com.lgc.wsh.opt
Class ScalarSolver

java.lang.Object
  com.lgc.wsh.opt.ScalarSolver

public class ScalarSolver
extends java.lang.Object

Search a single variable for a value that minimizes a function

Nested Class Summary

static interface

ScalarSolver.Function Implement a function of one variable to be minimized

Constructor Summary

ScalarSolver(ScalarSolver.Function function)
Constructor

Method Summary

double

solve(double scalarMin, double scalarMax, double okError, double okFraction, int numberIterations, Monitor monitor) Minimize a function of scalar and return the optimum value.

Methods inherited from class java.lang.Object

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

Constructor Detail

ScalarSolver

public ScalarSolver(ScalarSolver.Function function)

Constructor

Parameters:

function - Objective function to be minimized.

Method Detail

solve

public double solve(double scalarMin,
                    double scalarMax,
                    double okError,
                    double okFraction,
                    int numberIterations,
                    Monitor monitor)

Minimize a function of scalar and return the optimum value.

Parameters:

scalarMin - The minimum value allowed for the argument scalar.

scalarMax - The maximum value allowed for the argument scalar.

okError - The unknown error in scalar should be less than this fraction of the range: dscalar/(scalarMax-scalarMin) <= okError, where dscalar is the error bound on the returned value of scalar.

okFraction - The error in scalar should be less than this fraction of the minimum possible range for scalar: dscalar <= okFraction*(scalar-scalarMin), where dscalar is the error bound on the returned value of scalar.

numberIterations - The maximum number of iterations if greater than 6. The optimization performs at least 6 iterations -- the minimum necessary for a genuinely parabolic function. I recommend at least 20.

monitor - For reporting progress. Ignored if null.

Returns:

The optimum value minimizing the function.