Add eigenvalue decomposition to sparskit interface

[jacobmerson]

I am working on test cases for the stiffness matrix in tobinw/biotissue. For this I would like to check that the matrix is positive definite. It would be useful to have access to the sparskit tql2 method for this.

[wrtobin]

This is related to #3 as it is an additional high-level functionality we need more abstract access to. A wrapper API to retrieve these functions/variations of these functions for a compile-time configurable backend is needed.

[jacobmerson]

That seems a bit beyond what I am interested in implementing, as I am not super familiar with the details of las, and what you are thinking of for this api.

Were you planning on working on the "high-level" api anytime soon?

[wrtobin]

Probably not in the short term no, but I want to reference that issue in conjunction with this regardless.

Adding a specific eigenvalues calculation abstraction is a short term solution inline with the rest of the codebase at the moment. I should be able to get to this tomorrow morning.

[jacobmerson]

Awesome thanks!

[jacobmerson]

this may be a moot point. With the larger systems I really do not want to be computing the eigenvalues/eigenmatrices of the whole system. I may need to look at ARPACK or some other Lanczos type eigenvalue solver.

[wrtobin]

Do you still want this implemented? I was just about to get started on it.

[jacobmerson]

Not for now. Thanks.