bayespy.inference.vmp.nodes.gaussian.GaussianARDDistribution.compute_message_to_parent

GaussianARDDistribution.compute_message_to_parent(parent, index, u, u_mu_alpha)[source]

…

m = \begin{bmatrix} x \\ [-\frac{1}{2}, \ldots, -\frac{1}{2}] \\ -\frac{1}{2} \mathrm{diag}(xx^T) \\ [\frac{1}{2}, \ldots, \frac{1}{2}] \end{bmatrix}