[PYTHON MATRIX-SIG] Release 1.0alpha1 now available

Konrad Hinsen hinsen@physik.rwth-aachen.de
Sat, 10 Aug 1996 15:46:47 +0200

   The first alpha release of the final 1.0 version of the Numeric =
   Extensions is now available.  This release contains ZERO patches to the =

Great news! That makes me eager to get back to work ;-)
Unfortunately I won't have good enough net access until the end
of next week (right now I have an awfully slow link at 7 cents a minute,
but it works from my laptop...)

   Please beat hard on this and let me know where it breaks!

Sounds like fun!

   I'm getting tired of explaining to people why the Numeric Extensions =
   don't contain matrix inversion or fft operators.  I intend to fix this =
   by providing a standard library with the 1.0 release of the system.  The =
   primitive fftmodule in the current alpha1 release is my start on this.

Good idea. But we should make sure that this standard library is a subset
of some "full" liraries, even if these will not be ready immediately.
I don't want two incompatible matrix inversion functions!

   1) Based on LAPACK and FFTPACK FORTRAN libraries, but only using the =
   subset available with the RLab distribution.  This is a very nice =

I don't know this subset. Is it fully compatible with the current
versions of the full libraries?

   2) Implementing a bare minimum of functionality, in a simple, clean, =
   unsophisticated way.

   ie. fft( (0,0,1,1) ) will just work.  No need to setup work areas, etc.

That should be the default for any numeric library!

   3) Implementing the "basic" functions

   1D FFT
   Matrix Inversion
   Matrix Eigenvalues
   Matrix Determinant

I predict a lot of discussion about what is "basic". My only addition to
your list would be solution of linear equations (inversion is a
terribly inefficient way to do that). I would also like to have different
algorithms where appropriate, a fast one for standard problems and a
foolproof one for people who don't know what they are doing (i.e.
SVD for matrix inversion etc.)

As soon as I am back in business, I will update my LAPACK interface
for the new version. I don't pretend that it is the last word on
linear algebra libraries (some problems, like application to different
types of matrices (symmetric etc.) have been completely ignored
until now), but it has worked quite well for me in the past, and I have
a couple of application algorithms that use it.


Konrad Hinsen                     | E-Mail: hinsenk@ere.umontreal.ca
Departement de Chimie             | Tel.: +1-514-343-6111 ext. 3953
Universite de Montreal            | Fax:  +1-514-343-7586
C.P. 6128, succ. A                | Deutsch/Esperanto/English/Nederlands/
Montreal (QC) H3C 3J7             | Francais (phase experimentale)

MATRIX-SIG  - SIG on Matrix Math for Python

send messages to: matrix-sig@python.org
administrivia to: matrix-sig-request@python.org