On Mon, Oct 27, 2014 at 3:02 PM, Filippo Leonardi < [email protected]> wrote:
> On Monday 27 October 2014 14:31:45 Mark Adams wrote: > > > > I'm not sure what you are asking. starting i=0 or i=1, and with and > > without mg. So four possibilities? The solutions will be different for > > i=0 or 1 in this code: > > > > Yes, 4 cases: > i = 0, MG: wrong solution *(for *any* solution, even afterwards, i.e. i >= > 0)* > (and error, if debug) > OK, so the first time there is no RHS and so not eigen estimate and the next solves do not check to see if a valid eigen estimate has been created. Jed: could we just check for zero iterations in the eigen estimator and then call the random thing. Or something along those lines. > i = 1, MG: ok > i = 0, CG: (or anything): ok > i = 1, CG (or anything): ok > > > -ksp_chebyshev_estimate_eigenvalues_random > > Can try this but the problem, to me, seems deeper (because inherited from > each > solve afterwards).
