// CS 1538 Fall 2009
// This class has some methods to generate random variates

import java.util.Random;
import java.lang.Math;

public class RandDist
{
	private static boolean haveOne = false;
	private static double Z1, Z2;

	// Transform U(0,1) into the exponential distribution using the
	// inverse transform technique.
	public static double exponential(double lambda)
	{
		double R = Math.random();
		double mean = 1/lambda;
		double nextran = -mean * Math.log(R);
		return nextran;
	}

	// Transform U(0,1) into the normal distribution using the polar
	// technique.  Note that this technique generates two values at
	// a time, so the second is saved and used at the next call.
	public static double normal(double mean, double stddev)
	{

		if (haveOne)
		{
			haveOne = false;
			return (mean + stddev*Z2);
		}
		else
		{
			double R1 = Math.random();
			double R2 = Math.random();
			double theta = 2 * Math.PI * R2;
			double left = Math.sqrt((-2)*Math.log(R1));
			Z1 = left*Math.cos(theta);
			Z2 = left*Math.sin(theta);
			haveOne = true;
			return (mean + stddev*Z1);
		}
	}

	// Transform U(0,1) into the Erlang distribution using convolution.  This
	// distribution is the sum of k exponential variates, each with mean 1/k*theta.
	// Using some algebra as explained in the text and notes, we can calculate the
	// variate a bit more efficiently.
	public static double erlang(int k, double theta)
	{
		double R = 1.0;
		for (int i = 1; i <= k; i++)
		{
			R = R * Math.random();
		}
		double logR = Math.log(R);
		return ((-1/(k*theta)) * logR);
	}

	// Discrete Uniform distribution, as discussed in lecture
	public static int uniform(int low, int high)
	{
		double R = Math.random();
		return (low + (int)((high - low + 1)*R));
	}
}