Re: alternative to fmins

[Dominique Richard]

Le Mardi 19 Janvier 2016 18:06 CET, Olaf Till <address@hidden> a écrit:

> On Tue, Jan 19, 2016 at 09:50:35AM +0100, RICHARD Dominique wrote:
> > Hi Olaf,
> >  Thanks for your help.
> > I tried 'optim_doc' (both in a terminal at Octave prompt and in the
> > command line of QtOctave) but it doesn't work, probably because I am
> > running Octave 3.2.4. So I tried at home on another computer running
> > Octave 4.0.0, but it didn't work either.
>
> 'optim_doc', as well as 'fmins', 'nonlin_residmin', and
> 'nonlin_curvefit', are in the optim package, not in core
> Octave. Whether 'optim_doc' is present depends on your installed (and
> loaded) version of the optim package.
>
I found a machine with Octave 3.8.1 which includes 'nonlin_curvefit' and  
'nonlin_residmin' but apparently not 'optim_doc'.
The version of the Optim package seems to be 1.3.0 but I am not sure since the 
QtOctave Package Manager indicates :
************************************************
Categories: differential equations
Name: Optim
Name: Optim Version: 1.0.15
Date: 2010-10-25
************************************************
but also :
************************************************
URL: http://downloads.sourceforge.net/octave/optim-1.0.15.tar.gz?download
Candidate version: 1.0.15
Installed version: 1.3.0
Installed in: /usr/share/octave/packages/optim-1.3.0
************************************************

and the funny thing (well not so funny indeed) is that when I first run the 
test scripts of 'nonlin_curvefit' and  'nonlin_residmin' (found on 
Octave-Forge, or at the end of the functions), they runned nicely, but now they 
produce error messages like :
error: subscript indices must be either positive integers less than 2^31 or  
logicals
>>> _ide_reload_variables_list( whos() );
error: called from:
error:   /usr/share/octave/packages/optim-1.3.0/private/__nonlin_residmin__.m a
t line 482, column 9
error:   /usr/share/octave/packages/optim-1.3.0/nonlin_curvefit.m at line 66, c
olumn 18
error:   /home/dri/Documents/Octave/bsi/code_data_condex/test_optimization/test
_nonlin_curvefit.m at line 16, column 26

and I suspect some trouble (which i cannot identify) with QtOctave.

I'll try to have it reinstalled and the use nonlin_curvefit.



--
Dominique RICHARD
Laboratoire de Génie des Procédés Catalytiques - CNRS-CPE Lyon
3, rue Victor Grignard F-69616 Villeurbanne cedex
Tél. (33) 4 72 43 17 52 - Fax (33) 4 72 43 16 73