[SciPy-user] question about using speigs.ARPACK_eigs

christopher taylor christopher.paul.taylor@gmail....
Thu Feb 5 14:13:29 CST 2009


I'm currently working with sparse matrix of a size of roughly 65K by 65K.

I'd like to compute the first 2 eigenvectors of this 65K x 65K matrix.

I've been told to use speigs.ARPACK_eigs:

# data_mat sizes
data_mat_width=256
data_mat_height=256

#256*256 = 65536

#
# i've noticed there's an assertion that will kill a call to this
function if the last argument is >=4
#
eigvals, eigvecs = speigs.ARPACK_eigs( data_mat.matvec,
data_mat_width*data_mat_height, 2)

Unfortunately, the function is only computing a result of 2x2 matrix.
I need a result that's a matrix with a height of, roughly, 65536 and a
width of 2.

Any recommendations or tips?

Thanks,
ct


More information about the SciPy-user mailing list