#include <Gaussian.h>

Inheritance diagram for muq::Modeling::Gaussian:

Detailed Description

Definition at line 14 of file Gaussian.h.

Public Types
enum	Mode { Covariance , Precision }
	Are we specifying the mean, covariance matrix, or precision matrix. More...

enum	ExtraInputs { None = 1 << 0 , Mean = 1 << 1 , DiagCovariance = 1 << 2 , DiagPrecision = 1 << 3 , FullCovariance = 1 << 4 , FullPrecision = 1 << 5 }

typedef uint8_t	InputMask

Public Member Functions
	Gaussian (unsigned int dim, InputMask extraInputs=ExtraInputs::None)

	Gaussian (Eigen::VectorXd const &mu, InputMask extraInputs=ExtraInputs::None)
	Construct a Gaussian with scaled identity covariance/precision. More...

	Gaussian (Eigen::VectorXd const &mu, Eigen::MatrixXd const &obj, Gaussian::Mode mode=Gaussian::Mode::Covariance, InputMask extraInputs=ExtraInputs::None)
	Construct a Gaussian by specifying both the mean and covariance or precision matrix. More...

virtual	~Gaussian ()=default

Mode	GetMode () const

virtual Eigen::MatrixXd	ApplyCovariance (Eigen::Ref< const Eigen::MatrixXd > const &x) const override

virtual Eigen::MatrixXd	ApplyPrecision (Eigen::Ref< const Eigen::MatrixXd > const &x) const override

virtual Eigen::MatrixXd	ApplyCovSqrt (Eigen::Ref< const Eigen::MatrixXd > const &x) const override

virtual Eigen::MatrixXd	ApplyPrecSqrt (Eigen::Ref< const Eigen::MatrixXd > const &x) const override

Eigen::MatrixXd	GetCovariance () const
	Get the covariance. More...

Eigen::MatrixXd	GetPrecision () const

void	SetCovariance (Eigen::MatrixXd const &newCov)
	Set the covariance matrix. More...

void	SetPrecision (Eigen::MatrixXd const &newPrec)
	Set the precision matrix. More...

Gaussian::InputMask	GetInputTypes () const

std::shared_ptr< Gaussian >	Condition (Eigen::MatrixXd const &obsMat, Eigen::VectorXd const &data, Eigen::MatrixXd const &obsCov) const
	Returns a new Gaussian distribution conditioned on a linear observation. More...

void	ResetHyperparameters (ref_vector< Eigen::VectorXd > const &params) override

virtual double	LogDeterminant () const override

Public Member Functions inherited from muq::Modeling::GaussianBase
	GaussianBase (unsigned int dim)

	GaussianBase (unsigned int dim, Eigen::VectorXi const &hyperSizesIn)

	GaussianBase (Eigen::VectorXd const &meanIn)

	GaussianBase (Eigen::VectorXd const &meanIn, Eigen::VectorXi const &hyperSizesIn)

virtual	~GaussianBase ()=default

virtual unsigned int	Dimension () const

virtual Eigen::VectorXd const &	GetMean () const

virtual void	SetMean (Eigen::VectorXd const &newMu)

virtual Eigen::VectorXd	GradLogDensityImpl (unsigned int wrt, ref_vector< Eigen::VectorXd > const &inputs) override

virtual Eigen::VectorXd	ApplyLogDensityHessianImpl (unsigned int wrt1, unsigned int wrt2, ref_vector< Eigen::VectorXd > const &inputs, Eigen::VectorXd const &vec) override

Public Member Functions inherited from muq::Modeling::Distribution
	Distribution (unsigned int varSizeIn, Eigen::VectorXi const &hyperSizesIn=Eigen::VectorXi())

virtual	~Distribution ()=default

virtual double	LogDensity (ref_vector< Eigen::VectorXd > const &inputs)
	Evaluate the log-density. More...

virtual double	LogDensity (std::vector< Eigen::VectorXd > const &inputs)

	VARIADIC_TO_REFVECTOR (LogDensity, Eigen::VectorXd, double)

virtual Eigen::VectorXd	GradLogDensity (unsigned int wrt, std::vector< Eigen::VectorXd > const &inputs)

virtual Eigen::VectorXd	GradLogDensity (unsigned int wrt, ref_vector< Eigen::VectorXd > const &inputs)

template<typename... Args>
Eigen::VectorXd	GradLogDensity (unsigned int wrt, Args... args)

virtual Eigen::VectorXd	ApplyLogDensityHessian (unsigned int const inWrt1, unsigned int const inWrt2, ref_vector< Eigen::VectorXd > const &input, Eigen::VectorXd const &vec)

virtual Eigen::VectorXd	ApplyLogDensityHessian (unsigned int const inWrt1, unsigned int const inWrt2, std::vector< Eigen::VectorXd > const &input, Eigen::VectorXd const &vec)

Eigen::VectorXd	Sample (ref_vector< Eigen::VectorXd > const &inputs)
	Sample the distribution. More...

