Page 1 of 1

Imaginary Eigenvalues

Posted: 15 Aug 2011, 19:39
by huffte
In creating a very simple model with 3 DOF and known natural periods, I kept getting "Imaginary Eigenvalue" results.

I gave up for the day and when I opened the file the next morning, I ran without changing anything. The results were then fine and the periods were correct.

So it seems that upon the initial creation of some models - this one had only rigid links, bilinear kinetic links and lumped masses - one might get unexpected results, but that closing and re-opening the project may clear things up.

Re: Imaginary Eigenvalues

Posted: 16 Aug 2011, 15:10
by seismosoft
Dear huffte,

Thanks for this report. Could you please send us (support@seismosoft.com) the model in question, so that we have a look at things?

Thanks again,

Seismosoft Support

Re: Imaginary Eigenvalues

Posted: 16 Aug 2011, 19:12
by huffte
The project files have been sent and they are very simple. It should be easy enough to re-create the model in a few minutes and see if the same problem occurs on your computer. If not, then it may be a processor issue instead of a code issue.

Re: Imaginary Eigenvalues

Posted: 18 Aug 2011, 14:04
by seismosoft
Dear huffte,

By default, the Ritz-Jacobi eigenvalue algorithm is set up to search for 12 vibration modes. Given that your model only has three degrees of freedom, the eigenvalue algorithm was outputting imaginary or spurious modes, with no physical meaning.

When you reduced the number of requested modes from 12 to 3 (the file you sent us did feature such a setting), then the imaginary/spurious vibration modes disappeared.

Hope this helps.

Seismosoft Support