Class UnconLeastSqLevenbergMarquardt_F64<S extends DMatrix>

java.lang.Object

org.ddogleg.optimization.GaussNewtonBase_F64<ConfigLevenbergMarquardt, HessianLeastSquares<S>>

org.ddogleg.optimization.lm.LevenbergMarquardt_F64<S, HessianLeastSquares<S>>

org.ddogleg.optimization.lm.UnconLeastSqLevenbergMarquardt_F64<S>

All Implemented Interfaces:

Serializable, IterativeOptimization, UnconstrainedLeastSquares<S>, UnconstrainedLeastSquaresBase<S, FunctionNtoMxN<S>>, VerbosePrint

public class UnconLeastSqLevenbergMarquardt_F64<S extends DMatrix>
extends LevenbergMarquardt_F64<S, HessianLeastSquares<S>>
implements UnconstrainedLeastSquares<S>

Implementation of LevenbergMarquardt_F64 for UnconstrainedLeastSquares.

See Also:

• Serialized Form

Nested Class Summary

Nested classes/interfaces inherited from class GaussNewtonBase_F64

GaussNewtonBase_F64.Mode

Field Summary

Fields

Modifier and Type	Field	Description
protected FunctionNtoMxN<S>	functionJacobian
protected FunctionNtoM	functionResiduals

Fields inherited from class LevenbergMarquardt_F64

diagOrig, diagStep, jacobianType, lambda, lossFunc, lossFuncGradient, math, MAX_LAMBDA, nu, residuals, storageLossGradient

Fields inherited from class GaussNewtonBase_F64

config, ftest_val, fx, gradient, gtest_val, hessian, hessianScaling, mode, p, postUpdateAdjuster, sameStateAsCost, totalFullSteps, totalSelectSteps, verbose, verboseLevel, x, x_next

Constructor Summary

Constructors

Constructor	Description
UnconLeastSqLevenbergMarquardt_F64(MatrixMath<S> math, HessianLeastSquares<S> hessian)

Method Summary

Modifier and Type	Method	Description
protected void	computeResiduals(DMatrixRMaj x, DMatrixRMaj residuals)	Computes the residuals at state 'x'
protected void	functionGradientHessian(DMatrixRMaj x, boolean sameStateAsResiduals, DMatrixRMaj gradient, HessianLeastSquares<S> hessian)	Computes the gradient and Hessian at 'x'.
double	getFunctionValue()	Returns the value of the objective function being evaluated at the current parameters value.
double[]	getParameters()	After each iteration this function can be called to get the current best set of parameters.
void	initialize(double[] initial, double ftol, double gtol)	Specify the initial set of parameters from which to start from.
boolean	isConverged()	Indicates if iteration stopped due to convergence or not.
boolean	isUpdated()	True if the parameter(s) being optimized have been updated
void	setFunction(FunctionNtoM function, @Nullable FunctionNtoMxN<S> jacobian)	Specifies a set of functions and their Jacobian.
void	setPostUpdate(AdjustArray_F64 adjuster)	After a step has been computed and applied this function is called with the new candidate state passed in.

Methods inherited from class LevenbergMarquardt_F64

acceptNewState, checkConvergenceFTest, computeStep, computeStep, configure, costFromResiduals, getJacobianType, initialize, maximumLambdaNu, setLoss, updateDerivates

Methods inherited from class GaussNewtonBase_F64

applyHessianScaling, checkConvergenceGTest, computeHessianScaling, computeHessianScaling, computePredictedReduction, initialize, iterate, mode, setVerbose, setVerbose, undoHessianScalingOnParameters

Methods inherited from class Object

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

Methods inherited from interface IterativeOptimization

iterate, setVerbose

Methods inherited from interface UnconstrainedLeastSquaresBase

getJacobianType, setLoss

Methods inherited from interface VerbosePrint

setVerbose

Field Details

functionResiduals

protected FunctionNtoM functionResiduals

functionJacobian

protected FunctionNtoMxN<S extends DMatrix> functionJacobian

Constructor Details

UnconLeastSqLevenbergMarquardt_F64

public UnconLeastSqLevenbergMarquardt_F64(MatrixMath<S> math,
 HessianLeastSquares<S> hessian)

Method Details

setFunction

public void setFunction(FunctionNtoM function,
 @Nullable
 @Nullable FunctionNtoMxN<S> jacobian)

Description copied from interface: UnconstrainedLeastSquaresBase

Specifies a set of functions and their Jacobian. See class description for documentation on output data format.

