[SciPy-user] General_purpose root-finding routines

H Jansen h.jansen at fel.tno.nl
Wed Jan 29 11:44:49 EST 2003


So am I! I've been working on a nontrivial optimal control problem over
the last year, and missing a constraint nonlinear solver in Python, it
has been a pain to do this all in C++. I've posted for its need within a
Python environment before but is seems it's still on the wanted list.

Instead, I've been using the Omuses/HQP SQP (sequential quadratic
problem, sparse matrix) solver which is available on Sourceforge
(http://hqp.sourceforge.net and
http://www.systemtechnik.tu-ilmenau.de/~fg_opt/omuses/hqp.html) and is
free under GPL Version 2 (the only constraint nonlinear solver that is
free, as far as I know) --- I would propose it as a candidate for
Python! A few aspects must be mentioned:
1. It has already a tcl interface for controlling the solution procedure
(parameters and order of execution) --- it should be replace by Python
ASAP;
2. Too bad, it lacks a scripting interface for problem specification ...
(for how long ...?)
3. The optimization problem (the relation between independent (i.e.
control) and dependent variables) must be specified with a special type
of double, so-called adouble (=active double, provided by the free
ADOL-C library): it provides "on the fly" gradient information of the
problem at hand.

Not underestimating the work that should be done, I do believe that it's
a viable candidate to be interfaced with Python!

Henk Jansen

eric jones wrote:
> 
> I'm all for more solvers/optimizers -- especially a constrained solver
> for nonlinear optimization problems.  The best I know of is cfsqp (Andre
> Tits is the author I believe), but it isn't open source.  If we could
> get one of comparable quality in SciPy, that would be a *huge* addition.
> 
> There are probably a lot of more specialized solvers for specific
> problem types that would also be big additions.
> 
> I don't know of anyone working on such things.
> 
> eric
> 
> ----------------------------------------------
> eric jones                    515 Congress Ave
> www.enthought.com             Suite 1614
> 512 536-1057                  Austin, Tx 78701
> 
> > -----Original Message-----
> > From: scipy-user-admin at scipy.net [mailto:scipy-user-admin at scipy.net]
> On
> > Behalf Of Nils Wagner
> > Sent: Wednesday, January 29, 2003 5:52 AM
> > To: scipy-user at scipy.net
> > Subject: [SciPy-user] General_purpose root-finding routines
> >
> > Hi all,
> >
> > Afaik, thers is only one nonlinear m u l t i-variable equation solver
> in
> > scipy so far, that is fsolve.
> > Is there someone thinking about an extension with regard to this field
> ?
> >
> > Nils
> > _______________________________________________
> > SciPy-user mailing list
> > SciPy-user at scipy.net
> > http://www.scipy.net/mailman/listinfo/scipy-user
> 
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.net
> http://www.scipy.net/mailman/listinfo/scipy-user
-------------- next part --------------
A non-text attachment was scrubbed...
Name: h.jansen.vcf
Type: text/x-vcard
Size: 479 bytes
Desc: Card for H Jansen
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20030129/dd10c3ba/attachment.vcf>


More information about the SciPy-User mailing list