BinomialBandit

statrl.settings.bandits.stochastic.anytime.envs.parametric.BinomialBandit(means, repetitions=200, name='MAB-Binomial')

Build a Binomial bandit from a vector of per-trial success probabilities.

Parameters:

• means (array-like of float) – One success probability per arm. Note each arm’s reward mean is repetitions * means[a], not means[a].

• repetitions (int, default=200) – Number of Bernoulli trials summed into one reward.

• name (str, default='MAB-Binomial') – Prefix of the instance name.

Returns:

A bandit with integer rewards in range(repetitions + 1).

Return type:

StochasticBanditEnv