[Bioc-devel] Bioc-devel Digest, Vol 32, Issue 2

[Oliver Bembom]

Hi Tony,

I agree that that solution would be optimal in some sense. If I wrote
an API to donlp2, the user would have to supply R functions for
evaluating the objective function, its gradient, the constraint
functions, and their gradients. I haven't worked much with calling R
functions from within C, but I would be concerned that the many calls
to these R functions from within C would slow down the function quite
a bit.

Apart from that, it would be a good amount of work, since my R package
is basically a wrapper to a C function I wrote first which
incorporates the optimizer. So disentangling the optimization from
that wouldn't be trivial.

Oliver

On 11/2/06, A.J. Rossini <blindglobe at gmail.com> wrote:
> > From: "Oliver Bembom" <bembom at gmail.com>
> > Subject: Re: [Bioc-devel] Licensing question
>
>
> Another more painful approach would be to separate out DONLP2 into a
> general constrained optimization package, provide a common API, and
> then
> you could distribute your components as you feel fit.
>
> Then cosmo could be licensed any which way, and DONLP2 could
> eventually be replaced by an open source constrained optimizer.
>
> This is just a suggestion, not a recommendation!
>
> best,
> -tony
>
> blindglobe at gmail.com
> Muttenz, Switzerland.
> "Commit early,commit often, and commit in a repository from which we can easily
> roll-back your mistakes" (AJR, 4Jan05).
>