RASHO: A Restricted Additive Schwarz Preconditioner with Harmonic Overlap
نویسندگان
چکیده
A restricted additive Schwarz (RAS) preconditioning technique was introduced recently for solving general nonsymmetric sparse linear systems [1, 3, 4, 7, 8, 9, 11]. The RAS preconditioner improves the classical additive Schwarz preconditioner (AS), [10], in the sense that it reduces the number of iterations of the iterative method, such as GMRES, and also reduces the communication cost per iteration when implemented on distributed memory computers. However, RAS in its original form is a nonsymmetric preconditioner and therefore the cannot be used with the Conjugate Gradient method (CG). In this paper, we provide an extension of RAS for symmetric positive definite problems using the so-called harmonic overlaps (RASHO). Both RAS and RASHO outperform their counterparts of the classical additive Schwarz variants. Roughly speaking, the design of RASHO is based on a much deeper understanding of the behavior of Schwarz type methods in the overlapping regions, and in the construction of the overlap. Under RASHO, the overlap is obtained by extending the nonoverlapping subdomains only in the directions that do not cut the boundaries of other subdomains, and all functions are made harmonic in the overlapping regions. As a result, the subdomain problems in RASHO are smaller than those of AS, and the communication cost is also smaller when implemented on distributed memory computers, since the right-hand sides of discrete harmonic systems are always zero that do not need to be communicated. We will show numerically that RASHO preconditioned CG takes less number of iterations than the corresponding AS preconditioned CG. An almost optimal convergence theory will be
منابع مشابه
Restricted Additive Schwarz Preconditioners with Harmonic Overlap for Symmetric Positive Definite Linear Systems
A restricted additive Schwarz (RAS) preconditioning technique was introduced recently for solving general nonsymmetric sparse linear systems. In this paper, we provide one-level and two-level extensions of RAS for symmetric positive definite problems using the so-called harmonic overlaps (RASHO). Both RAS and RASHO outperform their counterparts of the classical additive Schwarz variants (AS). T...
متن کاملRestricted Additive Schwarz Method with Harmonic Overlap
In this paper, we introduce a new Schwarz preconditioner and with a new coarse space. We construct the preconditioner by grouping together, in one preconditioner, features from the additive overlapping Schwarz methods, from iterative substructuring methods and from a class of restricted additive Schwarz methods. The preconditioner is symmetric and considered as a symmetrized version of restrict...
متن کاملA Restricted Additive Schwarz Preconditioner with Harmonic Overlap for Symmetric Positive Definite Linear Systems ; CU-CS-920-01
متن کامل
A scaled and minimum overlap restricted additive Schwarz method with application to aerodynamics
A variant of the classical additive Schwarz preconditioner (AS) is presented and applied to the solution of a general class of twoand three-dimensional ̄ow problems. The scaled restricted additive Schwarz (RAS) with minimal overlap preconditioner is easy to parallelize since all the local communications among processors only involve information pertaining to the interface of the nonoverlapping ...
متن کاملSchwarz preconditioning for high order edge element discretizations of the time-harmonic Maxwell's equations
We focus on high order edge element approximations of waveguide problems. For the associated linear systems, we analyze the impact of two Schwarz preconditioners, the Optimized Additive Schwarz (OAS) and the Optimized Restricted Additive Schwarz (ORAS), on the convergence of the iterative solver.
متن کامل