com.itemanalysis.psychometrics.optimization

Class QNMinimizer.QNInfo

java.lang.Object

com.itemanalysis.psychometrics.optimization.QNMinimizer.QNInfo

Enclosing class:

QNMinimizer

public class QNMinimizer.QNInfo
extends java.lang.Object

The QNInfo class is used to store information about the Quasi Newton update. it holds all the s,y pairs, updates the diagonal and scales everything as needed.

Field Summary

Fields

Modifier and Type	Field and Description
double[]	d
QNMinimizer.eScaling	scaleOpt

Constructor Summary

Constructors

Constructor and Description
QNInfo()
QNInfo(int size)
QNInfo(java.util.List<double[]> sList, java.util.List<double[]> yList)

Method Summary

Modifier and Type	Method and Description
double[]	applyInitialHessian(double[] x)
void	clear()
void	free()
double	getRho(int ind)
double[]	getS(int ind)
double[]	getY(int ind)
void	setHistory(java.util.List<double[]> sList, java.util.List<double[]> yList)
int	size()
int	update(double[] newX, double[] x, double[] newGrad, double[] grad, double step)
int	update(double[] newS, double[] newY, double yy, double sy, double sg, double step)
void	useDiagonalScaling()
void	useScalarScaling()

Methods inherited from class java.lang.Object

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

Field Detail

d

public double[] d

scaleOpt

public QNMinimizer.eScaling scaleOpt

Constructor Detail

QNInfo

public QNInfo(int size)

QNInfo

public QNInfo()

QNInfo

public QNInfo(java.util.List<double[]> sList,
              java.util.List<double[]> yList)

Method Detail

size

public int size()

getRho

public double getRho(int ind)

getS

public double[] getS(int ind)

getY

public double[] getY(int ind)

useDiagonalScaling

public void useDiagonalScaling()

useScalarScaling

public void useScalarScaling()

free

public void free()

clear

public void clear()

setHistory

public void setHistory(java.util.List<double[]> sList,
                       java.util.List<double[]> yList)

applyInitialHessian

public double[] applyInitialHessian(double[] x)

update

public int update(double[] newX,
                  double[] x,
                  double[] newGrad,
                  double[] grad,
                  double step)
           throws QNMinimizer.SurpriseConvergence

Throws:

QNMinimizer.SurpriseConvergence

update

public int update(double[] newS,
                  double[] newY,
                  double yy,
                  double sy,
                  double sg,
                  double step)