The m-files in this directory are the power system simulation programs, which I made for testing the application the chaos synchnization. They are GNU Octave, Matlab and MATCOM compatible. (I belive :-) In my environment # Vine Linux 1.0, which is similar to RedHat 5.2 # Celeron 400MHz, RAM 128MB. the program runs about 80-times faster when compiled to an executable by MATCOM than when interpreted by GNU Octave, like this: --------------------------------------------------------- $ time trsdual > out1.dat 1.340u 0.010s 0:01.35 100.0% 0+0k 0+0io 511pf+0w $ time octave -q trsdual.m > out2.dat 115.380u 0.180s 1:55.65 99.9% 0+0k 0+0io 614pf+0w ---------------------------------------------------------- The reason is : * The program uses a lot of scalar variables. * And it has the Runge-Kutta integration loop in it. (for the research purpose) * So many %# scalar and %# fastindex directives are embedded in the m-files. (manual optimization) The last one is fairly important. Without all these directives, it seems to run only several times faster. The tar-gzipped m-files are: http://www.eecs.kumamoto-u.ac.jp/~t-kita/matcom-bm1/matcom-bm1.tar.gz The proceeding paper (in Japanese) including the results that have been computed with the program: (gzipped PostScript file; 400KB) http://www.eecs.kumamoto-u.ac.jp/~t-kita/pse99.ps.gz ---------- KITA Toshihiro mailto:t-kita@eecs.kumamoto-u.ac.jp Web page : http://www.eecs.kumamoto-u.ac.jp/~t-kita