|
|
Blitz Support : |
From: Marc Vinyes (mvinyes_at_[hidden])
Date: 2005-05-04 10:08:17
Apologies for cross-posting. I already sent a message to the list without
subscribing but now I have read more things and I have some other
questions.
http://www.oonumerics.org/MailArchives/blitz-support/2004/01/0993.php
I got the same problem:
Array<float,2> B(2,3),C(3,1);
B=1,2,3,4,5,6;
cout<<B;
C=1,2,3;
cout<<C;
Array<float,2> A(B*C);
cout<<A;
//Has as a result:
// 2 x 3
// [ 1 2 3
// 4 5 6 ]
// 3 x 1
// [ 1
// 2
// 3 ]
// 1 x 1
// [ 1 ]
And,also, for example:
Array<int,2> B(2,3),C(3,1);
B=1,2,3,4,5,6;
cout<<B;
C=1,2,3;
cout<<C;
Array<int,2> A(2,1);
A=B*C;
cout<<A;
//obtaining:
//2 x 3
//[ 1 2 3
// 4 5 6 ]
//3 x 1
//[ 1
// 2
// 3 ]
//2 x 1
//[ 1
// 4 ]
when "the result sould be intuitively":
2x1
[ 14
32]
I understood from the first mail that the practical solution consists in
switching to the "Matrix definition" but then I lose all the power of the
blitz operators like firstIndex, pow2(A),etc... (which give me compilation
errors)
So what do you recommend me? I really need the matrix multiplication and
some operator overloading to manage easily matrix operations and translate
from my "matlab drafts" efficiently.
Maybe for my purposes, I should use other much simpler and
semi-inefficient libraries like: gMatVec
http://gama.fsv.cvut.cz/~cepek/matvec/doc/html/
or http://www.robertnz.net/nm_intro.htm... or maybe I could try to add
some tricks to the blitz code...
I'm just working in my degree's final project in
http://www.iua.upf.es/mtg/eng/, doing some audio processing and filter
training...
Thank you in advance. Congratulations for your work, blitz++ looks
promising but I didn't expect to find this kind of issues...
Marc Vinyes