Algorithmically, Gauss elimination and LU factorization are essentially the same. When arithmetic is not strictly associative
(IEEE floats, for example), there would be differences. Actually, LU factorization has variations in the order of calculations
(Doolittle and Crout).
Oh, if binding ratmx to true, cures this problem, I'd say let it be done.
--Barton
________________________________________
From: maxima-bounces at math.utexas.edu [maxima-bounces at math.utexas.edu] on behalf of Robert Dodier [robert.dodier at gmail.com]
Sent: Friday, January 04, 2013 13:53
To: maxima at math.utexas.edu
Subject: Re: [Maxima] rtest_eigen test 13 doesn't finish?
On 2013-01-04, Barton Willis <willisb at unk.edu> wrote:
> Some time ago, I think I said that I didn't feel the love for the
> adjoint method for matrix inversion. But it does have the advantage of
> never needing to check if some huge expression vanishes..
Maybe so, but it was invertmx which was called from within the
simplifier for matrix to a negative exponent (SIMPNCEXPT -> POWERX ->
$INVERTMX) and invertmx used Gaussian elimination. I think the key to
success is that invertmx bound ratmx to true -- I am assuming there
isn't a fundamental advantage of Gaussian elimination over LU. Or is
there?
best
Robert Dodier
_______________________________________________
Maxima mailing list
Maxima at math.utexas.edu
http://www.math.utexas.edu/mailman/listinfo/maxima