currently the parameters used to specify various distributions are assumed to be independent, but in principle they could be correlated. if a user could provide a covariance matrix in addition to each parameter's mean, we could draw from a multivariate normal to pull distribution parameters representative of the covariance relationship.
currently the parameters used to specify various distributions are assumed to be independent, but in principle they could be correlated. if a user could provide a covariance matrix in addition to each parameter's mean, we could draw from a multivariate normal to pull distribution parameters representative of the covariance relationship.