Search results

I got the same results as you did. If you flip the covariance matrix or the weights you get the results from the book. math.sqrt(tracking_errors.dot(np.flip(covs)).dot(tracking_errors))