#include <PartitionOfUnityMapping.hpp>

Inheritance diagram for precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >:

Collaboration diagram for precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >:

Public Member Functions
	PartitionOfUnityMapping (Mapping::Constraint constraint, int dimension, RADIAL_BASIS_FUNCTION_T function, Polynomial polynomial, unsigned int verticesPerCluster, double relativeOverlap, bool projectToInput)

void	computeMapping () final override

void	clear () final override
	Clears a computed mapping by deleting the content of the `_clusters` vector.

void	tagMeshFirstRound () final override
	tag the vertices required for the mapping

void	tagMeshSecondRound () final override
	nothing to do here

std::string	getName () const final override
	name of the pum mapping

Public Member Functions inherited from precice::mapping::Mapping
	Mapping (Constraint constraint, int dimensions, bool requiresGradientData, InitialGuessRequirement initialGuessRequirement)
	Constructor, takes mapping constraint.

Mapping &	operator= (Mapping &&)=delete

virtual	~Mapping ()=default
	Destructor, empty.

void	setMeshes (const mesh::PtrMesh &input, const mesh::PtrMesh &output)
	Sets input and output meshes carrying data to be mapped.

const mesh::PtrMesh &	getInputMesh () const

const mesh::PtrMesh &	getOutputMesh () const

Constraint	getConstraint () const
	Returns the constraint (consistent/conservative) of the mapping.

MeshRequirement	getInputRequirement () const
	Returns the requirement on the input mesh.

MeshRequirement	getOutputRequirement () const
	Returns the requirement on the output mesh.

bool	hasComputedMapping () const
	Returns true, if the mapping has been computed.

virtual bool	hasConstraint (const Constraint &constraint) const
	Checks whether the mapping has the given constraint or not.

bool	isScaledConsistent () const
	Returns true if mapping is a form of scaled consistent mapping.

bool	requiresInitialGuess () const
	Return true if the mapping requires an initial guess.

const Eigen::VectorXd &	initialGuess () const
	Return the provided initial guess of a mapping using an initialGuess.

Eigen::VectorXd &	initialGuess ()

bool	hasInitialGuess () const
	True if initialGuess().size() == 0.

void	map (int inputDataID, int outputDataID)

void	map (int inputDataID, int outputDataID, Eigen::VectorXd &initialGuess)

void	map (const time::Sample &input, Eigen::VectorXd &output)
	Maps an input Sample to output data from input mesh to output mesh.

void	map (const time::Sample &input, Eigen::VectorXd &output, Eigen::VectorXd &initialGuess)
	Maps an input Sample to output data from input mesh to output mesh, given an initialGuess.

virtual void	scaleConsistentMapping (const Eigen::VectorXd &input, Eigen::VectorXd &output, Constraint type) const
	Scales the consistently mapped output data such that the surface integral of the values on input mesh and output mesh are equal.

bool	requiresGradientData () const
	Returns whether the mapping requires gradient data.

Private Member Functions
virtual void	mapConservative (const time::Sample &inData, Eigen::VectorXd &outData) override
	Maps data using a conservative constraint.

virtual void	mapConsistent (const time::Sample &inData, Eigen::VectorXd &outData) override
	Maps data using a consistent constraint.

void	exportClusterCentersAsVTU (mesh::Mesh &centers)

Private Attributes
precice::logging::Logger	_log {"mapping::PartitionOfUnityMapping"}
	logger, as usual

std::vector< SphericalVertexCluster< RADIAL_BASIS_FUNCTION_T > >	_clusters
	main data container storing all the clusters, which need to be solved individually

RADIAL_BASIS_FUNCTION_T	_basisFunction
	Radial basis function type used in interpolation.

const unsigned int	_verticesPerCluster
	Input parameters provided by the user for the clustering algorithm:

const double	_relativeOverlap
	overlap of vertex clusters

const bool	_projectToInput
	toggles whether we project the cluster centers to the input mesh

double	_clusterRadius = 0
	derived parameter based on the input above: the radius of each cluster

Polynomial	_polynomial
	polynomial treatment of the RBF system

Additional Inherited Members
Public Types inherited from precice::mapping::Mapping
enum	Constraint { CONSISTENT , CONSERVATIVE , SCALED_CONSISTENT_SURFACE , SCALED_CONSISTENT_VOLUME }
	Specifies additional constraints for a mapping. More...

enum class	MeshRequirement { UNDEFINED = 0 , VERTEX = 1 , FULL = 2 }
	Specifies requirements for the input and output meshes of a mapping. More...

enum class	InitialGuessRequirement : bool { Required = true , None = false }
	Specifies whether the mapping requires an initial guess. More...

Protected Member Functions inherited from precice::mapping::Mapping
mesh::PtrMesh	input () const
	Returns pointer to input mesh.

mesh::PtrMesh	output () const
	Returns pointer to output mesh.

void	setInputRequirement (MeshRequirement requirement)
	Sets the mesh requirement for the input mesh.

void	setOutputRequirement (MeshRequirement requirement)
	Sets the mesh requirement for the output mesh.

int	getDimensions () const

Protected Attributes inherited from precice::mapping::Mapping
bool	_hasComputedMapping = false
	Flag to indicate whether computeMapping() has been called.

bool	_requiresGradientData
	Flag if gradient data is required for the mapping.

