I have a csr 2d finite difference matrix where the coordinates are (r,z). It is banded where the inner bands represent the r-coupling and the outer bands represent the z-coupling. I have been experimenting with using a separate preconditioning matrix. With the current ordering of unknowns, if I extract the part of the csr matrix that represents the r-coupling, I have a compact tridiagonal matrix. If I extract the part of the csr matrix that represents the z-couping, I have a tridiagonal matrix with a large, non-compact bandwidth.
I'm wondering if petsc has support for reordering the unknowns under the hood so that for the z-coupling case, the matrix can be cast as a compact tridiagonal matrix. Essentially, this would be support for minimizing the bandwidth. Thanks, Dave
