Random.Range gives unexpected distribution

Unity Engine
,
1

Hi all,

I just started a new project and within the first few lines of code there’s already some unexpected behaviour. I create a list of nodes in the following manner:

Node X: 0 - 10 times
Node A: 1 time
Node X: 0 - 10 times
Node B: 1 time
Node X: 0 - 10 times

So, there 3 different kinds of nodes. The list always contains exactly one node of type A and of of type B. Before, in between and after these nodes there is a random amount (capped from 0 to 10, respectively) of nodes of type X. This amount is generated by Random.Range at runtime. The code looks like this:

public void CreateMap()
    {
        nodes.Clear();

        CoinNode coinNode;

        int i, j, k;

        for (i = 0; i < Random.Range(0, 11); i++)
        {
            coinNode = new CoinNode();
            nodes.Add(coinNode);
        }

        StartNode start = new StartNode();
        nodes.Add(start);

        for (j = 0; j < Random.Range(0, 11); j++)
        {
            coinNode = new CoinNode();
            nodes.Add(coinNode);
        }

        GraveyardNode graveyard = new GraveyardNode();
        nodes.Add(graveyard);

        for (k = 0; k < Random.Range(0, 11); k++)
        {
            coinNode = new CoinNode();
            nodes.Add(coinNode);
        }

        Debug.Log($"i: {i}\tj: {j}\tk:{k}");

        Map map = new Map(nodes);
    }

I’ve made 100 runs of this algorithm leading to 300 numbers generated with the following results for the Random.Range function:

0: 25 times
1: 42 times
2: 66 times
3: 56 times
4: 61 times
5: 31 times
6: 13 times
7: 4 times
8: 1 time
9: 1 time
10: 0 times

Now, there must be something off obviously, but I don’t know what. Can anybody help me out here?

Kind regards and thank you in advance!

2

I mean the most apparent thing I can see is using Random.Range in the for loop itself. That will calculate a random value for every iteration of the loop, which I don’t think is a behaviour you’d want.

Though what distribution are you expecting?

3 Likes
3

Isolate, isolate, isolate.

Remove ALL of that crazy wild list processing code and make a loop that pumps the random numbers out in a loop, tallying instances of each result.

That way you can be satisfied that the distribution is within reason. Over a sufficiently large run it will be.

If you are still seeing artifacts in your code, then you have a bug. Go fix your bug.

One approach is to make 10000 or whatever Random float numbers, store them in a text file as a giant JSON array, then use them again and again while you debug your code.

This isolate the random source from your code and lets you debug each individual value.

4

Oh, what an oversight. Didn’t see the forest for the trees there but that for sure solves it. Thank you.