models_learners.py 文件源码

def perf_pi_continuous(self, x):
        # Use history length 1 (Schreiber k=1), kernel width of 0.5 normalised units
        # learnerReward.piCalcC.initialise(40, 1, 0.5);
        # learnerReward.piCalcC.initialise(1, 1, 0.5);
        # src = np.atleast_2d(x[0:-1]).T # start to end - 1
        # dst = np.atleast_2d(x[1:]).T # 1 to end
        # learnerReward.piCalcC.setObservations(src, dst)

        # print "perf_pi_continuous", x
        # learnerReward.piCalcC.initialise(100, 1);
        # learnerReward.piCalcC.initialise(50, 1);
        learnerReward.piCalcC.initialise(10, 1);
        # src = np.atleast_2d(x).T # start to end - 1
        # learnerReward.piCalcC.setObservations(src.reshape((src.shape[0],)))
        # print "x", x.shape
        learnerReward.piCalcC.setObservations(x)
        # print type(src), type(dst)
        # print src.shape, dst.shape
        return learnerReward.piCalcC.computeAverageLocalOfObservations()# * -1