I’ve actually used Omega = quad_form_diag(rho,omega), where rho is the correlation matrix and omega is a vector of standard deviations. But this will return Omega to be a matrix with standard deviations on the diagonal and covariance on the off-diagonals. Then passed this in as a multivariate normal prior ~ multi_normal(beta,Omega). Can it be done this way? Or is the Omega matrix essentially wrong- should the diagonals be variances?