On Fri, Mar 7, 2014 at 4:56 PM, Jed Brown <[email protected]> wrote: > "Zou (Non-US), Ling" <[email protected]> writes: > > > Hi Jed, > > > > I was searching a topic on preconditioning in the mail list and found > this > > email. > > > > > > On Wed, Oct 10, 2012 at 2:55 PM, Jed Brown <[email protected]> wrote: > > > >> Then you aren't using a matrix-free method at all. > > > > > > In case of using SNESDefaultComputeJacobianColor to compute Jacobian > matrix > > and mf_operator command, is this exactly what you are describing ('aren't > > using a matrix-free method at all')? > > > > The SNESDefaultComputeJacobianColor I found in lines 149-155 from > > > http://www.mcs.anl.gov/petsc/petsc-2.3.3/src/snes/examples/tutorials/ex5.c.html > > 1. Please upgrade PETSc. I don't think we have any links to these > antique man pages, so I assume you specifically changed the version > number back rather than landing there accidentally. > Hmm... I don't know how come I got this old version document. We are using version 3.4 I believe. I simply did google 'SNESDefaultComputeJacobianColor' and the first search result linked me here. I searched 'SNESDefaultComputeJacobianColor' in the petsc 3.4 manual, and it is not existing.
> 2. It is NOT a matrix-free method because coloring is used to assemble > the matrix. It usually doesn't make sense to assemble a matrix using > coloring, but then not use it to apply the matrix (-snes_mf_operator > only uses the assembled matrix for preconditioning). > I agree... Ling
