QuickTopic free message boards logo
Skip to Messages

TOPIC:

Direct Methods for Sparse Least Squares Problems

  Spam messages 9-8 deleted by QuickTopic between 08-18-2010 02:03 AM and 07-21-2006 09:00 AM
7
Gary Tedeschi
11-30-2004
12:59 PM ET (US)
Has anyone used a non-FORTRAN, non-commercial (including MATLAB) general sparse matrix subroutine library that they would recommend?
6
Manmohan Chandraker
11-30-2004
12:36 PM ET (US)
For special cases, one can surely think of a series of transformations to change the column ordering individually. But such transformations would not be orthogonal and full rank. The advantage of a permutation matrix is that it is trivially full rank (its columns are basis vectors) and orthogonal. Thus, we only consider entire column and row orderings, not for individual elements.
5
Robin Hewitt
11-30-2004
11:49 AM ET (US)
Rasit,
re - change column ordering of rows individually -
I don't think you could do that without changing the matrix.
Edited 11-30-2004 11:50 AM
4
Rasit Topaloglu
11-29-2004
10:12 PM ET (US)
Would a low bandwidth be more adventageous at all times?

Would it be possible to change column ordering of rows individually to put a matrix into a band structure?
3
Manmohan Chandraker
11-29-2004
09:52 PM ET (US)
Stephen, the parts I will cover will mostly deal with graph theoretic approaches to dealing with sparsity. The relevant parts from the text would be Section 6.4 onwards, but some earlier results will be needed. Since it is a short presentation, I don't think I will be able to cover banded sparse systems. So the talk should be limited to general sparse systems.

The text uses results from graph theory, the proofs can be found in papers by R.E.Tarjan, such as "Graph Theory and Gaussian Elimination" and "Algorithmic Aspects of Vertex Elimination on Directed Graphs" by D.J.Rose and R.E. Tarjan.
Edited 11-29-2004 09:57 PM
2
Rasit Topaloglu
11-29-2004
09:37 PM ET (US)
.
Edited 11-29-2004 09:46 PM
1
Stephen Krotosky
11-29-2004
07:24 PM ET (US)
This is quite dense. If you had a specific sparse matrix type, it would be useful to apply the appropriate method. Is there a specific method that you're interested in for your work, or is you interest more general?

Print | RSS Views: 2275 (Unique: 1094 ) / Subscribers: 0 | What's this?