Re: lsode -- a do-it-all DGL solver?

[Thomas Treichl]

Matthias Brennwald schrieb:
> Ok, but: why do the odeXY solvers exist in Octave, then? Just for Matlab 
> compatibility? Aren't they faster for specific problems?
>
> Matthias

You're right, the final aim of OdePkg 'just' is to have a MatLab compatible 
solver-package to solve different types of problems. You think there should 
exist more goals? I don't see OdePkg as a competition - I see it as supporting 
the development of Octave and nothing more. When I started with this package I 
set the main focus on solving ODEs and after that I continued with interfaces 
for solvers of *explicitly* formulated DAEs whereas John's implementations 
(lsode, dassl...) can do ODEs and *implicitly* formulated DAEs (also called 
IDEs), ie. meanwhile we are able to solve '3 types of problems resp. formulations'.

When I'm talking about fast and slow then this is meant because the *.m and 
*.c->*.mex files if called from Octave are slower in general than a clean 
*.cc->*.oct implementation. If you have small problems (eg. order 1..50) you 
won't really notice the milliseconds of difference but I think you will notice 
if the order of the problem increases dramatically then. Currently I'm hacking 
my first *.cc (IDE) interface file ('just' another step forward) and sooner or 
later all current *.c files (currently ODE and DAE) will be ported to *.cc 
within OdePkg. Let's talk again about the speed of solving if this porting has 
been finished.

  Thomas