Base class for edges connecting exactly two vertices in the graph. More...

Inheritance diagram for BaseBinaryEdge< D, E, VertexXi, VertexXj >:

Collaboration diagram for BaseBinaryEdge< D, E, VertexXi, VertexXj >:

Public Types
typedef BaseEdge< D, E >::ErrorVector	ErrorVector
	The error vector type.
typedef Eigen::Map< Eigen::Matrix< g_type, Dj, Di >, Eigen::Matrix< g_type, Di, Dj >::Flags &Eigen::PacketAccessBit ? Eigen::Aligned :Eigen::Unaligned >	HessianBlockTransposedType
	Mapped transposed Hessian block type.
typedef Eigen::Map< Eigen::Matrix< g_type, Di, Dj >, Eigen::Matrix< g_type, Di, Dj >::Flags &Eigen::PacketAccessBit ? Eigen::Aligned :Eigen::Unaligned >	HessianBlockType
	Mapped Hessian block type.
typedef BaseEdge< D, E >::InformationType	InformationType
	The information matrix type.
typedef Eigen::Matrix< g_type, D, Di >::AlignedMapType	JacobianXiOplusType
	Jacobian type w.r.t. vertex Xi.
typedef Eigen::Matrix< g_type, D, Dj >::AlignedMapType	JacobianXjOplusType
	Jacobian type w.r.t. vertex Xj.
typedef BaseEdge< D, E >::Measurement	Measurement
	The measurement type.
typedef VertexXi	VertexXiType
	Type of the first (source) vertex.
typedef VertexXj	VertexXjType
	Type of the second (target) vertex.
Public Types inherited from BaseEdge< D, E >
typedef Eigen::Matrix< g_type, t_dims, 1 >	ErrorVector
	Fixed-size error vector type.
typedef Eigen::Matrix< g_type, t_dims, t_dims >	InformationType
	Fixed-size information (inverse covariance) matrix type.
typedef E	Measurement
	The measurement type stored by this edge.

Public Member Functions
	BaseBinaryEdge ()
	Constructs an empty binary edge with null-mapped Hessian and Jacobian memory.
bool	allVerticesFixed () const final override
	Returns true if both vertices are fixed.
void	constructQuadraticForm ()
	Constructs the quadratic form (Hessian blocks and gradient) from the linearized edge.
void	constructQuadraticForm2 ()
	Constructs the quadratic form using an optimized evaluation order.
const JacobianXiOplusType &	jacobianOplusXi () const
	returns the result of the linearization in the manifold space for the node xi
const JacobianXjOplusType &	jacobianOplusXj () const
	returns the result of the linearization in the manifold space for the node xj
virtual void	linearizeOplus ()
	Numerically linearizes the oplus operator by finite differences for both vertices.
void	linearizeOplusAndConstructQuadraticForm (JacobianWorkspace &jacobianWorkspace) final override
	Linearizes the oplus operator and constructs the quadratic form in one step.
void	mapHessianMemory (g_type *d, int i, int j, bool rowMajor) final override
	Maps the Hessian block memory to an external buffer.
virtual void	setVertex (uint04 i, HyperGraph::HGVertex *v) final override
	Sets vertex i to the given pointer, with type-checking.
virtual const HyperGraph::HGVertex *	vertex (uint04 i) const final override
	Returns a const pointer to vertex i (0 or 1).
virtual HyperGraph::HGVertex *	vertex (uint04 i) final override
	Returns a mutable pointer to vertex i (0 or 1).
virtual uint04	vertexCount () const final override
	Returns the number of vertices (always 2 for binary edges).
Public Member Functions inherited from BaseEdge< D, E >
	BaseEdge ()
	Default constructor.
g_type	chi2 () const final override
	Computes the chi-squared error: e^T * Omega * e.
virtual sint04	dimension () const final override
	Returns the dimension of the error vector.
const ErrorVector &	error () const
	Returns a const reference to the error vector.
const InformationType &	information () const
	Returns a const reference to the information matrix.
const g_type *	informationData () const final override
	Returns a const pointer to the raw information matrix data.
const Measurement &	measurement () const
	accessor functions for the measurement represented by the edge
int	rank () const
	Returns the rank (dimension) of the error.
g_type	scalerInformation () const
	information matrix of the constraint
void	setInformation (const InformationType &information)
	Sets the information matrix.
void	setMeasurement (const Measurement &m)
	Sets the measurement for this edge.
Public Member Functions inherited from OptimizableGraph::OGEdge
	OGEdge ()
	Default constructor.
void	clearRobustKernel ()
	Removes the robust kernel from this edge.
virtual void	computeError ()=0
	Computes the error of the edge and stores it internally.
long long	internalId () const
	the internal ID of the edge
int	level () const
	returns the level of the edge
RobustKernel *	robustKernel () const
	if NOT NULL, error of this edge will be robustifed with the kernel
void	setLevel (int l)
	sets the level of the edge
void	setRobustKernel (RobustKernel &ptr)
	specify the robust kernel to be used in this edge
Public Member Functions inherited from HyperGraph::HGEdge
	HGEdge (int id=-1)
	creates and empty edge with no vertices
int	id () const
	returns the id of this edge
void	setId (int id)
	sets the id of this edge

Static Public Attributes
static constexpr int	Di = VertexXiType::Dimension
	Dimension of the first vertex.
static constexpr int	Dj = VertexXjType::Dimension
	Dimension of the second vertex.
Static Public Attributes inherited from BaseEdge< D, E >
static constexpr sint04	Dimension
	Compile-time dimension of the error vector.

