R help - Apr 2012 - error estimating parameters with mle2

Hi all,

When I try to estimate the functional response of the Rogers type I 
equation (for the mle2 you need the package bbmle):
> RogersIbinom <- function(N0,attackR2_B,u_B) {attackR2_B+u_B*N0}
> RogersI_B <- 
mle2(FR~dbinom(size=N0,prob=RogersIbinom(N0,attackR2_B,u_B)/N0),start=list(attackR2_B=4.5,u_B=0.16),method="Nelder-Mead",data=data5)


 I get following error message

Error in optim(par = c(4.5, 0.16), fn = function (p)  : 
  function cannot be evaluated at initial parameters

Can someone tell me what I'm doing wrong? I used estimate starting values 
which were predicted with the nls function 

RogersI_N <- 
nls(FR~attackR2_N+u_N*N0,start=list(attackR2_N=1,u_N=4),control=list(maxiter=10000))

For some other models, I do the exact same thing and it works well, so I 
don't understand why this one doesn't work........................

Kind regards,
Met vriendelijke groeten,
Joachim

Don't waste paper! Think about the environment before printing this e-mail

______________________________________

Joachim Audenaert
Adviesdienst Gewasbescherming
Proefcentrum voor Sierteelt
Schaessestraat 18
B-9070 Destelbergen
Belgium
Tel. +32 9 353 94 71
Fax +32 9 353 94 95
E-mail: joachim.audenaert@pcsierteelt.be
www.pcsierteelt.be
______________________________________
	[[alternative HTML version deleted]]

Joachim Audenaert <Joachim.Audenaert <at> pcsierteelt.be> writes:
> When I try to estimate the functional response of the Rogers type I 
> equation (for the mle2 you need the package bbmle):
> 
> > RogersIbinom <- function(N0,attackR2_B,u_B) {attackR2_B+u_B*N0}
> > RogersI_B <- 
> mle2(FR~dbinom(size=N0,prob=RogersIbinom(N0,attackR2_B,u_B)/N0),
  
start=list(attackR2_B=4.5,u_B=0.16),method="Nelder-Mead",data=data5)
> 
>  I get following error message
> 
> Error in optim(par = c(4.5, 0.16), fn = function (p)  : 
>   function cannot be evaluated at initial parameters
> 
> Can someone tell me what I'm doing wrong? I used estimate starting
values
> which were predicted with the nls function 
> 
> RogersI_N <- 
> nls(FR~attackR2_N+u_N*N0,start=list(attackR2_N=1,u_N=4),
>  control=list(maxiter=10000))
  It's hard to know without a reproducible example.
  I'm a little confused by the broader context here because it looks
like your "Rogers functional response" is just linear?? (In the
context
of functional response models in ecology I think of the Rogers random
predator equation -- maybe I have the wrong idea.)

  If your predicted per capita attack probability is greater than
1 for any case, then you're going to be in trouble here ...

  Ben Bolker