Problems with using Eigs for a function

Hi,

I have problems with the interface of eigs(). I have a matrix T

T= [A B
    C D]

where matrix vector multiplication can be implemented with FFT for the block D, and the rest are sparse. Hence I have created a function for T rather than a matrix. I want to compute some of the eigenvalues of T, but have problems with the interface of eigs(). I have tried different ways of calling eigs but it always gives me errors.

Note: I did read the manual and when implementing T as an ordinary matrix it works fine... I'm also aware of that the function sent to eigs should implement "T\b" but now I'm just trying to figure out the interface.

When T is implemented as a sparse matrix, the following call works fine

[v_initial, eigenvalue] = eigs(T_old, 1, 'sm');

But when I use T as a function I get the following error:

Error using eigs/checkInputs (line 496)
Generalized matrix B must be the same size as A.

Error in eigs (line 93)
[A,Amatrix,isrealprob,issymA,n,B,classAB,k,eigs_sigma,whch, ...

In the debugger I can see that the matrix B now has gotten the value 'sm'. Just because I shifted from explicit matrix to function, the interface changed...

Question: How is the interface when you use a function instead of an explicit matrix?

Answer 1

[Steven Lord]

Replace the A matrix input with the two inputs AFUN and N. You need to tell MATLAB not only the function it should use to compute but also how large a vector X it should pass into that function as demonstrated by the use of the extra "n" input in the second example in the EIGS documentation as compared with the first example.

[Emil Ringh]

Silly me... Thanks a lot! :)