Inverse Transform Random Sampler

inv_transform.py

def rand_sampler(self, n, prob, u):
        """
        Generates a random sample of k numbers according to the discrete probability distribution
        n: sample size
        pro: probability distribution
        """
        cdf = []
        np.random.seed(100)
        for k in range(0, len(prob)):
            cdf.append(sum([prob[k] for k in range(0, k + 1)]))
        # u = np.random.sample(n)
        rds = []

        for j in range(0, len(cdf)):
            if j == 0:
                
                tmp = np.where(u <= cdf[j])[0]
                rds += np.repeat(j, len(tmp)).tolist()
            else:
                print j
                tmp = np.where(np.logical_and(u <= cdf[j], u > cdf[j - 1]))[0]
                rds += np.repeat(j, len(tmp)).tolist()

        return rds