[petsc-dev] Contribution proposal about the application of PCShell as left and right preconditioner

[Barry Smith]

  Sylvain,

   Thank you. I have put your patch into the branch barry/Sylvain-gmres-symmetric and merged into next for testing.

   Barry

> On Jan 9, 2016, at 3:06 PM, Sylvain Mercier <sylvainmercier85 at gmail.com> wrote:
> 
> Dear PETSc team,
> 
> During my phd thesis, I have worked on solving sequences of linear
> systems with slowly varying matrices using GMRES(restart). In
> particular, I have developed a preconditioning technique to improve
> the action of an existing "first-level preconditioner". This new
> method is usually applied as right preconditioner with a PCShell, although the first preconditioner is applied on the left. 
> 
> That is why I have developed routines in PETSc to be able to apply simultaneously a left and a right PCShells when we use a PC_SYMMETRIC PCSIDE. In fact, I have add two routines: PCShellSetApplySymmetricLeft and PCShellSetApplySymmetricRight (similarly to the routine PCShellSetApply).
> 
> Moreover, I have added the possibility to use a PC_SYMMETRIC preconditioner within GMRES.
> 
> I gently request your position on the integration of these developments in PETSc. I have some problems to provide a pull-request via a fork so I propose a patch. Let me know if you want extra integration work.
> 
> Best regards,
> Sylvain. 
> <0001-Application-symmetric-pchell-and-symmetric-pc-within.patch>