Re: Shooting method for ODE's

[Marc Compere]

Ravi,

    Hi, I've made some pretty good progress in utilizing existing FORTRAN routines to solve ODE's.  Actually, what's happened is that I stare at the code for a long time, make a little progress, gotten stuck again, then asked a knowledgeable programmer in my office for help, then slowly continue.  I'm not an expert by any means, but have ended up with something that works well as an interface between Octave and sdirk4.f, an implicit stiff RK ODE solver.

    I've also written some ODE solvers designed to be similar to the basic functionality of Matlab's ODE45.  You can find those at http://marc.me.utexas.edu/tmp/octave_ode_solvers.  They're easy to program and debug because they're Octave/Matlab m-file scripts, not compiled.

    I do not have a plug-in solution for your Bulerisch-Stoer request, however if you've already got some C or C++ or FORTRAN routines that you'd like to use at the Octave command line, you're welcome to use the following as a template to incorporate your own code:
    http://marc.me.utexas.edu/tmp/octave_ode_solvers/sdirk4_v1.00/sdirk4.cc

    You'll want to look at the rest of the files in http://marc.me.utexas.edu/tmp/octave_ode_solvers/sdirk4_v1.00 also.  Another tip is that the Octave distribution you use is important.  The names of functions change (sometimes) between different versions of the Octave libraries.  The readme file(s) in the directories I've pointed you to should tell you everything you need to know.

    From what I've found, there's no "easy" solution when you start programming interface routines between C/C++/FORTRAN and Octave.  It'll take (maybe several) days to understand what's going on and how you should tweak these examples to fit your particular application.

    Happy computing!  Octave is a *great* tool and only gets better as more and more people add to it's development.  I'm happy to say that, soon, I will have performed almost all my numerical analysis required for my Ph.D. in Octave.  John Eaton has also been very helpful with his suggestions during my struggles in developing the FORTRAN intreface code.  He's developed an awesome freeware tool I plan on using all over the world and throughout my career.

Best regards,
Marc
 

>Dear Colleagues,
>
>I am a new Octave user, and, I would be MOST grateful if someone could
>point me out to an Octave module which performs integration of systems of
>Ordinary Differential Equations (2nd order) for two point boundary value
>problems.  A preferable algorithm would be the Bulerisch-Stoer method, or
>an equivalent.
>
>
>Should this not be available, could someone please educate me as to how I
>can incorporate existing modules from LAPACK/EISPACK or Numerical Recipes
>in Fortran 77/90 and/or C into Octave.
>
>
>Thank you for your time and patience, and I anxiously await a prompt
>response.
>
>Regards
>
>Ravi Venkatesan
 
 
 
 

-- 
_________________________________________________
 Marc Compere, The University of Texas at Austin
 address@hidden, (512)471-7347,   <><
 http://nerdlab.me.utexas.edu/Šþcompere
_________________________________________________