Eecient Approximate Solution of Sparse Linear Systems
نویسنده
چکیده
We consider the problem of approximate solution e x of a linear system Ax = b over the reals, such that kAe x ? bk kbk; for a given ; 0 < < 1: This is one of the most fundamental of all computational problems. Let (A) = kAkkA ?1 k be the condition number of the n n input matrix A. Sparse, diagonally dominant (DD) linear systems appear very frequently in the solution of linear systems associated with PDEs and stochastic systems, and generally have polynomial condition number. While there is a vast literature on methods for approximate solution of sparse DD linear systems, most of the results are empirical, and to date there are no known proven linear bounds on the complexity of this problem. Using iterative algorithms, and building on the work of Vaidya V 91] and Gremban et al GMZ 94, GMZ 95, G 96], we provide the best known sequential work bounds for the solution of a number of major classes of DD sparse linear systems. Let = log((A)==). The sparsity graph of A is a graph whose nodes are the indices and whose edges represent pairs of indices of A with non-zero entries. The following results hold for a DD matrix A with non-zero oo-diagonal entries of bounded magnitude: (1) if A has a sparsity graph which is a regular d?dimensional grid for constant d; then our work is O(nn 2); (2) if A is a stochastic matrix with xed s(n)?separable graph as its sparsity graph, then our work is O((n + s(n) 2)): The following results hold for a DD matrix A with entries of unbounded magnitude: (3) if A is sparse (i.e. O(n) nonzeros), our work is less than O((n(+ log n)) 1:5); (4) if A has a sparsity graph in a family of graphs with constant size forbidden graph minors (e.g., planar graphs), then our work is bounded by O(n(+ log n) 1+o(1)) in the case log n = o(log) and O(n(+ log n)) 1+o(1) in the case log = o(log n). We use approximate preconditioned iterations to compute a sequence of iterative improvements to the preconditioned linear system. For class (1) of matrices (and class (2) with s(n) = O(p n)) we construct in O(n) work preconditioners, which reduce the condition number of the resulting preconditioned linear system to O(1); and our resulting 1 total work bounds to approximately solve …
منابع مشابه
Approximate Inverse Preconditioning of Iterative Methods for Nonsymmetric Linear Systems
A method for computing an incomplete factorization of the inverse of a nonsymmetric matrix A is presented. The resulting factorized sparse approximate inverse is used as a preconditioner in the iterative solution of Ax = b by Krylov subspace methods. 1. Introduction. We describe a method for computing an incomplete factorization of the inverse of a general sparse matrix A 2 IR nn. The resulting...
متن کاملExact and approximate solutions of fuzzy LR linear systems: New algorithms using a least squares model and the ABS approach
We present a methodology for characterization and an approach for computing the solutions of fuzzy linear systems with LR fuzzy variables. As solutions, notions of exact and approximate solutions are considered. We transform the fuzzy linear system into a corresponding linear crisp system and a constrained least squares problem. If the corresponding crisp system is incompatible, then the fuzzy ...
متن کاملApproximate Solution of Linear Volterra-Fredholm Integral Equations and Systems of Volterra-Fredholm Integral Equations Using Taylor Expansion Method
In this study, a new application of Taylor expansion is considered to estimate the solution of Volterra-Fredholm integral equations (VFIEs) and systems of Volterra-Fredholm integral equations (SVFIEs). Our proposed method is based upon utilizing the nth-order Taylor polynomial of unknown function at an arbitrary point and employing integration method to convert VFIEs into a system of linear equ...
متن کاملSolving Irregular Sparse Linear Systems On a Multicomputer Using the Cgnr Method
The eecient solution of irregular sparse linear systems on a distributed memory parallel computer is still a major challenge. Direct methods are concerned with unbalanced load processing or data distribution as well as diiculties pertaining to reuse eecient sequential codes. Iterative methods of the Krylov family are well suited for parallel computing but can provide disappointing convergence f...
متن کاملPspases: an Eecient and Scalable Parallel Sparse Direct Solver
Many problems in engineering and scientiic domains require solving large sparse systems of linear equations, as a computationally intensive step towards the nal solution. It has long been a challenge to develop eecient parallel formulations of sparse direct solvers due to several diierent complex steps involved in the process. In this paper, we describe PSPASES, one of the rst eecient, portable...
متن کامل