752f17f65ff851521b00f699e32f45e14945f069,tests/algorithms/test_td.py,,test_q_learning,#,19

Before Change

    pi, mdp, _ = initialize()
    alg = QLearning(pi, mdp.info, Parameter(.5))

    alg.Q.table = np.arange(np.prod(mdp.info.size)).reshape(mdp.info.size)

    alg._update(0, 1, 100, 1, 0)
    alg._update(1, 0, 10, 3, 1)
    alg._update(3, 1, 50, 3, 0)
    alg._update(2, 2, -100, 3, 1)

    test_q = np.array([[0, 53, 2, 3],
                       [7, 5, 6, 7],
                       [8, 9, -45, 11],
                       [12, 38, 14, 15]])

    assert np.array_equal(alg.Q.table, test_q)

After Change

    // Train
    core.learn(n_steps=100, n_steps_per_fit=1, quiet=True)

    test_q = np.array([[7.82042542, 8.40151978, 7.64961548, 8.82421875],
                       [8.77587891, 9.921875, 7.29316406, 8.68359375],
                       [7.7203125, 7.69921875, 4.5, 9.84375],
                       [0., 0., 0., 0.]])

    assert np.allclose(agent.Q.table, test_q)

Instances