vec_math
Class AbstractGradientModel

java.lang.Object
  vec_math.AbstractDataModel
      vec_math.AbstractGradientModel

All Implemented Interfaces:

DataModel, GradientModel

Direct Known Subclasses:

HarmonicFit.HarmonicModel, LevenbergMarquardt.LMSineModel, LevenbergMarquardt.SinusModel, MultipleFrequencyFit.FourierComponents, OrbitModel, Overscan.WifsipOverscan

public abstract class AbstractGradientModel
extends AbstractDataModel
implements GradientModel

This class is designed to work together with the LevenbergMarquardt class. Data models that should be optimized with the latter class should extends this class. Additionally to an abstract data model, we also have to supply the gradient matrix via GradientModel.getGradientMatrix(vec_math.VectorG).

Nested Class Summary

Nested classes/interfaces inherited from class vec_math.AbstractDataModel

AbstractDataModel.Construct, AbstractDataModel.LocalM

Field Summary

Fields inherited from class vec_math.AbstractDataModel

sigma, x, y

Constructor Summary

protected

AbstractGradientModel(VectorG[] times, double[] y, double[] err) Chains.

Method Summary

static void	check(AbstractGradientModel fit, VectorG exact) Check the gradient matrix with the exact solution.
static void	check(AbstractGradientModel fit, VectorG exact, double da) Check the gradient matrix with the exact solution.
static VectorG	getNegativeChi2HalfGradient(GradientModel grad, VectorG a) Returns half the negativ gradient of chi-square of this model with respect to the model parameters.
VectorG	getNegativeChi2HalfGradient(VectorG a) Returns half the negativ gradient of chi-square of this model with respect to the model parameters.
static QuadMatrix	getPseudoHessian(GradientModel grad, VectorG a) Returns the pseudo-hessian matrix consiting of the first derivatives only for the given set of parameters.
QuadMatrix	getPseudoHessian(VectorG a) Returns the pseudo-hessian matrix consiting of the first derivatives only for the given set of parameters.
static VectorG	LevenbergMarquardtSolver(GradientModel exp, VectorG start) We use a LevenbergMarquardt in standard form to solve for a minimum model.

Methods inherited from class vec_math.AbstractDataModel

getChiSquareModel, getChiSquareModel, getLorentzianModel, getMeasureCount, getMeasurementErrors, getMeasures, getModel, getModel, getResiduals, getResiduals, getRms, getRobustModel, getTimes, getTotalErrors, LorentzianSolver, RobustSolver, SimplexSolver

Methods inherited from class java.lang.Object

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

Methods inherited from interface vec_math.GradientModel

getGradientMatrix

Methods inherited from interface vec_math.DataModel

evaluateModel, getMeasureCount, getMeasurementErrors, getMeasures, getModel, getParameterCount, getResiduals, getTimes, getTotalErrors

Constructor Detail

AbstractGradientModel

protected AbstractGradientModel(VectorG[] times,
                                double[] y,
                                double[] err)

Chains.

Method Detail

getPseudoHessian

public static QuadMatrix getPseudoHessian(GradientModel grad,
                                          VectorG a)

Returns the pseudo-hessian matrix consiting of the first derivatives only for the given set of parameters.

getPseudoHessian

public QuadMatrix getPseudoHessian(VectorG a)

Returns the pseudo-hessian matrix consiting of the first derivatives only for the given set of parameters.

getNegativeChi2HalfGradient

public VectorG getNegativeChi2HalfGradient(VectorG a)

Returns half the negativ gradient of chi-square of this model with respect to the model parameters.

getNegativeChi2HalfGradient

public static VectorG getNegativeChi2HalfGradient(GradientModel grad,
                                                  VectorG a)

Returns half the negativ gradient of chi-square of this model with respect to the model parameters.

check

public static void check(AbstractGradientModel fit,
                         VectorG exact)

Check the gradient matrix with the exact solution.

check

public static void check(AbstractGradientModel fit,
                         VectorG exact,
                         double da)

Check the gradient matrix with the exact solution.

LevenbergMarquardtSolver

public static VectorG LevenbergMarquardtSolver(GradientModel exp,
                                               VectorG start)

We use a LevenbergMarquardt in standard form to solve for a minimum model.