Matt, The linear system is hard to solve and may be ill-conditioned. I got the condition number from PETSc, which does not look too bad (around 10^4), then the distribution of eigenvalues (spectrum) will be helpful for choosing suitable reordering/preconditioning types. Thanks, Qin
________________________________ From: Matthew Knepley <[email protected]> To: Qin Lu <[email protected]> Cc: "[email protected]" <[email protected]> Sent: Wednesday, February 26, 2014 4:46 PM Subject: Re: [petsc-users] Eigenvalues calculations in PETSc On Wed, Feb 26, 2014 at 2:41 PM, Qin Lu <[email protected]> wrote: Hello, > >I need to find the spectrum of the a large linear system (over 100,000 >unknowns). Option -ksp_compute_eigenvalues_explicitly ran out of memory since >the matrix is too big, then -ksp_compute_eigenvalues only gave a small number >of eigenvalues (around 30). My questions are: > >1. Is the number of computed eigenvalues determined by the number of GMRES >iterations to converge? >2. Can PETSc give the extreme eigenvalues now that the number of computed >eigenvalues is limited? > It will help to understand why you want these. Matt Thanks a lot for your info, >Qin > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener
