(), ProcessLib::Deformation::solidMaterialInternalVariablesToIntegrationPointWriter(), NumLib::NonlinearSolver< NonlinearSolverTag::Newton >::solve(), NumLib::NonlinearSolver< NonlinearSolverTag::Picard >::solve(), NumLib::PETScNonlinearSolver::solve(), ProcessLib::TimeLoop::solveCoupledEquationSystemsByStaggeredScheme(), ProcessLib::ComponentTransport::ComponentTransportProcess::solveReactionEquation(), BaseLib::splitString(), ProcessLib::PhaseField::PhaseFieldProcess< DisplacementDim >::updateConstraints(), anonymous_namespace{IdentifySubdomainMesh.cpp}::updateOrCheckExistingSubdomainProperty(), MathLib::TemplatePoint< T, DIM >::write(), and writeDataToMesh().

finalizeAssembly() [1/3]

template<typename Matrix >

void MathLib::LinAlg::finalizeAssembly

(

Matrix &

)

finalizeAssembly() [2/3]

void MathLib::LinAlg::finalizeAssembly

(

PETScMatrix &

A

)

Definition at line 162 of file LinAlg.cpp.

{
    A.finalizeAssembly(MAT_FINAL_ASSEMBLY);
}

References MathLib::PETScMatrix::finalizeAssembly().

Referenced by detail::applyKnownSolutions(), NumLib::TimeDiscretizedODESystem< ODESystemTag::FirstOrderImplicitQuasilinear, NonlinearSolverTag::Newton >::assemble(), NumLib::TimeDiscretizedODESystem< ODESystemTag::FirstOrderImplicitQuasilinear, NonlinearSolverTag::Picard >::assemble(), NumLib::LocalLinearLeastSquaresExtrapolator::calculateResiduals(), NumLib::LocalLinearLeastSquaresExtrapolator::extrapolate(), ProcessLib::ComponentTransport::ComponentTransportProcess::extrapolateIntegrationPointValuesToNodes(), ProcessLib::Process::setInitialConditions(), and ProcessLib::SmallDeformation::writeMaterialForces().

finalizeAssembly() [3/3]

void MathLib::LinAlg::finalizeAssembly

(

PETScVector &

x

)

Definition at line 167 of file LinAlg.cpp.

{
    x.finalizeAssembly();
}

References MathLib::PETScVector::finalizeAssembly().

matMult() [1/2]

template<typename Matrix , typename Vector >

void MathLib::LinAlg::matMult	(	Matrix const &	A,
		Vector const &	x,
		Vector &	y
	)

Computes $y = A \cdot x$.

Note

x must not be the same object as y. This restirction has been chosen in order to fulfill the requirements of the respective PETSc function.

Definition at line 114 of file LinAlg.h.

{
    assert(&x != &y);
    y = A*x;
}

matMult() [2/2]

void MathLib::LinAlg::matMult	(	PETScMatrix const &	A,
		PETScVector const &	x,
		PETScVector &	y
	)

Definition at line 141 of file LinAlg.cpp.

{
    // TODO check sizes
    assert(&x != &y);
    if (!y.getRawVector())
        y.shallowCopy(x);
    MatMult(A.getRawMatrix(), x.getRawVector(), y.getRawVector());
}

References MathLib::PETScMatrix::getRawMatrix(), MathLib::PETScVector::getRawVector(), and MathLib::PETScVector::shallowCopy().

Referenced by NumLib::NonlinearSolver< NonlinearSolverTag::Picard >::calculateNonEquilibriumInitialResiduum(), NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeResidual(), ProcessLib::computeResiduum(), NumLib::NonlinearSolver< NonlinearSolverTag::Picard >::solve(), and ProcessLib::ComponentTransport::ComponentTransportProcess::solveReactionEquation().

matMultAdd() [1/2]

template<typename Matrix , typename Vector >

void MathLib::LinAlg::matMultAdd	(	Matrix const &	A,
		Vector const &	v1,
		Vector const &	v2,
		Vector &	v3
	)

Computes $v_3 = A \cdot v_1 + v_2$.

Note

x must not be the same object as y. This restirction has been chosen in order to fulfill the requirements of the respective PETSc function.

