Improving Multifrontal Methods by Means of Block Low-Rank Representations
نویسندگان
چکیده
Matrices coming from elliptic Partial Differential Equations (PDEs) have been shown to have a low-rank property: well defined off-diagonal blocks of their Schur complements can be approximated by low-rank products. Given a suitable ordering of the matrix which gives to the blocks a geometrical meaning, such approximations can be computed using an SVD or a rank-revealing QR factorization. The resulting representation offers a substantial reduction of the memory requirement and gives efficient ways to perform many of the basic dense algebra operations. Several strategies have been proposed to exploit this property. We propose a low-rank format called Block Low-Rank (BLR), and explain how it can be used to reduce the memory footprint and the complexity of direct solvers for sparse matrices based on the multifrontal method. We present experimental results that show how the BLR format delivers gains that are comparable to those obtained with hierarchical formats such as Hierarchical matrices (H matrices) and Hierarchically Semi-Separable (HSS matrices) but provides much greater flexibility and ease of use which are essential in the context of a general purpose, algebraic solver. Key-words: sparse direct methods, multifrontal method, low-rank approximations, elliptic PDEs ∗ Also available as IRIT report RT/APO/12/6. † INPT(ENSEEIHT)-IRIT, Toulouse, France ({patrick.amestoy,clement.weisbecker}@enseeiht.fr). ‡ Livermore Software Technology Corporation, Livermore, CA, United States ([email protected]). § EDF Recherche et Développement, Clamart, France ([email protected]). ¶ CNRS-IRIT, Toulouse, France ([email protected]). ‖ INRIA-LIP, Lyon, France ([email protected]). ha l-0 07 76 85 9, v er si on 1 16 J an 2 01 3 Utilisation de représentations de rang faible par blocs dans les méthodes multifrontales Résumé : Il a été démontré que les matrices provenant d’équations aux dérivées partielles elliptiques ont une propriété de rang faible: certains blocs hors diagonaux de leurs compléments de Schur peuvent être approchés par des produits de rang faible. Etant donnée une permutation de la matrice qui donne un sens géométrique aux blocs, de telles approximations peuvent être calculées avec une décomposition en valeurs singulières ou une factorisation QR. La représentation correspondante offre une réduction significative en termes de besoins mémoire tout en fournissant des moyens plus efficaces d’effectuer les opérations d’algèbre linéaire dense. Plusieurs stratégies ont été proposées dans la littérature pour exploiter cette propriété. Nous proposons un format Block Low Rank (BLR) et montrons comment il peut être utilisé pour réduire les besoins mémoire et la complexité des solveurs directs pour matrices creuses basées sur la méthode multifrontale. Des résultats expérimentaux montrent que le format BLR donne des gains comparables à ceux obtenus avec des formats hiérarchiques, tout en fournissant une plus grande flexibilité et facilité d’utilisation, qui sont essentielles dans un solveur algébrique général. Mots-clés : matrices creuses, méthodes directes, méthode multifrontale, approximation de rang faible, EDP elliptique ha l-0 07 76 85 9, v er si on 1 16 J an 2 01 3 Improving multifrontal methods by means of block low-rank representations 3 AMS subject classifications. 05C50, 65F05, 65F50
منابع مشابه
Block Low-Rank (BLR) approximations to improve multifrontal sparse solvers
Matrices coming from elliptic Partial Differential Equations (PDEs) have been shown to have a lowrank property: well defined off-diagonal blocks of their Schur complements can be approximated by low-rank products. In the multifrontal context, this can be exploited within the fronts in order to obtain a substantial reduction of the memory requirement and an efficient way to perform many of the b...
متن کاملSuperfast Multifrontal Method for Structured Linear Systems of Equations
In this paper we develop a fast direct solver for discretized linear systems using the multifrontal method together with low-rank approximations. For linear systems arising from certain partial differential equations such as elliptic equations we discover that during the Gaussian elimination of the matrices with proper ordering, the fill-in has a low-rank property: all off-diagonal blocks have ...
متن کاملOn the Complexity of the Block Low-rank Multifrontal
Matrices coming from elliptic Partial Differential Equations have been shown to have a low4 rank property: well defined off-diagonal blocks of their Schur complements can be approximated by low-rank 5 products and this property can be efficiently exploited in multifrontal solvers to provide a substantial reduction 6 of their complexity. Among the possible low-rank formats, the Block Low-Rank fo...
متن کاملOn the Complexity of the Block Low-Rank Multifrontal Factorization
Matrices coming from elliptic Partial Differential Equations have been shown to have a low-rank property: well defined off-diagonal blocks of their Schur complements can be approximated by low-rank products and this property can be efficiently exploited in multifrontal solvers to provide a substantial reduction of their complexity. Among the possible low-rank formats, the Block Low-Rank format ...
متن کاملSuperfast Multifrontal Method for Large Structured Linear Systems of Equations
In this paper we develop a fast direct solver for large discretized linear systems using the supernodal multifrontal method together with low-rank approximations. For linear systems arising from certain partial differential equations such as elliptic equations, during the Gaussian elimination of the matrices with proper ordering, the fill-in has a low-rank property: all off-diagonal blocks have...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- SIAM J. Scientific Computing
دوره 37 شماره
صفحات -
تاریخ انتشار 2015