Package jdistlib.inference

Class BayesianModel

java.lang.Object

jdistlib.inference.BayesianModel

All Implemented Interfaces:

DifferentiableLogDensity, GradientProvider, LogDensity, PointwiseLogLikelihood

public final class BayesianModel
extends Object
implements DifferentiableLogDensity, GradientProvider, PointwiseLogLikelihood

Compiled named model evaluated on an unconstrained state space.

Method Summary

Modifier and Type	Method	Description
double[]	constrain(double[] state)
ModelData	data()
int	dimension()
ModelEvaluator	evaluator()
List<FactorSpec>	factors()
ModelGraph	graph()
boolean	hasAnalyticGradient()
double[]	initialState()
double	logDensity(double[] state)
double	logDensityAndGradient(double[] state, double[] gradient)	Evaluates the log density and replaces gradient with its gradient.
ObservationMetadata	observationMetadata()
Map<String,ParameterSpec>	parameters()
double[]	pointwiseLogLikelihood(double[] state)
ModelState	state(double[] unconstrainedState)	Returns a named constrained view backed by a fresh transformed state.
double[]	unconstrain(Map<String,double[]> values)	Converts a complete map of named constrained parameter values to sampler space.
BayesianModel	withPointwiseLikelihood(ObservationMetadata metadata, PointwiseLogLikelihoodEvaluator evaluator)	Returns a copy of this model with an additional pointwise likelihood evaluator.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface jdistlib.inference.PointwiseLogLikelihood

extractPointwiseLogLikelihood

Method Details

dimension

public int dimension()

initialState

public double[] initialState()

parameters

public Map<String,ParameterSpec> parameters()

factors

public List<FactorSpec> factors()

data

public ModelData data()

graph

public ModelGraph graph()

evaluator

public ModelEvaluator evaluator()

hasAnalyticGradient

public boolean hasAnalyticGradient()

Specified by:

hasAnalyticGradient in interface GradientProvider

observationMetadata

public ObservationMetadata observationMetadata()

Specified by:

observationMetadata in interface PointwiseLogLikelihood

pointwiseLogLikelihood

public double[] pointwiseLogLikelihood(double[] state)

Specified by:

pointwiseLogLikelihood in interface PointwiseLogLikelihood

withPointwiseLikelihood

public BayesianModel withPointwiseLikelihood(ObservationMetadata metadata,
 PointwiseLogLikelihoodEvaluator evaluator)

Returns a copy of this model with an additional pointwise likelihood evaluator.

logDensity

public double logDensity(double[] state)

Specified by:

logDensity in interface DifferentiableLogDensity

Specified by:

logDensity in interface LogDensity

logDensityAndGradient

public double logDensityAndGradient(double[] state,
 double[] gradient)

Description copied from interface: DifferentiableLogDensity

Evaluates the log density and replaces gradient with its gradient.

Specified by:

logDensityAndGradient in interface DifferentiableLogDensity

constrain

public double[] constrain(double[] state)

unconstrain

public double[] unconstrain(Map<String,double[]> values)

Converts a complete map of named constrained parameter values to sampler space.

state

public ModelState state(double[] unconstrainedState)

Returns a named constrained view backed by a fresh transformed state.