Re: [Toon-members] SVD

[E. Rosten]

On Tue, 18 Aug 2009, Georg Klein wrote:

> Hi,
>
> it'd be good to put a C++ version of the SVD into TooN - this turns out to be 
> much faster than LAPACK for small matrices and eliminates dependence on 
> LAPACK, both of which are good things for e.g. iphone stuff.

> Currently I'm using a c++-ified port of the numerical recipes algorithm, but 
> numerical recipes is notoriously difficult with copyright. Any guesses if 
> porting the c-code to C++ is enough to get around copyright, or better not?

It depends if it was ported from a direct copy or written with one eye on 
the book. The code in the optimization subdir is my own implementation of 
the numerical recipies algorithms. I didn't actually ever copy the code, 
but the algorithms are very close in structure. Plus also, the NR code 
isn't very nice to read, since none of the basic linear algebra bits are 
done automatically (like in TooN), and it's pre C89 at best, so all 
variables are declared at the start of the functions (net even block 
local).

If on the other hand you start off by copying the original code and then 
make successive modifications to it, then you're on slightly shakier 
ground.

By the way, NR says:

"if you analyze the ideas contained in a program and then express those 
ideas in your own completely different implementation, then that new 
program implementation belongs to you."

which sounds about right.

> Otherwise I can go back to the FORTRAN code which NumRec used to make their C 
> version.



-Ed

--
(You can't go wrong with psycho-rats.)(http://mi.eng.cam.ac.uk/~er258)

/d{def}def/f{/Times s selectfont}d/s{11}d/r{roll}d f 2/m{moveto}d -1
r 230 350 m 0 1 179{ 1 index show 88 rotate 4 mul 0 rmoveto}for/s 12
    d f pop 235 420 translate 0 0 moveto 1 2 scale show showpage