#18381: Cholesky decomposition should be real
-------------------------------------+-------------------------------------
Reporter: vdelecroix | Owner:
Type: defect | Status: positive_review
Priority: major | Milestone: sage-7.2
Component: linear algebra | Resolution:
Keywords: | Merged in:
Authors: Vincent Delecroix | Reviewers: Dima Pasechnik
Report Upstream: N/A | Work issues:
Branch: | Commit:
u/vdelecroix/18381 | c3db63db26a51c50344e7ecbac56dedd64e860c1
Dependencies: | Stopgaps:
-------------------------------------+-------------------------------------
Comment (by dimpase):
Replying to [comment:18 tdumont]:
> I don't think that Cholesky decomposition of sparse matrices would be so
interesting here. suitesparse
> works with ''very'' sparse matrices we encounter for example when
discretizing PDEs: that is, for a n x n matrix, we have O(n) non zero
terms. As LU and Cholesky decomposition create new non zero terms, one
must reorganise the graph of the matrix to minimize the amount of new
terms created when factorizing. This is for example what SuperLU does, and
the successors of SuperLU too.These are quite specialized softwares.
suitesparse has amd (also packages in cvxopt) to get a kind of reordering
you are talking about. Anyway it is a secondary question how to compute
Cholesky, or perhaps a more general factorisation involving a
permutation/reording, of a sparse [R,C]DF-matrix. It's more important to
get a backend for computations with sparse [R,C]DF-matrices.
--
Ticket URL: <http://trac.sagemath.org/ticket/18381#comment:19>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/d/optout.
