Turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- Intel Community
- Software Development SDKs and Libraries
- Intel® oneAPI Math Kernel Library & Intel® Math Kernel Library
- How to reorganize sparse matrix columns? dss_reorder

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Mute
- Printer Friendly Page

Roberto_Soares

Beginner

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

01-20-2011
01:55 PM

24 Views

How to reorganize sparse matrix columns? dss_reorder

I use MKL to solve FEM linear systems using the DSS library of commands. My question is related to ordering columns.

Currently, when I am currently setting up a matrix, my column numbers (due to connectivity) will not be in ascending order. Therefore I created a simple subroutine to reorganize them, however I feel that there is gotta be a better way. What happens if I send to the solver a unorganized matrix? And, is there a more efficient way to reorganize these matrices?

I know DSS_Reorder takes care of permutation issues. Does it has anything to do with my issue?

I appreciate any suggestion,

Roberto

I use Windows XP, 32 bits, Intel Fortran 11, VS2008.

Link Copied

1 Reply

Sergey_Solovev__Inte

New Contributor I

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

01-23-2011
09:26 PM

24 Views

Hi, Roberto,

The non-zero elements of each row of the matrix `A` must be stored in increasing order (its note from MKL PARDISO manual). Reordering step permutes matrix in order to decrease the time of computations on multi-core computers. This permutation stored in array perm(). Let me note, that input matrix (arrays rowIndex() and columns() ) arent changed during all phases (reordering, factorization and solve).

Sergey

For more complete information about compiler optimizations, see our Optimization Notice.