An Overlapping Domain Decomposition Method for Parameter Identification Problems∗
نویسندگان
چکیده
A parallel fully coupled one-level Newton-Krylov-Schwarz method is investigated for solving the nonlinear system of algebraic equations arising from the finite difference discretization of inverse elliptic problems. Both L and H least squares formulations are considered with the H regularization. We show numerically that the preconditioned iterative method is optimally scalable with respect to the problem size. The algorithm and our parallel software perform well on machines with modest number of processors, even when the level of noise is quite high.
منابع مشابه
A Domain Decomposition Method Combining a Boundary Element Method with a Meshless Local Petrov-Galerkin Method
A non-overlapping domain decomposition algorithm combining boundary element method with meshless local Petrov-Galerkin method is presented for solving the boundary value problem with discontinuous coefficient in this paper. The static relaxation parameter is employed to speed up the convergence rate. The convergence range and the optimal value of static relaxation parameter are studied, but the...
متن کاملA Two-Stage Multi-Splitting Method for Non-Overlapping Domain Decomposition for Parabolic Equations
In domain decomposition for parabolic partial differential equations (PDE) several approaches have been developed— breaking the domain into multiple subdomains of either overlapping or non-overlapping type, or using algebraic type splittings— cf. [CM94] for an overview. An important aspect is how to present the boundary conditions across interfaces or across common unknown points of subdomains,...
متن کاملA Parallel Overlapping Time-Domain Decomposition Method for ODEs
We introduce an overlapping time-domain decomposition for linear initial-value 7 problems which gives rise to an efficient solution method for parallel computers without 8 resorting to the frequency domain. This parallel method exploits the fact that homogeneous 9 initial-value problems can be integrated much faster than inhomogeneous problems by using 10 an efficient Arnoldi approximation for ...
متن کاملRestricted overlapping balancing domain decomposition methods and restricted coarse problems for the Helmholtz problem
Overlapping balancing domain decomposition methods and their combination with restricted additive Schwarz methods are proposed for the Helmholtz equation. These new methods also extend previous work on non-overlapping balancing domain decomposition methods toward simplifying their coarse problems and local solvers. They also extend restricted Schwarz methods, originally designed to overlapping ...
متن کاملAn Iterative Non-overlapping Domain Decomposition Method for Optimal Boundary Control Problems Governed by Parabolic Equations
In this paper, we consider a numerical method for solving optimal boundary control problems governed by parabolic equations. In order to avoid large amounts of calculation produced by traditional numerical methods, we establish an iterative non-overlapping domain decomposition method. The whole domain is divided into many non-overlapping subdomains, and the optimal boundary control problem is d...
متن کامل