sample_discrete

int  sample_discrete(int nvalues, float u)

int  sample_discrete(float weights[], float u)

nvalues

The returned integer will be uniform in the range [0,nvalues-1], returning 0 if u==0, and returning nvalues-1 if u==1. The output will be clamped to that range in case u is out of the range [0,1), to reduce the risk of roundoff on u causing problems.

weights

Relative weights, (the sum does not need to be 1), of each integer value in the range [0,len(weights)-1].

u

A number between 0 and 1.

Returns an integer, based on u, either uniformly weighted from 0 to nvalues-1, or weighted based on the weights array from 0 to len(weights)-1. Given uniform random u values in [0,1), the version taking nvalues will return uniform random integers in [0,nvalues-1], and the version taking weights will return random integers in [0,len(weights)-1], where the probability of i is weights[i]/sum_of_weights.