Sparse approximations of the Schur complement for parallel algebraic hybrid linear solvers in 3D
نویسندگان
چکیده
In this report we study the computational performance of variants of an algebraic additive Schwarz preconditioner for the Schur complement for the solution of large sparse linear systems. In earlier works, the local Schur complements were computed exactly using a sparse direct solver. The robustness of the preconditioner comes at the price of this memory and time intensive computation that is the main bottleneck of the approach for tackling huge problems. In this work we investigate the use of sparse approximation of the dense local Schur complements. These approximations are computed using a partial incomplete LU factorization. Such a numerical calculation is the core of the multi-level incomplete factorization such as the one implemented in pARMS. The numerical and computing performance of the new numerical scheme is illustrated on a set of large 3D convection-diffusion problems; preliminary experiments on linear systems arising from structural mechanics are also reported. Key-words: Hybrid direct/iterative solver, parallel scientific computing, scalable preconditioner, incomplete/partial factorization, large 3D problems, high performance computing. ∗ INRIA Bordeaux Sud-Ouest HiePACS joint INRIA-CERFACS Lab. [email protected] † Innovative Computing Laboratory, Computer Science Department, University of Tennessee, USA. [email protected] ‡ Department of Computer Science and Engineering, University of Minnesota, USA. [email protected]. The work of this author was supported by the US Department Of Energy under grant DE-FG-08ER25841 and by the Minnesota Supercomputer Institute. in ria -0 04 66 82 8, v er si on 1 6 Ap r 2 01 0 Approximation creuse du complément de Schur pour des solveurs linéaires hybrides pour des problèmes 3D Résumé : Dans ce rapport nous étudions les performances de variantes d’un préconditionneur de type Schwarz additif pour le complément de Schur pour la résolution de systèmes linéaires creux de grande taille. Dans des travaux précédents, les compléments de Schur pour ce préconditionneur étaient calculés explicitement en utilisant un solveur direct creux. La robustesse de ce préconditionneur avait un coût calculatoire et mémoire important qui constituait le principal frein pour la résolution de problèmes de très grande dimension. Dans ce travail, nous étudions l’utilisation d’approximation creuse de ces compléments de Schur denses. Ces approximations sont calculées via une factorisation LU incomplète partielle telle que celle utilisée dans pARMS. Les performances numériques et calculatoires de ce nouveau schéma sont illustrées sur un ensemble de problèmes de convection-diffusion en 3D; des résultats préliminaires sur des systèmes issus de mécanique des structures sont également présentés. Mots-clés : Solveurs linéaires hybrides directs/itératifs, calcul paralléle scientifique, préconditionneurs scalables, fatorisation incompléte/partielle, problémes 3D de grandes tailles, calcul haute performance. in ria -0 04 66 82 8, v er si on 1 6 Ap r 2 01 0 Sparse approximations for parallel algebraic hybrid linear solvers 3
منابع مشابه
Sparse approximations of the Schur complement for parallel algebraic hybrid solvers in 3D
In this paper we study the computational performance of variants of an algebraic additive Schwarz preconditioner for the Schur complement for the solution of large sparse linear systems. In earlier works, the local Schur complements were computed exactly using a sparse direct solver. The robustness of the preconditioner comes at the price of this memory and time intensive computation that is th...
متن کاملOn techniques to improve robustness and scalability of the Schur complement method
A hybrid linear solver based on the Schur complement method has great potential to be a general purpose solver scalable on tens of thousands of processors. It is imperative to exploit two levels of parallelism; namely, solving independent subdomains in parallel and using multiple processors per subdomain. This hierarchical parallelism can lead to a scalable implementation which maintains numeri...
متن کاملOn Techniques to Improve Robustness and Scalability of a Parallel Hybrid Linear Solver
A hybrid linear solver based on the Schur complement method has great potential to be a general purpose solver scalable on tens of thousands of processors. For this, it is imperative to exploit two levels of parallelism; namely, solving independent subdomains in parallel and using multiple processors per subdomain. This hierarchical parallelism can lead to a scalable implementation which mainta...
متن کاملConception d'un solveur linéaire creux parallèle hybride direct-itératif. (Design of a parallel hybrid direct/iterative sparse linear solver)
This thesis presents a parallel resolution method for sparse linear systems which combines effectively techniques of direct and iterative solvers using a Schur complement approach. A domain decomposition is built ; the interiors of the subdomains are eliminated by a direct method in order to use an iterative method only on the interface unknowns. The system on the interface (Schur complement) i...
متن کاملPartitioning, Ordering, and Load Balancing in a Hierarchically Parallel Hybrid Linear Solver
Abstract PDSLin is a general-purpose algebraic parallel hybrid (direct/iterative) linear solver based on the Schur complement method. The most challenging step of the solver is the computation of a preconditioner based on an approximate global Schur complement. We investigate two combinatorial problems to enhance PDSLin’s performance at this step. The first is a multiconstraint partitioning pro...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2011