Accelerated Inexact Newton Schemes for Large Systems of Nonlinear Equations

摘要: Classical iteration methods for linear systems, such as Jacobi iteration, can be accelerated considerably by Krylov subspace like GMRES @. In this paper, we describe how inexact Newton nonlinear problems in a similar way and leads to general framework that includes many well-known techniques solving well new ones. Inexact are frequently used practice avoid the expensive exact solution of large system arising (possibly also inexact) linearization step Newton's process. Our acceleration "linear steps" "nonlinear The described class methods, (AIN) contains GMRESR Arnoldi JacDav{} eigenproblems, variants method, damped Newton, problems. As numerical experiments suggest, AIN{} approach may useful construction efficient schemes