gtsam  3.2.0
gtsam
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
gtsam::JacobianFactor Class Reference

Detailed Description

A Gaussian factor in the squared-error form.

JacobianFactor implements a Gaussian, which has quadratic negative log-likelihood

\[ E(x) = \frac{1}{2} (Ax-b)^T \Sigma^{-1} (Ax-b) \]

where \( \Sigma \) is a diagonal covariance matrix. The matrix \( A \), r.h.s. vector \( b \), and diagonal noise model \( \Sigma \) are stored in this class.

This factor represents the sum-of-squares error of a linear measurement function, and is created upon linearization of a NoiseModelFactor, which in turn is a sum-of-squares factor with a nonlinear measurement function.

Here is an example of how this factor represents a sum-of-squares error:

Letting \( h(x) \) be a linear measurement prediction function, \( z \) be the actual observed measurement, the residual is

\[ f(x) = h(x) - z . \]

If we expect noise with diagonal covariance matrix \( \Sigma \) on this measurement, then the negative log-likelihood of the Gaussian induced by this measurement model is

\[ E(x) = \frac{1}{2} (h(x) - z)^T \Sigma^{-1} (h(x) - z) . \]

Because \( h(x) \) is linear, we can write it as

\[ h(x) = Ax + e \]

and thus we have

\[ E(x) = \frac{1}{2} (Ax-b)^T \Sigma^{-1} (Ax-b) \]

where \( b = z - e \).

This factor can involve an arbitrary number of variables, and in the above example \( x \) would almost always be only be a subset of the variables in the entire factor graph. There are special constructors for 1-, 2-, and 3- way JacobianFactors, and additional constructors for creating n-way JacobianFactors. The Jacobian matrix \( A \) is passed to these constructors in blocks, for example, for a 2-way factor, the constructor would accept \( A1 \) and \( A2 \), as well as the variable indices \( j1 \) and \( j2 \) and the negative log-likelihood represented by this factor would be

\[ E(x) = \frac{1}{2} (A_1 x_{j1} + A_2 x_{j2} - b)^T \Sigma^{-1} (A_1 x_{j1} + A_2 x_{j2} - b) . \]

+ Inheritance diagram for gtsam::JacobianFactor:

Public Member Functions

 JacobianFactor (const GaussianFactor &gf)
 Convert from other GaussianFactor.
 
 JacobianFactor (const JacobianFactor &jf)
 Copy constructor.
 
 JacobianFactor (const HessianFactor &hf)
 Conversion from HessianFactor (does Cholesky to obtain Jacobian matrix)
 
 JacobianFactor ()
 default constructor for I/O
 
 JacobianFactor (const Vector &b_in)
 Construct Null factor.
 
 JacobianFactor (Key i1, const Matrix &A1, const Vector &b, const SharedDiagonal &model=SharedDiagonal())
 Construct unary factor.
 
 JacobianFactor (Key i1, const Matrix &A1, Key i2, const Matrix &A2, const Vector &b, const SharedDiagonal &model=SharedDiagonal())
 Construct binary factor.
 
 JacobianFactor (Key i1, const Matrix &A1, Key i2, const Matrix &A2, Key i3, const Matrix &A3, const Vector &b, const SharedDiagonal &model=SharedDiagonal())
 Construct ternary factor.
 
template<typename TERMS >
 JacobianFactor (const TERMS &terms, const Vector &b, const SharedDiagonal &model=SharedDiagonal())
 Construct an n-ary factor. More...
 
template<typename KEYS >
 JacobianFactor (const KEYS &keys, const VerticalBlockMatrix &augmentedMatrix, const SharedDiagonal &sigmas=SharedDiagonal())
 Constructor with arbitrary number keys, and where the augmented matrix is given all together instead of in block terms. More...
 
