On Tue, 9 Feb 2010, Nuno Santos wrote:
< So given a matrix I want to find the its LU decomposition.
<
< So I could use
<
< > A: matrix( [1, 2, 3], [4, 5, 6], [7, 8, 9] );
< > lu_factor(A)$
< > get_lu_factors(%);
<
< and that would give me L and U in %[2], and [%3], respectively.
<
< However, the LU decomposition, but I am looking for it's:
< L - lower triangular matrix (not lower triangular matrix with unitary
< diagonal, which is what get_lu_factors returns)
< U - as upper triangular matrix with unitary diagonal. (not upper triangular
< without unitary diagonal, which is what get_lu_factors returns)
<
< This is the Khaletsky method. It's not very known. A search in google yields
< very few interesting results, but yet, I would like to be able to compute it
< that way.
<
< You can find a reference to this method here
< http://simlab.mas.bg.ac.rs/ZaSkidanje/LinAlgEq.pdf at slide 17.
<
< So, do you have any idea how I can achieve that?
I think if you want to change this behaviour, you should compute the
factorization of transpose(A), then transpose L & U.
Leo
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.