dcde2348aed207dbd12959aa2ee2096d016feb64,skopt/gbrt_opt.py,,gbrt_minimize,#Any#Any#Any#Any#Any#,52

Before Change

        // XXX so right now BFGS can"t do anything useful
        // XXX -> equivalent to random sampling
        x0 = _random_point(lower_bounds, upper_bounds, rng)
        next_x, _, _ = fmin_l_bfgs_b(
            _expected_improvement,
            np.asfortranarray(x0),
            args=(rgr, best_y),

After Change

                           n_points=20,
                           random_state=rng)
        aq = _expected_improvement(x0, rgr, best_y)
        best = np.argmin(aq)

        Xi[i] = x0[best].ravel()
        yi[i] = func(x0[best])

        if yi[i] < best_y:

Instances