Date:         Wed, 14 Jul 2010 07:35:12 -0400
Reply-To:     Ryan Black <ryan.andrew.black@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Ryan Black <ryan.andrew.black@GMAIL.COM>
Subject:      Adaptive Importance Sampling in NLMIXED
Content-Type: text/plain; charset=ISO-8859-1

Hi,

NLMIXED offers various estimation methods (e.g., adaptive quadrature,
adaptive importance sampling). Which nonlinear mixed modeling
scenario(s) might the use of adaptive importance sampling be
preferable to the use of adaptive quadrature? I'm thinking of a large
nested random effects model where it becomes computationally
infeasible to employ the adaptive quadrature method. One could, I
suppose, force the number of quadrature points to be considerably
small (e.g. 1) in such a scenario. But would that be preferable to an
adaptive importance sampling method? Any other examples where adaptive
importance sampling would be the method of choice? General thoughts on
this topic are welcome.

Thanks,

Ryan