hopsy.Gaussian.log_curvature#

Gaussian.log_curvature(self x)#

Computes the expected fisher information of a multivariate Gaussian model in \(n-k\) dimensions at x. This turns out to be just the reduced covariance matrix. Note that x still has to have dimension \(n\).

Parameters:

x (numpy.ndarray[n, 1]) – Input vector

Returns:

The expected Fisher information matrix, which we call log_curvature in this context.

Return type:

numpy.narray[n, n]