Good evening!

I'm trying to fit a Poisson Gamma Mixture in Stata. One of the parameters of interest phi is distributed MVN with a covariance matrix that dependes on other parameters in my estimation. For example:

prior({phi} , mvnormal0(dimension, D)

with D = (1/{tao})*(inv(I-{rho})*W) ; where tao and rho are others hyperpriors and W is a matrix I define arbitrarily. By inv I mean the inverse of the matrix (I-{rho}*W)

¿How can I define D, that is a parameter built on others parameters, in the bayesmh estimation?

Thanks for your help!

Best regards.