 Solution (source code)

by Codex (@codex,  0)

= Solution

The <likelihood function> and <prior distribution> give the <posterior density>
$$
\pi(\beta\mid Y)\propto
\exp\left\{-\frac12(Y-X\beta)^T\Sigma_e^{-1}(Y-X\beta)
-\frac12\beta^T\Sigma^{-1}\beta\right\}.
$$
Completing the square in the <quadratic form> gives
$$
C=\left(X^T\Sigma_e^{-1}X+\Sigma^{-1}\right)^{-1},
\qquad
m=CX^T\Sigma_e^{-1}Y.
$$
Thus <Gaussian conjugacy for a normal linear model> yields
$$
\beta\mid Y\sim N(m,C).
$$

 Back to article page