Protected Attributes
HessianBlockType	_hessian
	Mapped Hessian block for the off-diagonal (i,j) entry.
bool	_hessianRowMajor
	Whether the Hessian block is stored in row-major order.
HessianBlockTransposedType	_hessianTransposed
	Mapped transposed Hessian block.
JacobianXiOplusType	_jacobianOplusXi
	Jacobian of the error w.r.t. vertex Xi.
JacobianXjOplusType	_jacobianOplusXj
	Jacobian of the error w.r.t. vertex Xj.
VertexXi *	m_a = nullptr
	Pointer to the first (source) vertex.
VertexXj *	m_b = nullptr
	Pointer to the second (target) vertex.
Protected Attributes inherited from BaseEdge< D, E >
ErrorVector	_error
	The current error vector.
InformationType	_information
	The information (inverse covariance) matrix.
Measurement	_measurement
	The stored measurement for this edge.
Protected Attributes inherited from OptimizableGraph::OGEdge
int	_internalId = 0
	Internal sequential id assigned on insertion.
int	_level = 0
	Optimization level for multi-level optimization.
RobustKernel *	_robustKernel = nullptr
	Optional robust kernel for this edge.
Protected Attributes inherited from HyperGraph::HGEdge
int	_id
	unique id

Additional Inherited Members
Protected Member Functions inherited from BaseEdge< D, E >
InformationType	robustInformation (const Eigen::Vector3< g_type > &rho) const
	calculate the robust information matrix by updating the information matrix of the error

Detailed Description

template<sint04 D, typename E, typename VertexXi, typename VertexXj>
class BaseBinaryEdge< D, E, VertexXi, VertexXj >

Base class for edges connecting exactly two vertices in the graph.

Template Parameters

D	The dimension of the error vector.
E	The measurement type.
VertexXi	The type of the first vertex.
VertexXj	The type of the second vertex.

Definition at line 17 of file base_binary_edge.h.

Member Function Documentation

◆ linearizeOplusAndConstructQuadraticForm()

template<sint04 D, typename E, typename VertexXi, typename VertexXj>

void BaseBinaryEdge< D, E, VertexXi, VertexXj >::linearizeOplusAndConstructQuadraticForm ( JacobianWorkspace & jacobianWorkspace )

inlinefinaloverridevirtual

Linearizes the oplus operator and constructs the quadratic form in one step.

Parameters

[in] jacobianWorkspace Workspace providing temporary memory for Jacobian computation.

Implements OptimizableGraph::OGEdge.

Definition at line 247 of file base_binary_edge.h.

References _jacobianOplusXi, _jacobianOplusXj, constructQuadraticForm(), Di, Dj, and linearizeOplus().

◆ mapHessianMemory()

template<sint04 D, typename E, typename VertexXi, typename VertexXj>

void BaseBinaryEdge< D, E, VertexXi, VertexXj >::mapHessianMemory	(	g_type *	d,
		int	i,
		int	j,
		bool	rowMajor )

inlinefinaloverridevirtual

Maps the Hessian block memory to an external buffer.

Parameters

[in]	d	Pointer to the external memory.
[in]	i	Index of vertex i (unused, asserted to be 0).
[in]	j	Index of vertex j (unused, asserted to be 1).
[in]	rowMajor	If true, writes the transposed block.

Implements OptimizableGraph::OGEdge.

Definition at line 335 of file base_binary_edge.h.

References _hessian, _hessianRowMajor, and _hessianTransposed.

◆ setVertex()

template<sint04 D, typename E, typename VertexXi, typename VertexXj>

virtual void BaseBinaryEdge< D, E, VertexXi, VertexXj >::setVertex	(	uint04	i,
		HyperGraph::HGVertex *	v )

inlinefinaloverridevirtual

Sets vertex i to the given pointer, with type-checking.

Parameters

[in]	i	The vertex index (0 or 1).
[in]	v	The vertex pointer to assign.

Implements HyperGraph::HGEdge.

Definition at line 74 of file base_binary_edge.h.

References cast(), m_a, and m_b.

◆ vertex() [1/2]

template<sint04 D, typename E, typename VertexXi, typename VertexXj>

virtual const HyperGraph::HGVertex * BaseBinaryEdge< D, E, VertexXi, VertexXj >::vertex ( uint04 i ) const

inlinefinaloverridevirtual

Returns a const pointer to vertex i (0 or 1).

Parameters

[in] i The vertex index.

Implements HyperGraph::HGEdge.

Definition at line 55 of file base_binary_edge.h.

References m_a, and m_b.

◆ vertex() [2/2]

template<sint04 D, typename E, typename VertexXi, typename VertexXj>

virtual HyperGraph::HGVertex * BaseBinaryEdge< D, E, VertexXi, VertexXj >::vertex ( uint04 i )

inlinefinaloverridevirtual

Returns a mutable pointer to vertex i (0 or 1).

Parameters

[in] i The vertex index.

Implements HyperGraph::HGEdge.

Definition at line 64 of file base_binary_edge.h.

References m_a, and m_b.

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

GraphOptimization/Headers/base_binary_edge.h

Public Types

Public Member Functions

Static Public Attributes

Protected Attributes

Additional Inherited Members

Detailed Description

Member Function Documentation

◆ linearizeOplusAndConstructQuadraticForm()

◆ mapHessianMemory()

◆ setVertex()

◆ vertex() [1/2]

◆ vertex() [2/2]