Re: Interested about number crunching in Ada

[Jerry <lanceboyle@qwest.net>]

On Aug 15, 8:42 pm, holst <henrikhols...@gmail.com> wrote:
> Hi!
>
> I have stumbled upon Ada95 and I have found that a recent addition was
> made to the language standard [1]. An addition I, a student of
> scientific computing, are highly interested in.
>
> What is the best online resource to get into the core of the new high
> performance vector and matrix features? Does there exist some book
> (yet) which covers this area? Or any other field which might be
> related to me (concurrency, Fortran bindings etc.)? I know C and
> Pascal good and I have a good start into Fortran 90/95.

The new numerical aspects of Ada (Annex G.3) are excellent, providing
a number of types and function overloads. The new facilities are
rather basic as far as actual algorithms, but see a very recent
discussion regarding linking to BLAS and LAPACK, if your installation
doesn't already do that. (It seems that BLAS and LAPACK are quasi-
officially recommended --the Ada designers weren't foolish enough to
ignore these venerable numerical packages.)

More broadly as to the appropriateness of using Ada for numerical
work, I personally haven't run across a better solution. I'm a
relatively new user of Ada and am stunned at how well it works for
numerical work. I have used Fortran, Pascal, Matlab/Octave,
Mathematica, Maple, Igor Pro, and some others too obscure to mention
or remember. Ada tops them all for programming. (Mathematica, Maple,
Igor Pro e.g. have many other reasons to recommend them.)

What I (and many others) have done is to write some overloaded procs
and functions to handle vector-matrix things and whatever other
structures your work requires (For example, vectors and matrices of
transfer functions for signal processing and control systems.) With a
few overloaded functions, you can write concise yet clear code that
Matlab aspires to but doesn't entirely succeed at. And you can do
better than Matlab thanks to Ada's strong typing. If you have a vector
x, Matlob will not allow you to compute 1.0/x but Ada will (with an
overload).

I'd be glad to share my collection of overloads that allow mixing
arithmetic between Integers, Long_Floats, Complex, and real and
complex vectors and matrices. I know that there are a lot of
combinations to fully flesh out all of these, but I've found that not
all are required; and if I run across one that I don't have yet, it's
just a couple of minutes to write it.

Jerry