distributed-randomizer

DistRND.js

DistRND is a small PRNG (pseudorandom number generator) which also evenly distributes the random values over the given range.

■ Install

<script src="DistRND.min.js"></script>

■ Install (NPM)

npm i distributed-randomizer

■ Use

//Add this line when using NPM:
//const DistRND = require('distributed-randomizer');
 
const rand = new DistRND(min, max);
rand.next();

■ Parameters

argument	data type	description	default
min	positive int	Minimum value
max	positive int	Maximum value
spread	positive int	High spread will distribute numbers more randomly	0

Note: next() will behave more and more like Math.random() with increasing high spread value.

Tip: Start with spread = 1 and increase slowly to see the effects.

■ Showcase

Let's assume we want to generate 10.000 random numbers between 1 and 10. We will count the occurrence of each number for demonstrating purposes:

Comparison:

let data1 = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0];
let data2 = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0];
for (let i = 0; i < 10000; i++) {
    data1[Math.floor(Math.random() * 10)]++;    //with Math.random()
    data2[rand.next()]++;                       //with DistRND.js
}

Results:

data1 (sorted): [934, 956, 957, 997, 1007, 1016, 1018, 1034, 1039, 1042]

data2 (sorted): [1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000]

As you can see the normal Math.random() function can vary quite a bit from the average due to the nature of randomness or at least pseudorandomness in JavaScript. In this case the deviation from 1000 was ~3% on average and ~6.5% maximum.

■ Speed

DistRND.js can generate 1 Mio. random numbers between 1 and 1000 in less than 100 ms.

■ Implementation

• Generate random number x from active scope
• Add 1 to occurrence of x
• If new occurrence of x is greater than the average: Drop x from active scope
• If average updates: Reevaluate active scope

■ FAQ

• Is it still random?

Yes. The numbers in the active scope are still chosen at random with Math.random().

• Why manipulate randomness?

Example 1: When you want to train a Neural Network you could use this to train it with random sample data but also ensure that it won't get overtrained.

Example 2: When you have a game with multiple players you could use this to select random players but also ensure that every player gets the same amount of turns over the entire game.