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

Showing results for

- Intel Community
- Software Development Tools (Compilers, Debuggers, Profilers & Analyzers)
- Intel® Fortran Compiler
- dsygv not working as expected

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

Highlighted
##

I have a fortran program which use dsygv routine to get Eigen values for a generalized Eigen value problem. the problem is related to structural dynamics. I have the structure modeled in 3rd party software (sacs) and have all the results. I prepared the 2 matrices A and B in fortran and used dsygv to get the Eigen values - I am getting the first two Eigen values as negative and there after the eig(3) is matching eig(1) of software, eig(4) of fortran is matching eig(2) of software, eig(5)of fortran match eig(3) of software and so on. the matrices were generated inside fortran as result of calculations. I took them out in excel, fed them to Mathcad and the Eigen values generated in Mathcad match very well with software output - so why is there a discrepancy in fortran o/p. please see the attachment showing comparison. results.txt is fortran o/p

rajith_r_

Beginner

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

03-26-2017
05:45 AM

dsygv not working as expected

1 Reply

Highlighted
##

mecej4

Black Belt

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

03-27-2017
07:15 AM

Unless the documentation of a

Unless the documentation of a routine such as DSYGV states that a vector result follows a certain ordering convention, you should not expect any specific order, and it is up to you, after the DSYGV call has been completed, to sort the returned eigenvalues in whatever order you want.

I looked at the image that you attached, which contains numbers strewn all over the place. I have to admit that it conveyed no useful information to me.

The MKL forum is probably a better home for this thread.

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