Generate unique number within range (0 - X), keepi

Question

I ran into the challenge where I need a function that returns a random number within a given range from 0 - X. Not only that, but I require the number returned to be unique; not duplicating numbers that have already been returned on previous calls to the function.

Optionally, when this is done (e.g. the range has been 'exhausted'), just return a random number within the range.

How would one go about doing this?

Answer 1

This should do it:

function makeRandomRange(x) {
    var used = new Array(x),
        exhausted = false;
    return function getRandom() {
        var random = Math.floor(Math.random() * x);
        if (exhausted) {
            return random;
        } else {
            for (var i=0; i<x; i++) {
                random = (random + 1) % x;
                if (random in used)
                    continue;
                used[random] = true;
                return random;
            }
            // no free place found
            exhausted = true;
            used = null; // free memory
            return random;
        }
    };
}

Usage:

var generate = makeRandomRange(20);

var x1 = generate(),
    x2 = generate(),
    ...

---

Although it works, it has no good performance when the x-th random is generated - it searches the whole list for a free place. This algorithm, a step-by-step Fisher–Yates shuffle, from the question Unique (non-repeating) random numbers in O(1)?, will perform better:

function makeRandomRange(x) {
    var range = new Array(x),
        pointer = x;
    return function getRandom() {
        pointer = (pointer-1+x) % x;
        var random = Math.floor(Math.random() * pointer);
        var num = (random in range) ? range[random] : random;
        range[random] = (pointer in range) ? range[pointer] : pointer;
        return range[pointer] = num;
    };
}

(Demo at jsfiddle.net)

Extended version which does only generate one "group" of unique numbers:

function makeRandomRange(x) {
    var range = new Array(x),
        pointer = x;
    return function getRandom() {
        if (range) {
            pointer--;
            var random = Math.floor(Math.random() * pointer);
            var num = (random in range) ? range[random] : random;
            range[random] = (pointer in range) ? range[pointer] : pointer;
            range[pointer] = num;
            if (pointer <= 0) { // first x numbers had been unique
                range = null; // free memory;
            }
            return num;
        } else {
            return Math.floor(Math.random() * x);
        }
    };
}

(Demo)

Answer 2

You may try generating the number using the current date and time value which would make it unique. To make it within the range, you may have to use some mathematical function.

Answer 3

I wrote this function. It keeps its own array with a history of generated numbers, preventing initial duplicates, continuing to output a random number if all numbers in the range have been outputted once:

// Generates a unique number from a range
// keeps track of generated numbers in a history array
// if all numbers in the range have been returned once, keep outputting random numbers within the range
var UniqueRandom = { NumHistory: new Array(), generate: function(maxNum) {
        var current = Math.round(Math.random()*(maxNum-1));
        if (maxNum > 1 && this.NumHistory.length > 0) {
            if (this.NumHistory.length != maxNum) {
                while($.inArray(current, this.NumHistory) != -1) { current = Math.round(Math.random()*(maxNum-1)); }
                this.NumHistory.push(current);
                return current;
            } else {
                //unique numbers done, continue outputting random numbers, or we could reset the history array (NumHistory = [];)
                return current;
            }
        } else {
            //first time only
            this.NumHistory.push(current);
            return current;
        }
    }
};

Here's a working Fiddle

I hope this is of use to someone!

Edit: as pointed out by Pointy below, it might get slow with a large range (here is a fiddle, going over a range from 0-1000, which seems to run fine). However; I didn't require a very large range, so perhaps this function is indeed not suited if you look to generate and keep track of an enormous range.

Answer 4

You got some great programming answer. Here's one with a more theoretical flavor to complete your panorama :-)

Your problem is called "sampling" or "subset sampling" and there are several ways you could do this. Let N be the range you are sampling frame (i.e., N=X+1) and M be the size of your sample (the number of elements you want to pick).

• if N is much larger than M, you'll want to use an algorithm such as the one suggested by Bentley and Floyd in his column "Programming Pearls: a sample of brilliance" (temporarily available without ACM's lock screen here), I really recommend this as they explicitly give code and discuss in terms of hash tables, etc.; there a few neat tricks in there

• if N is within the same range as M, then you might want to use the Fisher-Yates shuffle but stop after only M steps (instead of N)

• if you don't really know then the algorithm on page 647 of Devroye's book on random generation is pretty fast.