configuration = {
    # Finite horizon of the simulation
    "horizon": 10000,
    # number of repetitions
    "repetitions": 100,
    # Number of cores for parallelization
    "n_jobs": 4,
    # Verbosity for the joblib
    "verbosity": 5,
    # Environment configuration, yeah you can set up more than one.
    # I striped some code that were not published yet, but you can implement
    # your own arms.
    "environment": [
        {
            "arm_type": Bernoulli,
            "probabilities": [0.02, 0.02, 0.02, 0.10, 0.05, 0.05, 0.05, 0.01, 0.01, 0.01]
        }
    ],
    # Policies that should be simulated, and their parameters.
    "policies": [
        {
            "archtype": UCB,
            "params": {}
        },
        {
            "archtype": Thompson,
            "params": {}
        },
        {
            "archtype": klUCB,
            "params": {}
        },
        {
            "archtype": AdBandit,
            "params": {
                "alpha": 0.5,
                "horizon": 10000
            }
        }
    ]
}