First I'd like to say that I admire your open mind and the tolerance you
display here. Maybe I should be a little less zealous as well :-)
But my university pays about 500 k€ each year (550 k$ !) only to be able to
use Matlab and we even have the cheap academic license ! That sometimes
irritates me ...
About Octave : the speed of both Octave and Matlab largely depends on the
Blas and Lapack libraries that sit behind the curtain. If you use a heavily
optimised Blas (like the Atlas stuff) Octave will be ways faster as well.
And ... Blas and Lapack are also Open Source ...
Using Blas and/or Lapack directly from C instead of through Matlab could be
much faster and isn't that difficult although it uses a somewhat archaic
Fortran based calling mechanism.
And another old but very good (C oriented) library for Linear Algebra is
Meschach : very intuitive and probably 5 times faster than Matlab for many
// And solve the system A.x = b using the Meschach routines :
LU = m_get(A->m,A->n);
LU = m_copy(A,LU);
pivot = px_get(A->m);
x = LUsolve(LU,pivot,b,VNULL);
Also Python is definitely worth looking at : Numpy, Scipy , Sympy and
Matplotlib are just as easy to use as Matlab.
Discussion closed ?
On Friday 01 June 2012, Curtis Olson wrote:
> On Fri, Jun 1, 2012 at 3:20 AM, Kees Lemmens
> > Ji,
> > Just wondering : why are some of you using an Open Source Flight
> > Simulator only to connect it to one of the most expensive (read least
> > value for money)
> > mathematical software packages in the world ?
> > You could at least consider trying to hookup to Octave (an Open Source
> > Matlab
> > lookalike), Scilab or Numpy/Scipy instead : that would fit the ideas
> > behind Flightgear a lot better ;-)
> > I suggest the people who want to connect to Matlab to try a commercial
> > simulator like MS-FlightSimulator instead ...
> To be fair, many people using matlab are students or work for larger
> companies that have purchased this software already. I'm a big open
> source fan, but one matlab script I have here ran about 1/10th the
> speed in octave -- although octave did produce correct results (within
> the margins of floating point rounding variations.) If you are working
> with large data sets or you are working on your code and need to
> iteratively run/test/edit many times, that extra speed may be worth the
> cost of the package -- but everyone's situation is different, and
> everyone is working on different problems with different data. Many
> people (myself usually too) have more time than money.
> I used to be more of a software/platform/os advocate than I am now --
> I've come to think that if a person is happy and comfortable and
> productive with the tools they are using, then that is a good thing --
> of course as long as that includes FlightGear. :-) I see no evil in
> purchasing software if it helps you get a task done. Personally I run
> Linux and use open-source tools almost exclusively, but there are times
> when a commercial package does work better or offers some feature that
> an open-source package does not. There are other times when an
> open-source solution works best.
> If someone is looking for a tool or a solution to a problem, that's
> definitely the time to suggest open-source alternatives in my view. But
> if someone is settled in and happy, let them work in peace. :-)