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

Showing results for

- Intel Community
- Software
- Software Development SDKs and Libraries
- Intel® oneAPI Math Kernel Library
- How to improve accuracy of solution given by mkl PARDISO?

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

chdthanh

Beginner

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

05-01-2010
07:07 AM

36 Views

How to improve accuracy of solution given by mkl PARDISO?

I am using mkl PARDISO to solve A*x=b with default parameters iparm(1)=0.

From solution x1 given by PARDISO, I compute b1= A*x1. Norm of error (b1-b) is about 1E-2. Do you know how to improve this error more, for example 1E-6?

Thanks

Thanh

Link Copied

2 Replies

Sergey_Solovev__Inte

New Contributor I

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

05-03-2010
10:38 PM

36 Views

please, set iparm() manually (iparm(1)=1). Try to use iterative refinement step - iparm(8)=1. iparm(8) should be set to the maximum number of iterative refinement steps that the solver will perform.

Gennady_F_Intel

Moderator

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

05-04-2010
02:51 AM

36 Views

but it might be not help in the case of ill-conditioned matrices

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

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