package jsat.distributions;
import jsat.linear.Vec;
/**
*
* @author Edward Raff
*/
public final class Cauchy extends ContinuousDistribution
{
private static final long serialVersionUID = -5083645002030551206L;
private double location;
private double scale;
public Cauchy(double x0, double y)
{
setScale(y);
setLocation(x0);
}
public Cauchy()
{
this(0, 1);
}
public void setLocation(double x0)
{
this.location = x0;
}
public void setScale(double y)
{
if(y <= 0)
throw new ArithmeticException("The scale parameter must be > 0, not " + y);
this.scale = y;
}
public double getScale()
{
return scale;
}
public double getLocation()
{
return location;
}
@Override
public double pdf(double x)
{
return 1.0 / ( Math.PI*scale* (1 + Math.pow((x-location)/scale, 2)) );
}
@Override
public double cdf(double x)
{
return Math.atan((x-location)/scale)/Math.PI + 0.5;
}
@Override
public double invCdf(double p)
{
return location + scale * Math.tan( Math.PI * (p - 0.5) );
}
@Override
public double min()
{
return Double.NEGATIVE_INFINITY;
}
@Override
public double max()
{
return Double.POSITIVE_INFINITY;
}
@Override
public String getDistributionName()
{
return "Cauchy";
}
@Override
public String[] getVariables()
{
return new String[] {"x0", "y"};
}
@Override
public double[] getCurrentVariableValues()
{
return new double[] {location, scale};
}
@Override
public void setVariable(String var, double value)
{
if(var.equals("y"))
setScale(value);
else if(var.equals("x0"))
setLocation(value);
}
@Override
public ContinuousDistribution clone()
{
return new Cauchy(location, scale);
}
@Override
public void setUsingData(Vec data)
{
data = data.sortedCopy();
//approximate y by taking | 1st quant - 3rd quantile|
int n = data.length();
setScale(Math.abs(data.get(n/4) - data.get(3*n/4)));
//approximate x by taking the median value
//Note, technicaly, any value is equaly likely to be the true median of a chachy distribution, so we dont care about the exact median
setLocation(data.get(n/2));
}
/**
* The Cauchy distribution is unique in that it does not have a mean value (undefined).
* @return {@link Double#NaN} since there is no mean value
*/
@Override
public double mean()
{
return Double.NaN;
}
@Override
public double median()
{
return location;
}
@Override
public double mode()
{
return location;
}
/**
* The Cauchy distribution is unique in that it does not have a variance value (undefined).
* @return {@link Double#NaN} since there is no variance value
*/
@Override
public double variance()
{
return Double.NaN;
}
/**
* The Cauchy distribution is unique in that it does not have a standard deviation value (undefined).
* @return {@link Double#NaN} since there is no standard deviation value
*/
@Override
public double standardDeviation()
{
return Double.NaN;
}
@Override
public double skewness()
{
return Double.NaN;
}
@Override
public int hashCode() {
final int prime = 31;
int result = 1;
long temp;
temp = Double.doubleToLongBits(location);
result = prime * result + (int) (temp ^ (temp >>> 32));
temp = Double.doubleToLongBits(scale);
result = prime * result + (int) (temp ^ (temp >>> 32));
return result;
}
@Override
public boolean equals(Object obj) {
if (this == obj) {
return true;
}
if (obj == null) {
return false;
}
if (getClass() != obj.getClass()) {
return false;
}
Cauchy other = (Cauchy) obj;
if (Double.doubleToLongBits(location) != Double
.doubleToLongBits(other.location)) {
return false;
}
if (Double.doubleToLongBits(scale) != Double
.doubleToLongBits(other.scale)) {
return false;
}
return true;
}
}
Ads help maintain this website.