Eigen::VectorXd	Sample (std::vector< Eigen::VectorXd > const &inputs)

Eigen::VectorXd	Sample ()
	Sample the distribution with no inputs. More...

	VARIADIC_TO_REFVECTOR (Sample, Eigen::VectorXd, Eigen::VectorXd)

std::shared_ptr< Density >	AsDensity ()
	Returns a density built from this distribution. More...

std::shared_ptr< RandomVariable >	AsVariable ()
	Returns a random variable built from this distribution. More...

Additional Inherited Members
Public Attributes inherited from muq::Modeling::Distribution
const unsigned int	varSize

const Eigen::VectorXi	hyperSizes

Member Typedef Documentation

◆ InputMask

typedef uint8_t muq::Modeling::Gaussian::InputMask

Definition at line 35 of file Gaussian.h.

Member Enumeration Documentation

◆ ExtraInputs

enum muq::Modeling::Gaussian::ExtraInputs

Enumerator
None
Mean
DiagCovariance
DiagPrecision
FullCovariance
FullPrecision

Definition at line 27 of file Gaussian.h.

◆ Mode

enum muq::Modeling::Gaussian::Mode

Are we specifying the mean, covariance matrix, or precision matrix.

Enumerator
Covariance	We are specifying the covariance.
Precision	We are specifying the precision.

Definition at line 18 of file Gaussian.h.

Constructor & Destructor Documentation

◆ Gaussian() [1/3]

Gaussian::Gaussian	(	unsigned int	dim,
		InputMask	extraInputs = `ExtraInputs::None`
	)

Definition at line 8 of file Gaussian.cpp.

References ComputeNormalization().

◆ Gaussian() [2/3]

Gaussian::Gaussian	(	Eigen::VectorXd const &	mu,
		InputMask	extraInputs = `ExtraInputs::None`
	)

Construct a Gaussian with scaled identity covariance/precision.

Parameters

[in] mu The mean

Definition at line 17 of file Gaussian.cpp.

References ComputeNormalization().

◆ Gaussian() [3/3]

Gaussian::Gaussian	(	Eigen::VectorXd const &	mu,
		Eigen::MatrixXd const &	obj,
		Gaussian::Mode	mode = `Gaussian::Mode::Covariance`,
		InputMask	extraInputs = `ExtraInputs::None`
	)

Construct a Gaussian by specifying both the mean and covariance or precision matrix.

Parameters

[in]	mu	The mean vector
[in]	obj	A matrix holding either the covariance or precision
[in]	mode	A flag indicating whether the matrix should be treated as the covariance or precision

Definition at line 27 of file Gaussian.cpp.

References CheckInputTypes(), ComputeNormalization(), covPrec, muq::Modeling::GaussianBase::mean, mode, and sqrtCovPrec.

◆ ~Gaussian()

virtual muq::Modeling::Gaussian::~Gaussian ( )

virtualdefault

Member Function Documentation

◆ ApplyCovariance()

Eigen::MatrixXd Gaussian::ApplyCovariance ( Eigen::Ref< const Eigen::MatrixXd > const & x ) const

overridevirtual

Applies the covariance matrix to a vector \(x\)

Parameters

[in] x A reference to the input vector

Returns: \(\Sigma x\), where \(\Sigma\) is the covariance matrix.

Implements muq::Modeling::GaussianBase.

Definition at line 248 of file Gaussian.cpp.

References covPrec, mode, and sqrtCovPrec.

Referenced by Condition(), muq::SamplingAlgorithms::SMMALAProposal::LogDensity(), and muq::SamplingAlgorithms::SMMALAProposal::Sample().

◆ ApplyCovSqrt()

Eigen::MatrixXd Gaussian::ApplyCovSqrt ( Eigen::Ref< const Eigen::MatrixXd > const & x ) const

overridevirtual

Applies a matrix square root of the covariance to a vector x.

