Re: It Complies, links, & runs...

[Brian Gough <bjg at network-theory dot co dot uk>]

LAPACK was a major effort by many people over a more than a decade, so
rewriting it is not something people are hurrying to do.  

It is straightforward to call the FORTRAN lapack from C (or C++) --
once you know the appropriate tricks.

I can send you an example program if you like, or just look at the
file CmplxSVD.cc in the Octave source code.

regards
Brian Gough


Evan Carew writes:
 > Hmmm... Sounds like a gentle way to say no to me. Can't say as I blame 
 > you, however, the whole reason I am going down this path is because I am 
 > having so much trouble using CLAPACK's svd. I was hoping a well written 
 > functional equivalent in a more flexible language than Fortran would be 
 > easier to use and understand.
 > 
 > Having used your system I have begun to wonder why there hasn't been a 
 > credible rewrite of LAPACK. Compared to LAPACK, the GSL is much easier 
 > to use and understand. I was therefore incouraged to note that is was 
 > fairly easy to go back to my old linear algebra books, read your code, 
 > and make the adjustments to get it to work with complex numbers. I 
 > thinking what would happen next was someone would take a look at my 
 > code, make any necessary minor tweaks and include it (that or tell me I 
 > was definately not on the right track).
 > 
 > I must say that at this point I am a little depressed. The most common 
 > comment I got when asking for help on the sci.math.num-analysis was "Why 
 > aren't you using Fortran?". <sigh>. Waiting another 25 years for the 
 > last of the Fortran programmers to retire is not an option I am happy about.