[ADMB Users] The importance of parameterizing a nonlinear model well

[Ben Bolker]

On 13-08-21 01:36 PM, dave fournier wrote:
> Hi Anders,
> 
> I guess my main problem with the paper is that by failing to address the
> reparameterization issue,
> it is more favourable to R. In fact on first reading It seems to imply
> that one can fit the model with R,
> but the issue is a bit muddled.  A user would be justified in concluding
> that  while ADMB was a bit more
> stable R is easier to use.
> 
> A big advantage of ADMB is the automatic calculation of std devs via the
> delta method.   If you use
> R you will have to calculate the derivatives and  delta method
> calculations by hand.  All in all, for
> this particular problem then I think that ADMB is easier than R.
> 
>      Cheers,
> 
>       Dave
> ___________________________________

  Catching up on e-mail ...
  I don't really care about who wins ("more favourable to R vs more
favourable to ADMB").  I actually agree with you that I would have liked
to see the reparameterization issue discussed more thoroughly in this
example ...

  cheers
   Ben