aaronkl / OptSizeChooser

0 stars 1 forks source link

Speeding up variance prediction in Gaussian Process #16

Open simbartonels opened 10 years ago

simbartonels commented 10 years ago

Computing the variance is still inefficient. What's missing is a faster way to compute: np.diag(np.dot(X,Y))

A solution could be: (X * Y.T).sum(-1)

See here:http://stackoverflow.com/questions/14758283/is-there-a-numpy-scipy-dot-product-calculating-only-the-diagonal-entries-of-the