|ANN: LAPACK and ScaLAPACK new functionality survey firstname.lastname@example.org (2004-06-06)|
|Re: ANN: LAPACK and ScaLAPACK new functionality survey email@example.com (2004-06-06)|
|From:||firstname.lastname@example.org (Nick Maclaren)|
|Date:||6 Jun 2004 21:34:50 -0400|
|Organization:||University of Cambridge, England|
|Posted-Date:||06 Jun 2004 21:34:50 EDT|
Edward Jason Riedy <email@example.com> wrote:
>LAPACK and ScaLAPACK new functionality survey
A couple of quick suggestions that may be a trifle too radical.
The first is FFTs and related thingumajigs. My experience is that
they come in shortly after eigenvalues as the most heavily used
numerically intensive matrix methods.
The second is graph-theoretic algorithms. Let's view a matrix as
a distance graph - there are LOTS of important algorithms that
munge over those things.
Sven will not be surprised that I have posted these suggestions;
he has had me go on about them often enough :-)
Return to the
Search the comp.compilers archives again.