Definition at line 127 of file LinAlg.h.

{
    assert(&v1 != &v3);
    v3 = v2 + A*v1;
}

matMultAdd() [2/2]

void MathLib::LinAlg::matMultAdd	(	PETScMatrix const &	A,
		PETScVector const &	v1,
		PETScVector const &	v2,
		PETScVector &	v3
	)

Definition at line 151 of file LinAlg.cpp.

{
    // TODO check sizes
    assert(&v1 != &v3);
    if (!v3.getRawVector())
        v3.shallowCopy(v1);
    MatMultAdd(A.getRawMatrix(), v1.getRawVector(), v2.getRawVector(),
               v3.getRawVector());
}

References MathLib::PETScMatrix::getRawMatrix(), MathLib::PETScVector::getRawVector(), and MathLib::PETScVector::shallowCopy().

Referenced by NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeResidual(), ProcessLib::computeResiduum(), and NumLib::MatrixTranslatorGeneral< ODESystemTag::FirstOrderImplicitQuasilinear >::computeRhs().

norm()

template<typename MatrixOrVector >

double MathLib::LinAlg::norm	(	MatrixOrVector const &	x,
		MathLib::VecNormType	type
	)

Definition at line 88 of file LinAlg.h.

{
    switch (type) {
        case MathLib::VecNormType::NORM1:
            return norm1(x);
        case MathLib::VecNormType::NORM2:
            return norm2(x);
        case MathLib::VecNormType::INFINITY_N:
            return normMax(x);
        default:
            OGS_FATAL("Invalid norm type given.");
    }
}

References MathLib::INFINITY_N, norm1(), MathLib::NORM1, norm2(), MathLib::NORM2, normMax(), and OGS_FATAL.

Referenced by LayeredVolume::addLayerBoundaries(), MathLib::calcProjPntToLineAndDists(), NumLib::ConvergenceCriterionDeltaX::checkDeltaX(), NumLib::ConvergenceCriterionResidual::checkDeltaX(), NumLib::ConvergenceCriterionResidual::checkResidual(), NumLib::TimeDiscretization::computeRelativeChangeFromPreviousTimestep(), MathLib::KelvinVector::Invariants< KelvinVectorSize >::equivalentStress(), GeoLib::Polyline::getDistanceAlongPolyline(), GeoLib::Grid< POINT >::getNearestPoint(), MathLib::KelvinVector::Invariants< KelvinVectorSize >::J2(), MathLib::KelvinVector::Invariants< KelvinVectorSize >::J3(), norm1(), norm2(), normMax(), and GeoLib::PointVec::resetInternalDataStructures().

norm1() [1/2]

template<typename MatrixOrVector >

double MathLib::LinAlg::norm1

(

MatrixOrVector const &

x

)

Computes the Manhattan norm of x.

norm1() [2/2]

template<>

double MathLib::LinAlg::norm1

(

PETScVector const &

x

)

Definition at line 82 of file LinAlg.cpp.

{
    PetscScalar norm = 0.;
    VecNorm(x.getRawVector(), NORM_1, &norm);
    return norm;
}

References MathLib::PETScVector::getRawVector(), and norm().

Referenced by norm().

norm2() [1/2]

template<typename MatrixOrVector >

double MathLib::LinAlg::norm2

(

MatrixOrVector const &

x

)

Computes the Euclidean norm of x.

norm2() [2/2]

template<>

double MathLib::LinAlg::norm2

(

PETScVector const &

x

)

Definition at line 92 of file LinAlg.cpp.

{
    PetscScalar norm = 0.;
    VecNorm(x.getRawVector(), NORM_2, &norm);
    return norm;
}

References MathLib::PETScVector::getRawVector(), and norm().

Referenced by norm().

normMax() [1/2]

template<typename MatrixOrVector >

double MathLib::LinAlg::normMax

(

MatrixOrVector const &

x

)

Computes the maximum norm of x.

normMax() [2/2]

template<>

double MathLib::LinAlg::normMax

(

PETScVector const &

x

)

Definition at line 102 of file LinAlg.cpp.