Detailed Description

template<typename RADIAL_BASIS_FUNCTION_T>
class precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >

Mapping using partition of unity decomposition strategies: The class here inherits from the Mapping class and orchestrates the partitions (called vertex clusters) in order to represent a partition of unity. This means in particular that the class computes the weights for the evaluation vertices and the necessary association between evaluation vertices and the clusters during initialization and traverses through all vertex clusters when evaluating the mapping.

Definition at line 29 of file PartitionOfUnityMapping.hpp.

Constructor & Destructor Documentation

◆ PartitionOfUnityMapping()

template<typename RADIAL_BASIS_FUNCTION_T >

precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >::PartitionOfUnityMapping	(	Mapping::Constraint	constraint,
		int	dimension,
		RADIAL_BASIS_FUNCTION_T	function,
		Polynomial	polynomial,
		unsigned int	verticesPerCluster,
		double	relativeOverlap,
		bool	projectToInput )

Constructor, which mostly sets the mesh connectivity requirements and initializes member variables.

Parameters

[in]	constraint	Specifies mapping to be consistent or conservative.
[in]	dimension	Dimensionality of the meshes
[in]	function	Radial basis function type used in interpolation
[in]	polynomial	The handling of the polynomial in the RBF system. Valid choices are 'off' and 'separate'
[in]	verticesPerCluster	Target number of vertices to be clustered together
[in]	relativeOverlap	Overlap between clusters: The parameter here determines the distance between two cluster centers, given the cluster radius (already determined through `verticesPerCluster` ). A value of 1 would correspond to no distance between cluster centers (i.e. completely overlapping clusters), 0 to distance of 2 x radius between clusters centers.
[in]	projectToInput	if enabled, places the cluster centers at the closest vertex of the input mesh. See also mapping::impl::createClustering()

Definition at line 117 of file PartitionOfUnityMapping.hpp.

Here is the call graph for this function:

Member Function Documentation

◆ clear()

template<typename RADIAL_BASIS_FUNCTION_T >

void precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >::clear ( )

finaloverridevirtual

Clears a computed mapping by deleting the content of the _clusters vector.

Implements precice::mapping::Mapping.

Definition at line 415 of file PartitionOfUnityMapping.hpp.

◆ computeMapping()

template<typename RADIAL_BASIS_FUNCTION_T >

void precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >::computeMapping ( )

finaloverridevirtual

Computes the clustering for the partition of unity method and fills the _clusters vector, which allows to travers through all vertex cluster computed. Each vertex cluster in the vector directly computes local mapping matrices and matrix decompositions. In addition, the method computes the normalized weights (Shepard's method) for the partition of unity method and stores them directly in each relevant vertex cluster. In debug mode, the function also exports the partition centers as a separate mesh for visualization purpose.

Implements precice::mapping::Mapping.

Definition at line 143 of file PartitionOfUnityMapping.hpp.

Here is the call graph for this function:

◆ exportClusterCentersAsVTU()

template<typename RADIAL_BASIS_FUNCTION_T >

void precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >::exportClusterCentersAsVTU ( mesh::Mesh & centers )

private

export the center vertices of all clusters as a mesh with some additional data on it such as vertex count only enabled in debug builds and mainly for debugging purpose

Definition at line 366 of file PartitionOfUnityMapping.hpp.

Here is the call graph for this function:

◆ getName()

template<typename RADIAL_BASIS_FUNCTION_T >

std::string precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >::getName ( ) const

finaloverridevirtual

name of the pum mapping

Implements precice::mapping::Mapping.

Definition at line 425 of file PartitionOfUnityMapping.hpp.

◆ mapConservative()

template<typename RADIAL_BASIS_FUNCTION_T >

void precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >::mapConservative	(	const time::Sample &	inData,
		Eigen::VectorXd &	outData )

overrideprivatevirtual

Maps data using a conservative constraint.

Parameters

[in]	input	Sample to map data from
[in]	output	Values to map to

If requiresInitialGuess(), then the initial guess is available via initialGuess(). Provide a new initial guess by overwriting it. The mapping has full control over its size.

See also: For mappings requiring an initialGuess: initialGuess() hasInitialGuess()

Implements precice::mapping::Mapping.

Definition at line 264 of file PartitionOfUnityMapping.hpp.

Here is the call graph for this function:

◆ mapConsistent()

template<typename RADIAL_BASIS_FUNCTION_T >

void precice::mapping::PartitionOfUnityMapping< RADIAL_BASIS_FUNCTION_T >::mapConsistent	(	const time::Sample &	inData,
		Eigen::VectorXd &	outData )

overrideprivatevirtual

Maps data using a consistent constraint.

Parameters

[in]	input	Sample to map data from
[in]	output	Values to map to

If requiresInitialGuess(), then the initial guess is available via initialGuess(). Provide a new initial guess by overwriting it. The mapping has full control over its size.