
406 Current Trends in Bayesian Methodology with Applications
ables we recommend that of [5]. We set δ = 3 and D = 0.4I
6
+ 0.6J
6
(I
6
and
J
6
denote the identity matrix and the matrix of all ones, respectively).
We sampled 300 observations fro m a Multivariate Normal with mean zero
and precision matrix
18.18 −6.55 0 2.26 −6.27 0
−6.55 14.21 0 −4.90 0 0
0 0 10.47 0 0 −3.6 5
2.26 −4.90 0 10.69 0 0
−6.27 0 0 0 27.26 0
0 0 −3.65 0 0 7.41
(19.8)
FIGURE 19.4
Geometric graphs cor responding to snapshots of posterior samples (a snapshot
was taken from every 100 iterations) from model of Equation (19.6a).