{
    PetscScalar norm = 0.;
    VecNorm(x.getRawVector(), NORM_INFINITY, &norm);
    return norm;
}

References MathLib::PETScVector::getRawVector(), and norm().

Referenced by norm().

scale() [1/3]

template<typename MatrixOrVector >

void MathLib::LinAlg::scale	(	MatrixOrVector &	x,
		double const	a
	)

Scales x by a.

Definition at line 44 of file LinAlg.h.

{
    x *= a;
}

scale() [2/3]

void MathLib::LinAlg::scale	(	PETScMatrix &	A,
		double const	a
	)

Definition at line 117 of file LinAlg.cpp.

{
    MatScale(A.getRawMatrix(), a);
}

References MathLib::PETScMatrix::getRawMatrix().

scale() [3/3]

void MathLib::LinAlg::scale	(	PETScVector &	x,
		double const	a
	)

Definition at line 44 of file LinAlg.cpp.

{
    VecScale(x.getRawVector(), a);
}

References MathLib::PETScVector::getRawVector().

Referenced by ProcessLib::computeResiduum(), MaterialPropertyLib::Property::description(), MaterialPropertyLib::ClausiusClapeyron::dMolarMass(), MaterialPropertyLib::RelPermBrooksCorey::dValue(), MaterialPropertyLib::RelPermBrooksCoreyNonwettingPhase::dValue(), MaterialPropertyLib::RelPermLiakopoulos::dValue(), NumLib::BackwardEuler::getWeightedOldX(), VtkVisImageItem::Initialize(), main(), MaterialPropertyLib::ClausiusClapeyron::molarMass(), VtkVisTabWidget::on_scaleZ_textChanged(), ProcessLib::PhaseField::PhaseFieldProcess< DisplacementDim >::postNonLinearSolverConcreteProcess(), VtkVisTabWidget::setActiveItem(), VtkTextureOnSurfaceFilter::SetRaster(), VtkBGImageSource::SetRaster(), MaterialPropertyLib::Property::setScale(), VtkColorByHeightFilter::SetTableRangeScaling(), VtkVisPointSetItem::setTranslation(), MaterialPropertyLib::RelPermBrooksCorey::value(), MaterialPropertyLib::RelPermBrooksCoreyNonwettingPhase::value(), and MaterialPropertyLib::RelPermLiakopoulos::value().

set()

void MathLib::LinAlg::set	(	PETScVector &	x,
		double const	a
	)

Definition at line 32 of file LinAlg.cpp.

{
    VecSet(x.getRawVector(), a);
}

References MathLib::PETScVector::getRawVector().

Referenced by ProcessLib::ComponentTransport::intersection(), and ProcessLib::SmallDeformation::writeMaterialForces().

setLocalAccessibleVector()

void MathLib::LinAlg::setLocalAccessibleVector

(

PETScVector const &

x

)

Set local accessible vector in order to get entries. Call this before call operator[] or get(...) of x.

Definition at line 27 of file LinAlg.cpp.

{
    x.setLocalAccessibleVector();
}

References MathLib::PETScVector::setLocalAccessibleVector().

Referenced by ProcessLib::CoupledSolutionsForStaggeredScheme::CoupledSolutionsForStaggeredScheme(), ProcessLib::Process::assemble(), ProcessLib::Process::assembleWithJacobian(), NumLib::LocalLinearLeastSquaresExtrapolator::calculateResidualElement(), ProcessLib::Process::computeSecondaryVariable(), ProcessLib::Process::postIteration(), ProcessLib::TES::TESProcess::postIterationConcreteProcess(), ProcessLib::Process::postNonLinearSolver(), ProcessLib::Process::postTimestep(), ProcessLib::LIE::HydroMechanics::HydroMechanicsProcess< GlobalDim >::postTimestepConcreteProcess(), ProcessLib::Process::preIteration(), ProcessLib::Process::preTimestep(), ProcessLib::Process::setInitialConditions(), NumLib::transformVariableFromGlobalVector(), ProcessLib::PhaseField::PhaseFieldProcess< DisplacementDim >::updateConstraints(), and ProcessLib::SmallDeformation::writeMaterialForces().