High-performance and Parallel Inversion of a Symmetric Positive Definite Matrix
نویسندگان
چکیده
We present families of algorithms for operations related to the computation of the inverse of a Symmetric Positive Definite (SPD) matrix: Cholesky factorization, inversion of a triangular matrix, multiplication of a triangular matrix by its transpose, and one-sweep inversion of an SPD matrix. These algorithms are systematically derived and implemented via the Formal Linear Algebra Methodology Environment (FLAME), an approach for developing linear algebra algorithms. How different members of these families of algorithms are more or less suited for a given architecture is demonstrated via implementations for sequential, shared-memory, and distributed memory parallel architectures. Performance on various platforms is reported.
منابع مشابه
Toward Accelerating the Matrix Inversion Computation of Symmetric Positive-Definite Matrices on Heterogeneous GPU-Based Systems
The goal of this paper is to implement an efficient matrix inversion of symmetric positive-definite matrices on heterogeneous GPU-based systems. The matrix inversion procedure can be split into three stages: computing the Cholesky factorization, inverting the Cholesky factor and calculating the product of the inverted Cholesky factor with its transpose to get the final inverted matrix. Using hi...
متن کاملDECOMPOSITION METHOD FOR SOLVING FULLY FUZZY LINEAR SYSTEMS
In this paper, we investigate the existence of a positive solution of fully fuzzy linear equation systems. This paper mainly to discuss a new decomposition of a nonsingular fuzzy matrix, the symmetric times triangular (ST) decomposition. By this decomposition, every nonsingular fuzzy matrix can be represented as a product of a fuzzy symmetric matrix S and a fuzzy triangular matrix T.
متن کاملInvertastic: Large-scale Dense Matrix Inversion
Linear algebraic techniques are widely used in scientific computing, often requiring large-scale parallel resources such as those provided by the ARCHER service. Libraries exist to facilitate the development of appropriate parallel software, but use of these involves intricacies in decomposition of the problem, managing parallel input and output, passing messages and the execution of the linear...
متن کاملFast symmetric matrix inversion using modified Gaussian elimination
In this paper we present two different variants of method for symmetric matrix inversion, based on modified Gaussian elimination. Both methods avoid computation of square roots and have a reduced machine time’s spending. Further, both of them can be used efficiently not only for positive (semi-) definite, but for any non-singular symmetric matrix inversion. We use simulation to verify results, ...
متن کاملOn the numerical solution of generalized Sylvester matrix equations
The global FOM and GMRES algorithms are among the effective methods to solve Sylvester matrix equations. In this paper, we study these algorithms in the case that the coefficient matrices are real symmetric (real symmetric positive definite) and extract two CG-type algorithms for solving generalized Sylvester matrix equations. The proposed methods are iterative projection metho...
متن کامل