[ADMB Users] multiple random effects vectors with correlated Gaussian priors

[dave fournier]

   Like a lot of things in ADMB this was an idea to see how the simple
    special structure of guassian priors could be exploited. While the
    code was written with more than one object in mind it was never
    completely implemented.  It would need some work.

    the concentrated likelihood idea can't be used with
    SEPARABLE_FUNCTIONS