From: Gautier <gautier.demontmollin@maths.unine.ch>
To: John Halleck <nahaj@u.cc.utah.edu>
Subject: Re: Looking for Ada Technique Name and References
Date: 2000/02/22
Date: 2000-02-22T00:00:00+00:00 [thread overview]
Message-ID: <38B1D1A8.D30E6C55@maths.unine.ch> (raw)
In-Reply-To: 88kh6q$j4j$1@coward.ks.cc.utah.edu
Hi. The point with what you are trying to do is that the "*"
cannot know if its argument is transposed by nature...
unless you create a sort of matrices that are taken
as transposed! But it doesn't solve your pb.
The thing to do is "function Mult_transposed_by(..."
as suggested. No ambiguity... NB: For small matrices (3x3)
it's not so dramatic explicitely to transpose - since
you chose the functional writing that's is already a bit more expensive.
The real cost is with the multiplications. Transposing
is not too awful - no fp arithmetic and well cached. E.g.
I fill sparse matrices with tiny 4x4 up to 27x27 matrices
involved in expressions like
Transpose(Me4) + Transpose(Me3) + Re + Transpose(Me)
and the filling is very fast (DEC Ada on OpenVMS).
--
Gautier
_____\\________________\_______\
http://members.xoom.com/gdemont/
prev parent reply other threads:[~2000-02-22 0:00 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <88kegp$iso$1@coward.ks.cc.utah.edu>
[not found] ` <88kh6q$j4j$1@coward.ks.cc.utah.edu>
2000-02-18 0:00 ` Looking for Ada Technique Name and References Tucker Taft
2000-02-21 0:00 ` Diana Webster
2000-02-22 0:00 ` John Halleck
2000-02-22 0:00 ` tmoran
2000-02-22 0:00 ` David Starner
2000-02-23 0:00 ` tmoran
2000-02-22 0:00 ` Vladimir Olensky
2000-02-22 0:00 ` John Halleck
2000-02-23 0:00 ` Nick Roberts
2000-02-22 0:00 ` Jon S Anthony
2000-02-28 0:00 ` Charles D. Hixson
2000-02-28 0:00 ` Jon S Anthony
2000-02-29 0:00 ` Charles Hixson
2000-02-29 0:00 ` Brian Rogoff
2000-02-29 0:00 ` Jon S Anthony
2000-02-29 0:00 ` Brian Rogoff
2000-02-29 0:00 ` Jon S Anthony
2000-03-01 0:00 ` Brian Rogoff
2000-03-01 0:00 ` Jon S Anthony
2000-03-01 0:00 ` Charles Hixson
2000-03-04 0:00 ` Nick Roberts
2000-02-29 0:00 ` Wes Groleau
2000-02-29 0:00 ` Gautier
2000-03-01 0:00 ` Wes Groleau
2000-02-22 0:00 ` Gautier [this message]
replies disabled
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox