/*
* Copyright (C) 2015 Edward Raff
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see .
*/
package jsat.distributions.discrete;
import static jsat.math.SpecialMath.*;
import static java.lang.Math.*;
import java.util.Random;
import jsat.linear.DenseVector;
import jsat.math.SpecialMath;
/**
* The Poisson distribution is for the number of events occurring in a fixed
* amount of time, where the event has an average rate and all other occurrences
* are independent.
*
* @author Edward Raff
*/
public class Poisson extends DiscreteDistribution
{
private double lambda;
/**
* Creates a new Poisson distribution with λ = 1
*/
public Poisson()
{
this(1);
}
/**
* Creates a new Poisson distribution
* @param lambda the average rate of the event
*/
public Poisson(double lambda)
{
setLambda(lambda);
}
/**
* Sets the average rate of the event occurring in a unit of time
*
* @param lambda the average rate of the event occurring
*/
public void setLambda(double lambda)
{
if (Double.isNaN(lambda) || lambda <= 0 || Double.isInfinite(lambda))
throw new IllegalArgumentException("lambda must be positive, not " + lambda);
this.lambda = lambda;
}
/**
*
* @return the average rate of the event occurring in a unit of time
*/
public double getLambda()
{
return lambda;
}
@Override
public double logPmf(int x)
{
if(x < 0)
return -Double.MAX_VALUE;
//log(e^-lambda lambda^x / x!)
//log(x!) = log(Gamma(x+1))
return -lnGamma(x+1) - lambda + x * log(lambda);
}
@Override
public double pmf(int x)
{
if(x < 0)
return 0;
return Math.exp(logPmf(x));
}
@Override
public double cdf(int x)
{
if(x < 0)
return 0;
return gammaQ(x+1, lambda);
}
private double sampleOne(Random rand)
{
//From http://www.johndcook.com/blog/2010/06/14/generating-poisson-random-values/
double c = 0.767 - 3.36/lambda;
double beta = PI/sqrt(3.0*lambda);
double alpha = beta*lambda;
double k = log(c) - lambda - log(beta);
while(true)
{
double u = rand.nextDouble();
double x = (alpha - log((1.0 - u) / u)) / beta;
double n = floor(x + 0.5);
if (n < 0)
continue;
double v = rand.nextDouble();
double y = alpha - beta * x;
// double lhs = y + log(v/(1.0 + exp(y))^2);
//simplify right part as log(v)-2 log(e^y+1)
// double lhs = y + log(v/pow(1.0 + exp(y), 2));
double lhs = y + log(v) - 2 * log(exp(y) + 1);
// double rhs = k + n*log(lambda) - log(n!);
double rhs = k + n * log(lambda) - SpecialMath.lnGamma(n + 1);
if (lhs <= rhs)
return n;
}
}
@Override
public double[] sample(int numSamples, Random rand)
{
double[] samples = new double[numSamples];
if(lambda < 60)
{
//https://en.wikipedia.org/wiki/Poisson_distribution
double p_init = exp(-lambda);
for(int i = 0; i < numSamples; i++)
{
double u = rand.nextDouble();
double x = 0;
double p = p_init;
double s = p;
while(u > s)
{
x++;
p *= lambda/x;
s += p;
}
samples[i] = x;
}
}
else
{
for(int i = 0; i < numSamples; i++)
samples[i] = sampleOne(rand);
}
return samples;
}
@Override
public double mean()
{
return lambda;
}
@Override
public double mode()
{
//see https://math.stackexchange.com/questions/246496/the-mode-of-the-poisson-distribution/246507#246507
if(lambda < 1)
return 0;
else if(lambda > 1 && Math.rint(lambda) != lambda)
return Math.floor(lambda);
else//lambda is an integer
return lambda;//lamda-1 is also valid
}
@Override
public double variance()
{
return lambda;
}
@Override
public double skewness()
{
return 1/standardDeviation();
}
@Override
public double min()
{
return 0;
}
@Override
public double max()
{
return Double.POSITIVE_INFINITY;
}
@Override
public Poisson clone()
{
return new Poisson(lambda);
}
}
Ads help maintain this website.