FEAST Eigenvalue Solver

[JohnNichols]

This is an interesting observation from FEAST.  I am putting a bridge as the first big test of the HARRISON, PARDISO and FEAST.  The first thing in is the piles, each is the same vertical length, but they are not connected so one has an interesting matrix.  It all works fine, but one of the sixteen piles has a natural frequency that is slightly different to the rest.    It is not a problem as it is just part of the overall structure, but it has been intriguing.

Are we going to see the latest FEAST in MKL. 

I can get 584 vectors in 12 seconds on a DELL Precision. 

John

[JohnNichols]

When the math does not make sense -- check the data not the math. 

It is reasonably easy to create a DXF plot in Fortran - easiest way to get plots quickly -  see attached file. 

it is very easy in a text file of node to make a mistake - found the mistake.  Of course this points out an interesting thing with the Eigenvalues, from FEAST - as I am using a axisymmetric structure, I get the same eigenvalue for each identical element.  But in a real structure there are minor statistical differences, which is the noisy signals we are finding. nice to find a reason for it. 

FEAST is really great.

J

[Gennady_F_Intel]

John, are you talking about FEAST 3.0?  which new features from FEAST 3.0 are interesting in?

[JohnNichols]

Number of eigenvectors in search interval - really relevant to what we want.

I can send you an example privately if you like - attached file shows the rather interesting problems that some have and shows an interesting take on the analysis.

John