Stochasm

A JavaScript component to create functions that generate random values.

Why?

It can be very useful to generate random numbers that are governed by properties of different types of distributions. Such distributions are useful for modeling numerical behavior and response of systems.

Fork

This module was forked from https://github.com/heydenberk/stochasm that was created by Eric Heydenberk. Why did I fork it?

• Unmaintained
• Written in CoffeeScript
• Not a UMD
• No tests
• No support to modify the random number generation

Rather than pester Eric about changing any of the above, a fork seemed more reasonable.

Name

stochasm is a portmanteau of stochastic and chasm.

Install

Node.js/Browserify

npm install --save stochasm

Component

component install jprichardson/stochasm

Bower

bower install stochasm

Script

<script src="/path/to/stochasm.js"></script>

Usage

To create a stochasm object, simply invoke the function and pass it an options object with a kind property. If not provided, kind is 'float'.

Valid kinds include float, integer, set.

Floating-point Decimals

It's very easy generate a float between 0 and 1.

var stochasm = require('stochasm')
 
var generator = stochasm()
generator.next(); // 0.9854211050551385
generator.next(); // 0.8784450970124453
generator.next(); // 0.1592887439765036

This is not very exciting because it simply wraps the built-in Math.random method.

Floats from an Interval

Specifying a min and a max allows us to create random numbers in the interval (min, max), not inclusive.

var radianGenerator = stochasm({min: 0, max: Math.PI * 2})
radianGenerator.next(); // 3.7084574239999655
radianGenerator.next(); // 1.021138034566463
radianGenerator.next(); // 4.012664264853087

Floats from a Normal Distribution

We can also generate random floats from a normal distribution. Min and max are optional, and when provided will result in truncation of all results outside of [min, max].

var testScores = stochasm({mean: 75, stdev: 14, min: 0, max: 100})
testScores.next(); // 59.437160028200125
testScores.next(); // 80.18612670399554
testScores.next(); // 75.81242027226946

Integers

For integers, the interval [min, max] is inclusive.

var die = stochasm({kind: "integer", min: 1, max: 6})
die.next(); // 6
die.next(); // 1
die.next(); // 2

if next() feels out of place for your use case, just rename the method:

die.roll = die.next
die.roll() //4

Multiple Results

If the next method (or a method aliased to it) is passed an integer n, it will return an n-length array of results. Using the die instance from the previous example:

die.roll(1); // [5]
die.roll(2); // [5, 3]
die.roll(5); // [6, 3, 6, 6, 5]

From Sets

We can generate random values from arbitary sets.

var dayGenerator = new stochasm({
    kind: "set",
    values: ["monday", "tuesday", "wednesday", "thursday", "friday", "saturday", "sunday"]
});
dayGenerator.next(); // friday
dayGenerator.next(); // monday 
dayGenerator.next(); // monday

From Sets with Weights

What if we favor the weekend? Well, we can pass weights, an array of the same length as values consisting of probabilities out of 1 that correspond to values.

var biasedDayGenerator = new stochasm({
    kind: "set",
    values: ["monday", "tuesday", "wednesday", "thursday", "friday", "saturday", "sunday"],
    weights: [0.1, 0.1, 0.1, 0.1, 0.1, 0.25, 0.25]
});
biasedDayGenerator.next(); // thursday
biasedDayGenerator.next(); // sunday 
biasedDayGenerator.next(); // saturday

From Sets Without Replacement

Note: This functionality may be removed.

Passing a replacement property with a falsy value will result in each random value generation to be removed from the set.

var chores = new stochasm({
    kind: "set",
    values: ["floors", "windows", "dishes"],
    replacement: false
});
var myChore = chores.next(); // "windows"
var yourChore = chores.next(); // "floors"
var hisChore = chores.next(); // "dishes"
var noOnesChore = chores.next(); // undefined

Mutators

The constructor accepts an optional final argument which is passed the output of the random value generator. Its return value becomes the return value of next or its alias. To generate random boolean values, we can do:

var booleanGenerator = new stochasm({
    kind: "integer",
    min: 0,
    max: 1
}, Boolean);
 
booleanGenerator.next(); // false
booleanGenerator.next(); // true
booleanGenerator.next(); // true

We can map the previously mentioned radianGenerator to the cosine of its values.

var radianSineGenerator = new stochasm({
    min: 0,
    max: Math.PI * 2
}, Math.cos);
radianSineGenerator.next(); // -0.31173382958096524
radianSineGenerator.next(); // -0.6424354006937544
radianSineGenerator.next(); // 0.6475980728835664

Mutators remember their previous result and, at each generation, apply the results of a specified stochasm to create a new result.

(This is functionally equivalent to a Markov chain.)

var drunkardsWalk = new stochasm({
    kind: "integer",
    min: -1,
    max: 1
}, function(a, b) { return a + b; });
 
drunkardsWalk.value = 0; // Sets the initial value
drunkardsWalk.next(10); // [-1, -2, -2, -1, -1, -1, 0, 1, 1, 2]
drunkardsWalk.next(10); // [3, 3, 3, 2, 1, 0, -1, 0, 0, 0]
drunkardsWalk.next(10); // [0, 1, 0, -1, 0, 0, 1, 2, 1, 1]

Let's model a bank account's balance. How much money might you have after 10 years if you start with $1000, add $1000 every year, and get interest at a random rate between 1% and 5%?

var addInterest = function(interestRate, principal) {
    return (principal + 1000) * interestRate;
};
var savingsAccountBalance = new stochasm({
    kind: "float",
    min: 1.01,
    max: 1.05
}, addInterest);
 
savingsAccountBalance.value = 1000; // Sets the initital value
savingsAccountBalance.next(10);
/*
[
    2096.2402432970703,
    3177.3792999428224,
    4339.349049328612,
    5441.863800747634,
    6507.916293297546,
    7669.519280743041,
    9011.783840249629,
    10225.82489660009,
    11630.122217972781,
    12782.667463879243
]
*/

Multiple Generators

If the stochasm function is passed multiple configuration objects, next (or its alias) returns an array of each random generated value.

To generate a random point, we might do:

var x = { kind: 'integer', min: 0, max: 480 };
var y = { kind: 'integer', min: 0, max: 360 };
var mutator = function(values) {
    return {
        x: values[0],
        y: values[1]
    };
};
var randomPoint = new stochasm(x, y, mutator);
 
randomPoint.next(); // { x: 79, y: 65 }
randomPoint.next(); // { x: 151, y: 283 }
randomPoint.next(); // { x: 5, y: 253 }

Your Own Random Number Generator

Want to bring your own random number generator to the party? Whether you're working with Node.js and want to use crypto.getRandomValues() or in the browser and want to use window.crypto.getRandomValues(), you can. You could also return a constant to see how your system may respond to certain conditions in testing.

Note: At this time, it's assumed that any function that you set rand to will return a number in the range of [0,1).

var stochasm = require('stochasm')
 
var radianGenerator = stochasm({min: 0, max: Math.PI * 2})
radianGenerator.rand = function() { return 0.4 }

Credits

As stated above, this code was forked from the Node.js module stochator (https://github.com/heydenberk/stochasm) that was created by Eric Heydenberk. Eric Heydenberk deserves much of the credit for coming up with such an awesome idea.

License

It made sense to just keep the license the same. MPLv2.

(MPLv2 License)

Copyright 2013, Eric Heydenberk and JP Richardson