/* This routine is used to reparameterize the model, if desired. (Reparameterization is often desirable to make the form of the posterior closer to a multivariate normal. See documentation in directory above.) The code in this file is used for the univariate regression model; editing is required for new models. If parameters are not transformed, it is still necessary to copy the appropriate information from par into theta and return nk. Inputs: pri Logarithm of prior distribution at current parameter vector par Current record read from posterior simulation file; parameter values are typically contained within this record. npar Number of elements in par Outputs: pri Logarithm of prior distribution at transformed current parameter vector theta Transformed parameter vector nk Number of elements in transformed parameter vector */ proc repar; nk=(npar+1)/2; theta=par[1,1:nk]; theta[nk]=ln(par[npar]); pri=pri+theta[nk]; retp(""); endp;