Parameters

[in] x A reference to the input vector

Returns: \(\Sigma^{1/2} x\), where \(\Sigma^{1/2}\) is a square root of the covariance matrix (e.g., Cholesky factor).

Implements muq::Modeling::GaussianBase.

Definition at line 215 of file Gaussian.cpp.

References covPrec, mode, and sqrtCovPrec.

◆ ApplyPrecision()

Eigen::MatrixXd Gaussian::ApplyPrecision ( Eigen::Ref< const Eigen::MatrixXd > const & x ) const

overridevirtual

Applies the precision matrix (inverse covariance) to a vector \(x\)

Parameters

[in] x A reference to the input vector

Returns: \(\Sigma^{-1} x\), where \(\Sigma\) is the covariance matrix.

Implements muq::Modeling::GaussianBase.

Definition at line 264 of file Gaussian.cpp.

References covPrec, mode, and sqrtCovPrec.

◆ ApplyPrecSqrt()

Eigen::MatrixXd Gaussian::ApplyPrecSqrt ( Eigen::Ref< const Eigen::MatrixXd > const & x ) const

overridevirtual

Applies a matrix square root of the precision to a vector x. If \(\Sigma= L L^T\) for a covariance matrix \(\Sigma\) and a square root \(L\), then \(\Sigma^{-1} = L^{-T}L^{-1}\) is a decomposition of the precision matrix and \(L^{-T}\) is a square root of the precision matrix. This function returns the action of \(L^{-T}\) on a vector \(x\).

Parameters

[in] x A reference to the input vector

Returns: \(L^{-T} x\)

Implements muq::Modeling::GaussianBase.

Definition at line 231 of file Gaussian.cpp.

References covPrec, mode, and sqrtCovPrec.

◆ CheckInputTypes()

void Gaussian::CheckInputTypes	(	InputMask	extraInputs,
		Mode	mode
	)

staticprotected

Definition at line 48 of file Gaussian.cpp.

References mode.

Referenced by Gaussian().

◆ ComputeNormalization()

void Gaussian::ComputeNormalization ( )

protected

Compute the distribution's scaling constant.

Returns: Scaling constant

Definition at line 126 of file Gaussian.cpp.

References covPrec, logDet, mode, and sqrtCovPrec.

Referenced by Gaussian(), ResetHyperparameters(), SetCovariance(), and SetPrecision().

◆ Condition()

std::shared_ptr< Gaussian > Gaussian::Condition	(	Eigen::MatrixXd const &	obsMat,
		Eigen::VectorXd const &	data,
		Eigen::MatrixXd const &	obsCov
	)		const

Returns a new Gaussian distribution conditioned on a linear observation.

Definition at line 282 of file Gaussian.cpp.

References ApplyCovariance(), muq::Modeling::GaussianBase::Dimension(), GetCovariance(), muq::Modeling::GaussianBase::mean, and nlohmann::to_string().

◆ GetCovariance()

Eigen::MatrixXd Gaussian::GetCovariance ( ) const

Get the covariance.

Returns: The covariance

Definition at line 91 of file Gaussian.cpp.

References covPrec, muq::Modeling::GaussianBase::mean, and mode.

Referenced by Condition().

◆ GetExtraSizes()

Eigen::VectorXi Gaussian::GetExtraSizes	(	unsigned	dim,
		InputMask	extraInputs
	)

staticprotected

Definition at line 65 of file Gaussian.cpp.

◆ GetInputTypes()

Gaussian::InputMask muq::Modeling::Gaussian::GetInputTypes ( ) const

inline

Definition at line 88 of file Gaussian.h.

References inputTypes.

◆ GetMode()

Mode muq::Modeling::Gaussian::GetMode ( ) const

inline

Definition at line 61 of file Gaussian.h.

References mode.

◆ GetPrecision()

Eigen::MatrixXd Gaussian::GetPrecision ( ) const

Definition at line 108 of file Gaussian.cpp.

References covPrec, muq::Modeling::GaussianBase::mean, and mode.

◆ LogDeterminant()

virtual double muq::Modeling::Gaussian::LogDeterminant ( ) const

inlineoverridevirtual

Return the log determinant of the covariance matrix. Defaults to 0.0, so this should be overridden by child classes that are used in applications where the density needs to known completely, not only up to a normalizing constant.

Reimplemented from muq::Modeling::GaussianBase.

Definition at line 97 of file Gaussian.h.

References logDet.