 JacobianFactor (const GaussianFactorGraph &graph, boost::optional< const Ordering & > ordering=boost::none, boost::optional< const VariableSlots & > variableSlots=boost::none)
 Build a dense joint factor from all the factors in a factor graph. More...
 
virtual ~JacobianFactor ()
 Virtual destructor.
 
virtual GaussianFactor::shared_ptr clone () const
 Clone this JacobianFactor.
 
virtual void print (const std::string &s="", const KeyFormatter &formatter=DefaultKeyFormatter) const
 
virtual bool equals (const GaussianFactor &lf, double tol=1e-9) const
 Equals for testable.
 
Vector unweighted_error (const VectorValues &c) const
 
Vector error_vector (const VectorValues &c) const
 (A*x-b)
 
virtual double error (const VectorValues &c) const
 (A*x-b)/sigma
 
virtual Matrix augmentedInformation () const
 0.5*(A*x-b)'D(A*x-b) More...
 
virtual Matrix information () const
 Return the non-augmented information matrix represented by this GaussianFactor.
 
virtual VectorValues hessianDiagonal () const
 Return the diagonal of the Hessian for this factor.
 
virtual void hessianDiagonal (double *d) const
 Return the diagonal of the Hessian for this factor (raw memory version)
 
virtual std::map< Key, Matrix > hessianBlockDiagonal () const
 Return the block diagonal of the Hessian for this factor.
 
virtual std::pair< Matrix, Vector > jacobian () const
 Returns (dense) A,b pair associated with factor, bakes in the weights.
 
std::pair< Matrix, Vector > jacobianUnweighted () const
 Returns (dense) A,b pair associated with factor, does not bake in weights.
 
virtual Matrix augmentedJacobian () const
 Return (dense) matrix associated with factor. More...
 
Matrix augmentedJacobianUnweighted () const
 Return (dense) matrix associated with factor. More...
 
const VerticalBlockMatrixmatrixObject () const
 Return the full augmented Jacobian matrix of this factor as a VerticalBlockMatrix object. More...
 
VerticalBlockMatrixmatrixObject ()
 Mutable access to the full augmented Jacobian matrix of this factor as a VerticalBlockMatrix object. More...
 
virtual GaussianFactor::shared_ptr negate () const
 Construct the corresponding anti-factor to negate information stored stored in this factor. More...
 
virtual bool empty () const
 Check if the factor is empty. More...
 
bool isConstrained () const
 is noise model constrained ?
 
virtual DenseIndex getDim (const_iterator variable) const
 Return the dimension of the variable pointed to by the given key iterator todo: Remove this in favor of keeping track of dimensions with variables?
 
size_t rows () const
 return the number of rows in the corresponding linear system
 
size_t cols () const
 return the number of columns in the corresponding linear system
 
const SharedDiagonal & get_model () const
 get a copy of model
 
SharedDiagonal & get_model ()
 get a copy of model (non-const version)
 
const constBVector getb () const
 Get a view of the r.h.s. More...
 
constABlock getA (const_iterator variable) const
 Get a view of the A matrix for the variable pointed to by the given key iterator.
 
constABlock getA () const
 Get a view of the A matrix, not weighted by noise.
 
BVector getb ()
 Get a view of the r.h.s. More...
 
ABlock getA (iterator variable)
 Get a view of the A matrix for the variable pointed to by the given key iterator (non-const version)
 
ABlock getA ()
 Get a view of the A matrix.
 
Vector operator* (const VectorValues &x) const
 Return A*x.
 
void transposeMultiplyAdd (double alpha, const Vector &e, VectorValues &x) const
 x += A'*e. More...
 
void multiplyHessianAdd (double alpha, const VectorValues &x, VectorValues &y) const
 y += alpha * A'*A*x
 
void multiplyHessianAdd (double alpha, const double *x, double *y, std::vector< size_t > keys) const
 y += alpha * A'*A*x
 
void multiplyHessianAdd (double alpha, const double *x, double *y) const
 y += alpha * A'*A*x
 
VectorValues gradientAtZero () const
 A'*b for Jacobian.
 
virtual void gradientAtZero (double *d) const
 A'*b for Jacobian, eta for Hessian (raw memory version)
 
JacobianFactor whiten () const
 Return a whitened version of the factor, i.e. More...
 
std::pair< boost::shared_ptr
< GaussianConditional >
, boost::shared_ptr
< JacobianFactor > > 
eliminate (const Ordering &keys)
 Eliminate the requested variables. More...
 
void setModel (bool anyConstrained, const Vector &sigmas)
 set noiseModel correctly
 
boost::shared_ptr
< GaussianConditional
splitConditional (size_t nrFrontals)
 splits a pre-factorized factor into a conditional, and changes the current factor to be the remaining component. More...
 
