- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I have tried to understand if MKL provides a faster method for the solution of simple linear tri-diagonal systems which can be easily solved with simple Gaussian elimination.
I am using PARDISO in different cases.
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Pardiso is not the best choice for solving banded linear equations. Please consider using the MKL/Lapack banded equations solvers ?gbslv with bandwidths kl = ku = 1, see https://software.intel.com/en-us/node/468882 .
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
you may also try to evaluate ?gtsv ( Computes the solution to the system of linear equations with a tridiagonal coefficient matrix A and multiple right-hand sides )
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@Gennady,
Documentation of ?gtsv states:
The routine uses Gaussian elimination with partial pivoting.
How come it doesn't use Thomas Algorithm?
How does it compare to vanilla Thomas Algorithm in performance?
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page