Demo and test Right Over Here

Overview

Oftentimes while doing performance tests in javascript, I need to create test data with the following properties:

• extremely large
• sufficiently random
• reproducable

Javascript's Math.random can do the first 2 constraints, but the third one is left in the open and that's the problem; if you can't reliably reproduce the same set of psuedo random numbers every time, then that is yet another factor that may lead to a higher variance during test runs than necessary.

Seeing as the performance of this library could vary too and become a factor, there is a routine designed to generate a set of random numbers of a given length.

About

This library is modularized. It has a set a generic functions and two different generators.

Functions

• seed (x) : seeds the generator to x.
• generate () : generates a floating point between 0 and 1.
• generate (min, max) : generates a floating point between min and max.
• integer () : generates an integer between 0 and 1.
• integer (min, max) : generates an integer between min and max.
• set (count, min, max) : generates an array of floating point numbers between min and max.
• shuffle (array) : re-orders array using Fisher-Yates.
• base (len, radix) : generates len characters in base radix.
• UUID () : generates a UUID formatted string.

Usage

To use, create a new object of the type of generator you want to use. currently supported generators are:

• MT: MT19937 over 32-bit space
• PRNG: X[n + 1] = (a * X[n] + b) mod m

To start a new MT generator:

var generator = new Random.MT();

To start a new PRNG generator:

var generator = new Random.PRNG();

Then you can use it as follows:

generator.UUID();
generator.shuffle([1,2,3,4]);

And so on. Every generator is seeded at 0.

Drop in replacement / Apply on existing code-bases

You can override Math.random by using the _unitgenerator of an instance.

Math.random = generator._unitgenerator;

Extending

The functions above are written generically and can be used with a generator you author yourself.

To create a generator you run this:

var NewType = Random.Generator(gen);
var instance = new NewType();

Where gen is a function that returns an object with the following two keys defined:

• seed: function(number) - seeds the generator
• _unitgenerate: function() - returns an iteration of the generator scaled between 0 and 1.

The MT and the PRNG generators are written this way. You can look at the end of the source file to see how it was done.