How to Find Similar Eigenvectors

[sfhaider]

If i run a FE model with slight difference, say i changed the Thickness of couple of components, i will get two sets of Eigenvectors. Is there a way, one can find out whether some of these eigenvectors are similar. As the eigenvectors are normalized and linear, only substracting two eigenvector columns will not be enough. What can one do to find out if the two eigenvector columns are similar.
I am using NASTRAN and will use a DMAP code.

 

[GregLocock]

usual way is a least square fit, look at R^2.



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[gwolf2]

There are different normalisation options in most codes, this may help to show the differences better than the default which is usually scaled to max 1.0 displacement.

If you have a decent post-processor, I always animate mode shapes on the screen and spin them around a bit - it will help to reveal any significant differences in shape.

 

[pja]

if for some reason you want a quantitative measure you could find the angle between the two vectors in n-space.