a = (a - np.mean(a)) / (np.std(a) * len(a)) b = (b - np.mean(b)) / (np.std(b)) c = np.correlate(a, b, 'full')