Experimentations OneMax problem Q-learning Bayesian-inference on dices Upper-Confidence-Bounds approximation of PI different distribution law probabilities of the dice statistics on languages urn problem