[R] Constrained non-linear mixed models

[Hwange Project]

Dear r-helpers,
does anyone knows how to constrain the value of parameters in a non-linear
mixed model:
I've got something like that:
nlme(log(response)~z+y*log(var1)+x*log(var2)+log((b*var3+a)/(b*var4+a)^x)
where 0<x<1, a<1 and (b*(var3 or var4: range of values similar)+a) between 0
and 1.
or even better if you know how to linearize it !!

and another one :
What means ?
Error in chol((value + t(value))/2) : the leading minor of order 1 is not
positive definite
(my statical books do not go far enough in the detail.
It must be about the Cholesky something if I remember well what I read
long-time ago).

I'm sorry to bother you with this kind of things, but I'm in a remote place
(~18°45'S,26°45'E)
(even if e-mails are working at the incredible speed of 4.8 kbits/sec...)
without a lot of statistical or computer ressources. Thanks,
simon

---------------------------------------------------------
Simon Chamaillé
Hwange Project
CIRAD-CNRS
Hwange Main Camp Research
P. Bag 5776, Dete
(+00 263) 18-647
ciradhwg at mweb.co.zw <mailto:ciradhwg at mweb.co.zw>