[R] Nonlinear Optimization

[Thomas Lumley]

On Tue, 24 Feb 2004, Kissell, Robert [EQRE] wrote:

> Hi,
>
> I have been brought back to the "R-Side" from MatLab. I have used R in
> graduate econometrics but only for statistics and regression (linear and
> nonlinear). But now I need to run general nonlinear optimization.
>
> I know about the add-in quadprog but my problem is not QP. My problem is a
> general nonlinear (obj funct) with linear constraints.I know about the "ms"
> and "nls" functions, but these seem only for nonlinear regression, not
> nonlinear minimization. I am looking for a pure non-linear optimization
> module.
>
> The nonlinear optimization functions I used in MatLab are "fmincon" and
> "fminunc"
>
> Is there any nonlinear optimization algorithm for R?
>

There are three:

nlm(), using a Newton-type algorithm, unconstrained
optim(), with a palette of algorithms including the box-constrained
	L-BFGS-B
constrOptim(), using an adaptive log-barrier for linear constraints, with
	optim() used for the actual optimisation.

and probably more in add-on packages.

	-thomas