edu.northwestern.at.utils.math.distributions
Class ChiSquare

java.lang.Object
  edu.northwestern.at.utils.math.distributions.ChiSquare

public class ChiSquare
extends java.lang.Object

Chi-square distribution functions.

Constructor Summary

protected

ChiSquare() Make class non-instantiable but inheritable.

Method Summary

static double	chisquare(double chisq, double df) Compute probability for chi-square distribution.
static double	chisquareInverse(double p, double df) Compute percentage point for chi-square distribution.

Methods inherited from class java.lang.Object

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

Constructor Detail

ChiSquare

protected ChiSquare()

Make class non-instantiable but inheritable.

Method Detail

chisquare

public static double chisquare(double chisq,
                               double df)

Compute probability for chi-square distribution.

Parameters:

chisq - Percentage point of chi-square distribution

df - Degrees of freedom

Returns:

The corresponding probabiity for the chi-square distribution.

The probability is determined using the relationship between the incomplete gamma CDF and the chi-square distribution:

chisqprob(chisq,df) = incompleteGamma( chisq/2, df/2 )

The result is accurate to about 14 decimal digits.

chisquareInverse

public static double chisquareInverse(double p,
                                      double df)

Compute percentage point for chi-square distribution.

Parameters:

p - Probability level for which to compute percentage point

df - Degrees of freedom

Returns:

The corresponding percentage for the chi-square distribution.

Throws:

java.lang.IllegalArgumentException - If df <= 0 or p very near zero or p very near one. A tolerance of 1.0e-15 is used for testing the value of p.

Implements the method of Best and Roberts (1975). The result accuracy varies, but should be good to at least 10 decimal digits in all cases.