◆ ModeFromExtras()

Gaussian::Mode Gaussian::ModeFromExtras ( InputMask extraInputs )

staticprotected

Definition at line 56 of file Gaussian.cpp.

◆ ResetHyperparameters()

void Gaussian::ResetHyperparameters ( ref_vector< Eigen::VectorXd > const & params )

overridevirtual

Process the hyperparameter inputs. This should be overridden by children that have extra inputs.

Parameters

[in] params A vector of hyperparameter vectors.

Reimplemented from muq::Modeling::GaussianBase.

Definition at line 153 of file Gaussian.cpp.

References ComputeNormalization(), covPrec, inputTypes, muq::Modeling::GaussianBase::mean, sqrtCovPrec, and nlohmann::to_string().

◆ SetCovariance()

void Gaussian::SetCovariance ( Eigen::MatrixXd const & newCov )

Set the covariance matrix.

Parameters

[in] newcov The new covariance

Definition at line 182 of file Gaussian.cpp.

References ComputeNormalization(), covPrec, muq::Modeling::GaussianBase::mean, mode, and sqrtCovPrec.

◆ SetPrecision()

void Gaussian::SetPrecision ( Eigen::MatrixXd const & newPrec )

Set the precision matrix.

Parameters

[in] newprec The new precision

Definition at line 199 of file Gaussian.cpp.

References ComputeNormalization(), covPrec, muq::Modeling::GaussianBase::mean, mode, and sqrtCovPrec.

Member Data Documentation

◆ covPrec

Eigen::MatrixXd muq::Modeling::Gaussian::covPrec

protected

Definition at line 119 of file Gaussian.h.

Referenced by ApplyCovariance(), ApplyCovSqrt(), ApplyPrecision(), ApplyPrecSqrt(), ComputeNormalization(), Gaussian(), GetCovariance(), GetPrecision(), ResetHyperparameters(), SetCovariance(), and SetPrecision().

◆ inputTypes

Gaussian::InputMask muq::Modeling::Gaussian::inputTypes

protected

What form do the extra inputs take? Just the mean, or the mean and covariance?

Definition at line 116 of file Gaussian.h.

Referenced by GetInputTypes(), and ResetHyperparameters().

◆ logDet

double muq::Modeling::Gaussian::logDet

protected

The log determinant of the covariance matrix.

Definition at line 125 of file Gaussian.h.

Referenced by ComputeNormalization(), and LogDeterminant().

◆ mode

Gaussian::Mode muq::Modeling::Gaussian::mode

protected

Have we specified the covariance or the precision.

Definition at line 113 of file Gaussian.h.

Referenced by ApplyCovariance(), ApplyCovSqrt(), ApplyPrecision(), ApplyPrecSqrt(), CheckInputTypes(), ComputeNormalization(), Gaussian(), GetCovariance(), GetMode(), GetPrecision(), SetCovariance(), and SetPrecision().

◆ sqrtCovPrec

Eigen::LLT<Eigen::MatrixXd> muq::Modeling::Gaussian::sqrtCovPrec

protected

Definition at line 122 of file Gaussian.h.

Referenced by ApplyCovariance(), ApplyCovSqrt(), ApplyPrecision(), ApplyPrecSqrt(), ComputeNormalization(), Gaussian(), ResetHyperparameters(), SetCovariance(), and SetPrecision().

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

doxygen_prep/MUQ/Modeling/Distributions/Gaussian.h
doxygen_prep/modules/Modeling/src/Distributions/Gaussian.cpp

Detailed Description

Public Types

Public Member Functions

Additional Inherited Members

Member Typedef Documentation

◆ InputMask

Member Enumeration Documentation

◆ ExtraInputs

◆ Mode

Constructor & Destructor Documentation

◆ Gaussian() [1/3]

◆ Gaussian() [2/3]

◆ Gaussian() [3/3]

◆ ~Gaussian()

Member Function Documentation

◆ ApplyCovariance()

◆ ApplyCovSqrt()

◆ ApplyPrecision()

◆ ApplyPrecSqrt()

◆ CheckInputTypes()

◆ ComputeNormalization()

◆ Condition()

◆ GetCovariance()

◆ GetExtraSizes()

◆ GetInputTypes()

◆ GetMode()

◆ GetPrecision()

◆ LogDeterminant()

◆ ModeFromExtras()

◆ ResetHyperparameters()

◆ SetCovariance()

◆ SetPrecision()

Member Data Documentation

◆ covPrec

◆ inputTypes

◆ logDet

◆ mode

◆ sqrtCovPrec