bayespy.inference.vmp.nodes.gaussian.GaussianDistribution.compute_cgf_from_parents¶

GaussianDistribution.compute_cgf_from_parents(u_mu_Lambda)[source]¶: Compute $\mathrm{E}_{q(p)}[g(p)]$

$g (\boldsymbol{\mu}, \mathbf{\Lambda}) &= - \frac{1}{2} \operatorname{tr}(\boldsymbol{\mu\mu}^{\mathrm{T}} \mathbf{\Lambda} ) + \frac{1}{2} \log |\mathbf{\Lambda}|$