Re: I could not build octave 2.9.12 on cygwin.

[Tatsuro MATSUOKA]

Dear Prof. John W. Eaton

>
>   https://www.cae.wisc.edu/pipermail/bug-octave/2007-May/002487.html
>
>jwe
>
Thank you for your mail.  I have succeded in buiding Octave 2.9.12.
Thanks a lot.  


On cygwin, the compiler gcc 3.4.4-3 has still the problem.
With the compiler gcc 3.4.4-3 on cygwin, I could not build atlas.
I still have to use gcc 3.3.3-3. 
I hope the cygwin support will gcc ver 4. 

By the way
I tested LU decomposition by the Octave 2.9.12 built by cygwin. (with ATLAS)
lutest
Simple left division
ans =  10.229
LU decomposition
ans =  0.14200 

MSVC 2.9.12 by Michael Goffioul

Simple left division
ans =  5.4591
LU decomposition
ans =  0.19162

I feel performance the matrix clclation of 2.9.xx series.



*********lutest.m**************
% test LU decomposition
Num=1000; ItNum=10;
id=1:Num;
rand("seed",1);
A=rand(Num)-0.5;
rand("seed",2);
B=rand(Num,ItNum)-0.5;
[L U P]=lu(A);
%
disp("Simple left division");
ts=time();
for k=1:ItNum
  b=B(:,k);
  x=A\b;
end
time()-ts
x1=x;
%
ts=time();
disp("LU decomposition");
ts=time();
for k=1:ItNum
  b=B(:,k);
  c=P*b; y=L\c; x=U\y;
end
time()-ts
x2=x;
plot(id,x1,"o1",id,x2,"+2");
*********************************

I have a lecture about nummerical calclation for the chemincal Engineer course
student. (undergraduate, the 4th grade.)
I use the octave for the nummerical tool.
In Japan, the octave is not popular to the chemical engineer.

Although many of them of course know Prof. Levenspiel, but do not know 
the octave software. 

I am introducing the octave to my colleges of other univrersities in Japan.
***************************************************************
Dr. Tatsuro MATSUOKA
Associate Professor

Department of Molecular Design and Engineering
Graduate School of Engineering
Nagoya University
Furo-cho, Chikusa-ku, Nagoya, 464-8603, Japan
E-mail address@hidden
Tel.   +81(Japan)-52-789-3274
FAX    +81(Japan)-52-789-3273
****************************************************************