Re: 3.1 status report

[John W. Eaton]

On  3-Apr-2008, Jaroslav Hajek wrote:

| >  Another small but fairly important project is improving the
| >  compatibility of fsolve.  Is anyone interested in working on that?
| 
| I volunteer for this one (unless anyone else is working on that).

Not yet, no.  So it's all yours.

| As a future project, what about making fsolve reentrant?

Is there an C implementation of these functions in the GSL?  If so,
maybe we should use that.  But that project can wait until we decide
how to include GSL functions in Octave, probably in 3.2.  For now,
let's just focus on getting the fsolve scripting language interface to
be compatible.

| Further, what about including also the nonlinear least squares part of
| MINPACK (LMDIF/LMDER) to implement lsqnonlin?

| I see fzero, fmins and fminbnd are also missing. There may be some
| state-of-the art implementations of fzero and fminbnd available, or
| can be implemented according to the well-known books...

Yes all these are good projects.  If they can be done with GSL
functions, then I think we should use those.  Maybe in 3.2?  If you
want to write them now, then maybe they could go in Octave Forge until
we release 3.1 and start work on 3.2?

jwe