| Date: | Mon, 11 Apr 2011 11:39:07 -0400 |
|---|
| Reply-To: | R B <ryan.andrew.black@GMAIL.COM> |
|---|
| Sender: | "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU> |
|---|
| From: | R B <ryan.andrew.black@GMAIL.COM> |
|---|
| Subject: | Re: NLMIXED Versus Proc Mixed |
|---|
|
| In-Reply-To: | <cf6ee41e7970d0c62ea7d5785ac80457.squirrel@webmail.pitt.edu> |
|---|
| Content-Type: | text/plain; charset=ISO-8859-1 |
|---|
Francis,
It would help if you posted your code.
Ryan
On Mon, Apr 11, 2011 at 11:08 AM, SUBSCRIBE SAS-L Ameribrit
<FRP3@pitt.edu> wrote:
> HI Guys:
> I am running a bivariate model in SAS using proc mixed and it runs great
> and in a couple of seconds. What I would like to do is to reproduce these
> same results in NLMIXED.
>
> What I am finding is that it runs forever even with a single quadrature
> point. I have tried taking the values I get from Proc mixed with ML option
> and using these as starting values in NLMIXED but I get 'NO VALID
> PARAMETER POINTS FOUND:.
>
> Why does Proc Mixed give instant results yet NLMIXED takes exponentially
> longer or does not run at all on the same dataset.
>
> thanks
> Francis
>
|
