TY - JOUR

T1 - Nonlinear Preconditioning

T2 - How to Use a Nonlinear Schwarz Method to Precondition Newton's Method

AU - Dolean, V.

AU - Gander, M. J.

AU - Kheriji, W.

AU - Kwok, F.

AU - Masson, R.

N1 - Funding information:
This work was partially supported by TOTAL. The work of the fourth author was partially supported by the Hong Kong Research Grant Counci l (grant ECS/22300115) and by the NSFC Young Scientist Fund (grant 11501483).
Publisher copyright:
© 2016, Society for Industrial and Applied Mathematics

PY - 2016/11/1

Y1 - 2016/11/1

N2 - For linear problems, domain decomposition methods can be used directly as iterative solvers but also as preconditioners for Krylov methods. In practice, Krylov acceleration is almost always used, since the Krylov method finds a much better residual polynomial than the stationary iteration and thus converges much faster. We show in this paper that also for nonlinear problems, domain decomposition methods can be used either directly as iterative solvers or as preconditioners for Newton's method. For the concrete case of the parallel Schwarz method, we show that we obtain a preconditioner we call RASPEN (restricted additive Schwarz preconditioned exact Newton), which is similar to ASPIN (additive Schwarz preconditioned inexact Newton) but with all components directly defined by the iterative method. This has the advantage that RASPEN already converges when used as an iterative solver, in contrast to ASPIN, and we thus get a substantially better preconditioner for Newton's method. The iterative construction also allows us to naturally define a coarse correction using the multigrid full approximation scheme, which leads to a convergent two-level nonlinear iterative domain decomposition method and a two level RASPEN nonlinear preconditioner. We illustrate our findings with numerical results on the Forchheimer equation and a nonlinear diffusion problem.

AB - For linear problems, domain decomposition methods can be used directly as iterative solvers but also as preconditioners for Krylov methods. In practice, Krylov acceleration is almost always used, since the Krylov method finds a much better residual polynomial than the stationary iteration and thus converges much faster. We show in this paper that also for nonlinear problems, domain decomposition methods can be used either directly as iterative solvers or as preconditioners for Newton's method. For the concrete case of the parallel Schwarz method, we show that we obtain a preconditioner we call RASPEN (restricted additive Schwarz preconditioned exact Newton), which is similar to ASPIN (additive Schwarz preconditioned inexact Newton) but with all components directly defined by the iterative method. This has the advantage that RASPEN already converges when used as an iterative solver, in contrast to ASPIN, and we thus get a substantially better preconditioner for Newton's method. The iterative construction also allows us to naturally define a coarse correction using the multigrid full approximation scheme, which leads to a convergent two-level nonlinear iterative domain decomposition method and a two level RASPEN nonlinear preconditioner. We illustrate our findings with numerical results on the Forchheimer equation and a nonlinear diffusion problem.

KW - Nonlinear preconditioning

KW - Preconditioning Newton's method

KW - Two-level nonlinear Schwarz methods

UR - http://www.scopus.com/inward/record.url?scp=85007109685&partnerID=8YFLogxK

U2 - 10.1137/15M102887X

DO - 10.1137/15M102887X

M3 - Journal article

AN - SCOPUS:85007109685

SN - 1064-8275

VL - 38

SP - A3357-A3380

JO - SIAM Journal on Scientific Computing

JF - SIAM Journal on Scientific Computing

IS - 6

ER -