org.ddogleg.optimization.impl

Interface TrustRegionStep

All Known Implementing Classes:

CauchyStep, DoglegStepF, DoglegStepFtF

public interface TrustRegionStep

Computes the next step to take in the trust region approach and the expected reduction. This interface has been designed to facilitate efficient computation.

The predicted reduction is defined as L(0) - L(h)
where L(h) is a linear model:
L(h) = 0.5||f(x) + J(x)*h||²
which can be reduced to:
L(0) - L(h) = -f^T(x)*J(x)*h - 0.5*h^TJ(x)^T*J(x)*h

Method Summary

Modifier and Type	Method and Description
void	computeStep(double regionRadius, org.ejml.data.DenseMatrix64F step) Computes the next step to take for a given trust region.
void	init(int numParam, int numFunctions) Initialize internal data structures.
boolean	isMaxStep() Was a step equal to the regionRadius taken?
double	predictedReduction() Returns the predicted reduction for the step.
void	setInputs(org.ejml.data.DenseMatrix64F x, org.ejml.data.DenseMatrix64F residuals, org.ejml.data.DenseMatrix64F J, org.ejml.data.DenseMatrix64F gradient, double fx) Specifies the state of the system being optimized.

Method Detail

init

void init(int numParam,
          int numFunctions)

Initialize internal data structures. Only needs to be called once.

Parameters:

numParam - Number of parameters being optimizes. This is the length of 'x'

numFunctions - Number of functions. Number of outputs to f(x)

setInputs

void setInputs(org.ejml.data.DenseMatrix64F x,
               org.ejml.data.DenseMatrix64F residuals,
               org.ejml.data.DenseMatrix64F J,
               org.ejml.data.DenseMatrix64F gradient,
               double fx)

Specifies the state of the system being optimized. Call before computeStep(double, org.ejml.data.DenseMatrix64F).

Parameters:

x - Sample point being considered.

residuals - Function output: f(x)

J - Jacobian: J(x)

gradient - Gradient: J^T(x)*f(x)

fx - Residual at x: 0.5*f^T(x)*f(x)

computeStep

void computeStep(double regionRadius,
                 org.ejml.data.DenseMatrix64F step)

Computes the next step to take for a given trust region. Must invoke setInputs(org.ejml.data.DenseMatrix64F, org.ejml.data.DenseMatrix64F, org.ejml.data.DenseMatrix64F, org.ejml.data.DenseMatrix64F, double) before, but it only needs to be called once.

Parameters:

regionRadius - Size of the trust region.

step - Output, the computed step.

predictedReduction

double predictedReduction()

Returns the predicted reduction for the step. A linear model is used to predict the reduction. See class description for

Returns:

The predicted reduction.

isMaxStep

boolean isMaxStep()

Was a step equal to the regionRadius taken?

Returns:

true if maximum step and false if less than the maximum step