Pseudo-Random number generator that follows a Normal or Log-Normal distribution.

README.md

Marsaglia-polar.js

This script generates random numbers along a Normal or Log-normal distribution using the Marsaglia polar method.

There are four functions you can use:

• normalRandom: Generate random numbers that follow a Normal distribution.
• normalRandomInRange: Generate random numbers that follow a Normal distribution but are clipped to fit within a range
• normalRandomScaled: Generate random numbers that follow a Normal distribution with a given mean and standard deviation
• lnRandomScaled: Generate random numbers that follow a Log-normal distribution with a given geometric mean and geometric standard deviation

References:

• Marsaglia polar method
• Ziggurat algorithm

marsaglia-polar.js

var spareRandom = null;

function normalRandom()
{
	var val, u, v, s, mul;

	if(spareRandom !== null)
	{
		val = spareRandom;
		spareRandom = null;
	}
	else
	{
		do
		{
			u = Math.random()*2-1;
			v = Math.random()*2-1;

			s = u*u+v*v;
		} while(s === 0 || s >= 1);

		mul = Math.sqrt(-2 * Math.log(s) / s);

		val = u * mul;
		spareRandom = v * mul;
	}
	
	return val;
}

function normalRandomInRange(min, max)
{
	var val;
	do
	{
		val = normalRandom();
	} while(val < min || val > max);
	
	return val;
}

function normalRandomScaled(mean, stddev)
{
	var r = normalRandom();

	r = r * stddev + mean;

	return Math.round(r);
}

function lnRandomScaled(gmean, gstddev)
{
	var r = normalRandom();

	r = r * Math.log(gstddev) + Math.log(gmean);

	return Math.round(Math.exp(r));
}

Hi. I tested the normalRandomScaled() function and found it doesn't return a normal distribution with the specified standard deviation. I created sample sizes of 59 with a mean of 1752 and sd of 85 and returned distributions had sd's of around 6.

Kind regards,

Chris

@chrislawes79 thanks for the report. I've updated the code the the version that works. I'd made some silly changes to scale the data and that was messing things up.