[petsc-dev] Mark, what's the rational for keeping the Chebyshev tuning inside GAMG?

[Barry Smith]

> On Feb 21, 2015, at 12:07 AM, Jed Brown <jed at jedbrown.org> wrote:
> 
> Barry Smith <bsmith at mcs.anl.gov> writes:
> 
>>  Mark,
>> 
>>   What's the rational for keeping the Chebyshev tuning inside GAMG
>>   that duplicates that is already support by KSPChebyshev ? I'd like
>>   to remove this chunk of code from GAMG since it is duplicative and
>>   instead make sure that the Chebyshev tuning inside the KSPChebyshev
>>   provides everything that is needed.
> 
> Note that GAMG smoothing cannot use SOR preconditioning, but you
> insisted that PCMG should use Cheby/SOR instead of Cheby/Jacobi by
> default.

   Huhh? But anyways, even if GAMG turns on Cheby/Jacobi  as the smoother there is not reason the Chebyshev estimator you wrote cannot be used since it just uses whatever PC has been set. So your response seems completely orthogonal to my question. Can we merge and just have one Chebyshev estimator?


  Barry

> 
>>   Also in particular what about the /* zeroing out BC rows -- needed for crazy matrices */ should this be ported to KSPChebyshev ?
>> 
>>    /* create cheby smoothers */
>>    for (lidx = 1, level = pc_gamg->Nlevels-2; lidx <= fine_level; lidx++, level--) {
>>      KSP       smoother;
>>      PetscBool flag,flag2;
>>      PC        subpc;
>> 
>>      ierr = PCMGGetSmoother(pc, lidx, &smoother);CHKERRQ(ierr);
>>      ierr = KSPGetPC(smoother, &subpc);CHKERRQ(ierr);
>> 
>>      /* do my own cheby */
>> 
>> Thanks
>> 
>>   Barry