- Public Member Functions inherited from gtsam::GaussianFactor
 GaussianFactor ()
 Default constructor creates empty factor.
 
template<typename CONTAINER >
 GaussianFactor (const CONTAINER &keys)
 Construct from container of keys. More...
 
virtual ~GaussianFactor ()
 Destructor.
 
- Public Member Functions inherited from gtsam::Factor
Key front () const
 First key.
 
Key back () const
 Last key.
 
const_iterator find (Key key) const
 find
 
const FastVector< Key > & keys () const
 Access the factor's involved variable keys.
 
const_iterator begin () const
 Iterator at beginning of involved variable keys.
 
const_iterator end () const
 Iterator at end of involved variable keys.
 
size_t size () const
 
void print (const std::string &s="Factor", const KeyFormatter &formatter=DefaultKeyFormatter) const
 print
 
void printKeys (const std::string &s="Factor", const KeyFormatter &formatter=DefaultKeyFormatter) const
 print only keys
 
FastVector< Key > & keys ()
 
iterator begin ()
 Iterator at beginning of involved variable keys.
 
iterator end ()
 Iterator at end of involved variable keys.
 

Public Types

typedef JacobianFactor This
 Typedef to this class.
 
typedef GaussianFactor Base
 Typedef to base class.
 
typedef boost::shared_ptr< Thisshared_ptr
 shared_ptr to this class
 
typedef VerticalBlockMatrix::Block ABlock
 
typedef
VerticalBlockMatrix::constBlock 
constABlock
 
typedef ABlock::ColXpr BVector
 
typedef constABlock::ConstColXpr constBVector
 
- Public Types inherited from gtsam::GaussianFactor
typedef GaussianFactor This
 This class.
 
typedef boost::shared_ptr< Thisshared_ptr
 shared_ptr to this class
 
typedef Factor Base
 Our base class.
 
- Public Types inherited from gtsam::Factor
typedef FastVector< Key >::iterator iterator
 Iterator over keys.
 
typedef FastVector< Key >
::const_iterator 
const_iterator
 Const iterator over keys.
 

Protected Member Functions

template<typename TERMS >
void fillTerms (const TERMS &terms, const Vector &b, const SharedDiagonal &noiseModel)
 Internal function to fill blocks and set dimensions.
 
- Protected Member Functions inherited from gtsam::Factor
 Factor ()
 Default constructor for I/O.
 
template<typename CONTAINER >
 Factor (const CONTAINER &keys)
 Construct factor from container of keys. More...
 
template<typename ITERATOR >
 Factor (ITERATOR first, ITERATOR last)
 Construct factor from iterator keys. More...
 
bool equals (const This &other, double tol=1e-9) const
 check equality
 

Protected Attributes

VerticalBlockMatrix Ab_
 
noiseModel::Diagonal::shared_ptr model_
 
- Protected Attributes inherited from gtsam::Factor
FastVector< Keykeys_
 The keys involved in this factor.
 

Friends

class boost::serialization::access
 Serialization function.
 
GTSAM_EXPORT std::pair
< boost::shared_ptr
< GaussianConditional >
, boost::shared_ptr
< JacobianFactor > > 
EliminateQR (const GaussianFactorGraph &factors, const Ordering &keys)
 

Constructor & Destructor Documentation

template<typename TERMS >
gtsam::JacobianFactor::JacobianFactor ( const TERMS &  terms,
const Vector &  b,
const SharedDiagonal &  model = SharedDiagonal() 
)

Construct an n-ary factor.

