Solving large matrix equations

[Martin Heller]

Jaroslav Hajek wrote:

> Since your H matrix is full, you're solving a full system using
> LAPACK, and 10000 is really too much, at least for a typical computer.
> My advice is you avoid forming of the matrix H explicitly and rather
> express its action on a vector via a function (K should be doable with
> cumsum and D with diff). Then you'll be able to feed the function into
> pcg and get your solution.

Thanks for this. I will see if I can get it working and then submit the 
function to Octave-Forge.

Thanks to everybody for the responses -- both on the list and via 
private mail.