LowerBound.RdComputes the lower bound.
LowerBound(Pg, Omega, M, S, W, Wg, p, logSTW, logSTWg)
| Pg | Edge probability matrix (qxq). |
|---|---|
| Omega | Matrix with precision values common to all spanning trees (qxq). |
| M | Matrix of variational means (nxq). |
| S | Matrix of variational marginal variances (nxq). |
| W | Edges weight matrix. |
| Wg | Variational edges weight matrix. |
| p | Number of observed species. |
| logSTW | Log of the matrix tree run on the W matrix. |
| logSTWg | Log of the matrix tree run on the Wg matrix. |
A vector containing the different terms of the lower bound. More precisely :
J: value of the lower bound.
T1: expectation of log p(Z | T).
T2: expectation of the difference in log scale of the tree density and its variational approximation.
T3: Variational entropy of the latent gaussian parameters Z.