Template Parameters
TERMSA container whose value type is std::pair<Key, Matrix>, specifying the collection of keys and matrices making up the factor.
template<typename KEYS >
gtsam::JacobianFactor::JacobianFactor ( const KEYS &  keys,
const VerticalBlockMatrix augmentedMatrix,
const SharedDiagonal &  sigmas = SharedDiagonal() 
)

Constructor with arbitrary number keys, and where the augmented matrix is given all together instead of in block terms.

Note that only the active view of the provided augmented matrix is used, and that the matrix data is copied into a newly-allocated matrix in the constructed factor.

gtsam::JacobianFactor::JacobianFactor ( const GaussianFactorGraph graph,
boost::optional< const Ordering & >  ordering = boost::none,
boost::optional< const VariableSlots & >  variableSlots = boost::none 
)
explicit

Build a dense joint factor from all the factors in a factor graph.

If a VariableSlots structure computed for graph is already available, providing it will reduce the amount of computation performed.

Member Function Documentation

Matrix gtsam::JacobianFactor::augmentedInformation ( ) const
virtual

0.5*(A*x-b)'D(A*x-b)

Return the augmented information matrix represented by this GaussianFactor. The augmented information matrix contains the information matrix with an additional column holding the information vector, and an additional row holding the transpose of the information vector. The lower-right entry contains the constant error term (when \( \delta x = 0 \)). The augmented information matrix is described in more detail in HessianFactor, which in fact stores an augmented information matrix.

Implements gtsam::GaussianFactor.

Matrix gtsam::JacobianFactor::augmentedJacobian ( ) const
virtual

Return (dense) matrix associated with factor.

The returned system is an augmented matrix: [A b] weights are baked in

Implements gtsam::GaussianFactor.

Matrix gtsam::JacobianFactor::augmentedJacobianUnweighted ( ) const

Return (dense) matrix associated with factor.

The returned system is an augmented matrix: [A b] weights are not baked in

std::pair< boost::shared_ptr< GaussianConditional >, boost::shared_ptr< JacobianFactor > > gtsam::JacobianFactor::eliminate ( const Ordering keys)

Eliminate the requested variables.

virtual bool gtsam::JacobianFactor::empty ( ) const
inlinevirtual

Check if the factor is empty.

TODO: How should this be defined?

Implements gtsam::GaussianFactor.

const constBVector gtsam::JacobianFactor::getb ( ) const
inline

Get a view of the r.h.s.

vector b, not weighted by noise

BVector gtsam::JacobianFactor::getb ( )
inline

Get a view of the r.h.s.

vector b (non-const version)

const VerticalBlockMatrix& gtsam::JacobianFactor::matrixObject ( ) const
inline

Return the full augmented Jacobian matrix of this factor as a VerticalBlockMatrix object.

VerticalBlockMatrix& gtsam::JacobianFactor::matrixObject ( )
inline

Mutable access to the full augmented Jacobian matrix of this factor as a VerticalBlockMatrix object.

GaussianFactor::shared_ptr gtsam::JacobianFactor::negate ( ) const
virtual

Construct the corresponding anti-factor to negate information stored stored in this factor.

Returns
a HessianFactor with negated Hessian matrices

Implements gtsam::GaussianFactor.

GaussianConditional::shared_ptr gtsam::JacobianFactor::splitConditional ( size_t  nrFrontals)

splits a pre-factorized factor into a conditional, and changes the current factor to be the remaining component.

Performs same operation as eliminate(), but without running QR.

void gtsam::JacobianFactor::transposeMultiplyAdd ( double  alpha,
const Vector &  e,
VectorValues x 
) const

x += A'*e.

If x is initially missing any values, they are created and assumed to start as zero vectors.

JacobianFactor gtsam::JacobianFactor::whiten ( ) const

Return a whitened version of the factor, i.e.

with unit diagonal noise model.


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