Extend qp() with ARPACK or sparse functions?

[forkandwait]

Hi all,

I use qp() in my work (population forecasting), and get large sparse
matrices for H due to taking the Kronecker product of a bunch of 36 by 36
mostly sparse matrices.  I am unable to do some big analyses with qp()
bogging down, mostly in taking eigenvalues inside __qp__(), as best I can tell.

I was wondering if it would be helpful to extend __qp__() to use eigs() from
the ARPACK library, and maybe allow for sparse matrices, to get some
increased speed.

I was thinking of adding conditionally compiled sections to the __qp__() file.

I can maybe work on this after December.

Can anybody comment on this proposal?  

For reference, here is me rambling a few months ago:

https://mailman.cae.wisc.edu/pipermail/help-octave/2013-April/057969.html