TasDREAM::LikelihoodGaussIsotropic Class Reference

Implements likelihood under the assumption of isotropic white noise. More...

#include <tsgDreamLikelyGaussian.hpp>

Inheritance diagram for TasDREAM::LikelihoodGaussIsotropic:

Collaboration diagram for TasDREAM::LikelihoodGaussIsotropic:

Public Member Functions
	LikelihoodGaussIsotropic ()
	Default constructor for convenience, an object constructed with the default cannot be used until setData() is called.
	LikelihoodGaussIsotropic (double variance, const std::vector< double > &data_mean, size_t num_observe=1)
	Constructs the class and calls setData().
	~LikelihoodGaussIsotropic ()=default
	Default destructor.
void	setData (double variance, const std::vector< double > &data_mean, size_t num_observe=1)
	Set the noise magnitude (varaince) the observed data (data_mean) and number of observations (num_observe). More...
void	getLikelihood (TypeSamplingForm form, const std::vector< double > &model, std::vector< double > &likely) const override final
	Compute the likelihood of a set of model outputs.
void	getLikelihood (TypeSamplingForm form, double const model[], int num_samples, double likely[]) const override final
	Overload for raw-arrays, for interface purposes mostly, e.g., python.
int	getNumOutputs () const override
	Returns the size of the data_mean vector (for error checking purposes).
void	write (std::ostream &os, int outputs_begin=0, int outputs_end=-1) const
	Writes the data for a portion of the outputs into a stream. More...
void	read (std::istream &is)
	Reads the data from a stream, assumes write() has been used first.
Public Member Functions inherited from TasDREAM::TasmanianLikelihood
	TasmanianLikelihood ()
	Empty default constructor.
virtual	~TasmanianLikelihood ()
	Empty virtual destructor.
virtual	operator std::function< void (TypeSamplingForm, const std::vector< double > &, std::vector< double > &)>() const
	Automatically convert the likelihood into input for TasDREAM::posterior().

Detailed Description

Implements likelihood under the assumption of isotropic white noise.

Gaussian Likelihood

The general formula for Gaussian likelihood is where the are the data observations, y is the model output, and is the noise covariance matrix.

Isotropic Gaussian

The simplest isotopic case of Gaussian likelihood assumes that the covariance matrix is scaled identity, i.e., the noise has no correlation and has the same magnitude for each model output, then inverting the covariance matrix reduces to multiplying by the inverse of the noise variance (magnitude). Also, the sum corresponding to the multiple data samples can be replaced by scaled operation on the data mean (average).

Member Function Documentation

setData()

void TasDREAM::LikelihoodGaussIsotropic::setData	(	double	variance,
		const std::vector< double > &	data_mean,
		size_t	num_observe = 1
	)

Set the noise magnitude (varaince) the observed data (data_mean) and number of observations (num_observe).

Set the parameters of the likelihood.

Parameters

variance	must be a positive number indicating the magnitude of the noise.
data_mean	must have the same size as the number of model outputs and hold the average of all measurements.
num_observe	must be a positive integer indicating the number of samples.

write()

void TasDREAM::LikelihoodGaussIsotropic::write ( std::ostream &  os, int  outputs_begin = 0, int  outputs_end = -1  ) const

inline

Writes the data for a portion of the outputs into a stream.

The likelihood object does not store the raw inputs to setData(), instead optimized data-structures are used. This method writes either entire likelihood or the optimized data for a portion of the outputs.

Parameters

os	is the stream where the data will be written.
outputs_begin	is the first output to include in the write process.
outputs_end	is one more than the last output to write, use -1 to indicate all outputs after output_begin.

This method is used by the MPI scatter likelihood template.

---

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

• DREAM/tsgDreamLikelyGaussian.hpp