Specified by:

setFunction in interface UnconstrainedLeastSquaresBase<S extends DMatrix, FunctionNtoMxN<S extends DMatrix>>

Parameters:

function - Computes the output of M functions f_i(x) which take in N fit parameters as input.

jacobian - Computes the Jacobian of the M functions. If null a numerical Jacobian will be used.

setPostUpdate

public void setPostUpdate(AdjustArray_F64 adjuster)

Description copied from interface: UnconstrainedLeastSquaresBase

After a step has been computed and applied this function is called with the new candidate state passed in. It provides the user the opportunity to adjust the data in anyway which will not change the score. For example, if a vector is scale invariant you can ensure it always has a norm of 1.

Specified by:

setPostUpdate in interface UnconstrainedLeastSquaresBase<S extends DMatrix, FunctionNtoMxN<S extends DMatrix>>

Parameters:

adjuster - The new user provided adjuster

initialize

public void initialize(double[] initial,
 double ftol,
 double gtol)

Description copied from interface: UnconstrainedLeastSquaresBase

Specify the initial set of parameters from which to start from. Call after UnconstrainedLeastSquaresBase.setFunction(FunctionNtoM, FunctionJacobian) has been called.

Specified by:

initialize in interface UnconstrainedLeastSquaresBase<S extends DMatrix, FunctionNtoMxN<S extends DMatrix>>

Parameters:

initial - Initial parameters or guess with N elements..

ftol - Relative threshold for change in function value between iterations. 0 ≤ ftol ≤ 1. Try 1e-12

gtol - Absolute threshold for convergence based on the gradient's norm. 0 disables test. 0 ≤ gtol. Try 1e-12

getParameters

public double[] getParameters()

Description copied from interface: UnconstrainedLeastSquaresBase

After each iteration this function can be called to get the current best set of parameters.

Specified by:

getParameters in interface UnconstrainedLeastSquaresBase<S extends DMatrix, FunctionNtoMxN<S extends DMatrix>>

getFunctionValue

public double getFunctionValue()

Description copied from interface: UnconstrainedLeastSquaresBase

Returns the value of the objective function being evaluated at the current parameters value. If not supported then an exception is thrown.

Specified by:

getFunctionValue in interface UnconstrainedLeastSquaresBase<S extends DMatrix, FunctionNtoMxN<S extends DMatrix>>

Returns:

Objective function's value.

isUpdated

public boolean isUpdated()

Description copied from interface: IterativeOptimization

True if the parameter(s) being optimized have been updated

Specified by:

isUpdated in interface IterativeOptimization

Returns:

True if parameters have been updated

isConverged

public boolean isConverged()

Description copied from interface: IterativeOptimization

Indicates if iteration stopped due to convergence or not.

Specified by:

isConverged in interface IterativeOptimization

Returns:

True if iteration stopped because it converged.

functionGradientHessian

protected void functionGradientHessian(DMatrixRMaj x,
 boolean sameStateAsResiduals,
 DMatrixRMaj gradient,
 HessianLeastSquares<S> hessian)

Description copied from class: GaussNewtonBase_F64

Computes the gradient and Hessian at 'x'. If sameStateAsCost is true then it can be assumed that 'x' has not changed since the cost was last computed.

Specified by:

functionGradientHessian in class GaussNewtonBase_F64<ConfigLevenbergMarquardt, HessianLeastSquares<S extends DMatrix>>

Parameters:

sameStateAsResiduals - (Input) If true then when the cost (or residuals) was last called it had the same value of x

gradient - (Input) x

hessian - (Output) hessian

computeResiduals

protected void computeResiduals(DMatrixRMaj x,
 DMatrixRMaj residuals)

Description copied from class: LevenbergMarquardt_F64

Computes the residuals at state 'x'

Specified by:

computeResiduals in class LevenbergMarquardt_F64<S extends DMatrix, HessianLeastSquares<S extends DMatrix>>

Parameters:

x - (Input) state

residuals - (Output) residuals F(x) - Y