Efficient Methods for Out-of-Core Sparse Cholesky Factorization

  title={Efficient Methods for Out-of-Core Sparse Cholesky Factorization},
  author={Edward E. Rothberg and Robert S. Schreiber},
  journal={SIAM J. Sci. Comput.},
We consider the problem of sparse Cholesky factorization with limited main memory. The goal is to efficiently factor matrices whose Cholesky factors essentially fill the available disk storage, using very little memory (as little as 16 Megabytes (MBytes)). This would enable very large industrial problems to be solved with workstations of very modest cost. We consider three candidate algorithms. Each is based on a partitioning of the matrix into panels. The first is a robust, out-of-core… 

Tables from this paper

An out-of-core sparse Cholesky solver

The design and development of the first release of a new symmetric direct solver that aims to circumvent this limitation by allowing the system matrix, intermediate data, and the matrix factors to be stored externally.

The design and implementation of a new out-of-core sparse cholesky factorization method

A new out-of-core sparse Cholesky factorization method that uses the elimination tree to partition the matrix, an advanced subtree-scheduling algorithm, and both right-looking and left-looking updates is described.

Towards a Parallel Out-of-core Multifrontal Solver: Preliminary Study

A prototype implementation of an out-of-core extension to a parallel multifrontal solver (MUMPS), where disk is used to store data that cannot fit in memory, and it is shown that, by storing the factors to disk, larger problems can be solved on limited-memory machines with reasonable performance.

Scaling and pivoting in an out-of-core sparse direct solver

A new out-of-core multifrontal solver HSL_MA78 from the HSL mathematical software library that is designed to solve the unsymmetric sparse linear systems that arise from finite element applications is considered and how equilibration can be built into the solver without requiring the system matrix to be held in main memory is considered.

A Cholesky out-of-core factorization

The Design of I/O-Efficient Sparse Direct Solvers

This work guides us in designing a software library that implements an external memory sparse solver, and proves upper and lower bounds on these quantities for several model problems with useful sparsity.

Reducing the I/O Volume in Sparse Out-of-core Multifrontal Methods

This paper shows how to process the task dependency graph of multifrontal methods in a way that minimizes the input/output (I/O) requirements and shows that efficient memory management algorithms can be applied to all the variants proposed.

Locality of reference in sparse Cholesky factorization methods.

This paper analyzes the cache efficiency of two high-performance sparse Cholesky factorization algorithms: the multifrontal algorithm and the left-looking algorithm, and shows that while both algorithms sometimes enjoy a high level of data reuse in the cache, they are incomparable: there are matrices on which one is cache efficient and the other is not, and vice versa.

On the Out-Of-Core Factorization of Large Sparse Matrices. (Méthodes directes hors-mémoire (out-of-core) pour la résolution de systèmes linéaires creux de grande taille)

This thesis proposes and studies various out-of-core models that aim at limiting the overhead due to data transfers between memory and disks on uniprocessor machines and focuses on a particular factorization method, the multifrontal method, that it shows allows to solve large sparse linear systems efficiently.

Analysis of the solution phase of a parallel multifrontal approach

Block Sparse Cholesky Algorithms on Advanced Uniprocessor Computers

Two sparse Cholesky factorization algorithms are examined in a systematic and consistent fashion, both to illustrate the strengths of the blocking techniques in general and to obtain a fair evaluation of the two approaches.

An Evaluation of Left-Looking, Right-Looking and Multifrontal Approaches to Sparse Cholesky Factorization on Hierarchical-Memory Machines

A comprehensive analysis of the performance of a variety of sparse Cholesky factorization methods on hierarchical-memory machines finds that primitives that manipulate larger structures greatly increase the amount of exploitable data reuse, thus leading to dramatically higher perfommance on hierarchical -memory machines.

Secondary Storage Methods for Solving Symmetric, Positive Definite, Banded Linear Systems.

This work develops and analyzes methods for using secondary storage in cases where there is not enough primary memory to compute the band Cholesky factorization of matrices, and derives conditions under which the factorization is compute-bound.

The influence of relaxed supernode partitions on the multifrontal method

An algorithm for partitioning the nodes of a graph into supernodes is presented, which improves the performance of the multifrontal method for the factorization of large, sparse matrices on vector computers, and factorizes the extremely sparse electric power matrices faster than the general sparse algorithm.

An adaptive general sparse out-of-core cholesky factorization scheme

Experimental results on some large sparse practical problems are provided; they show significant reduction in storage requirement for Cholesky factors with little increase (and sometimes decrease) in execution time.

On the storage requirement in the out-of-core multifrontal method for sparse factorization

Two techniques are introduced to reduce the working storage requirement for the recent multifrontal method of Duff and Reid used in the sparse out-of-core factorization of symmetric matrices. For a

A Supernodal Approach to Sparse Partial Pivoting

A sparse LU code is developed that is significantly faster than earlier partial pivoting codes and compared with UMFPACK, which uses a multifrontal approach; the code is very competitive in time and storage requirements, especially for large problems.

Parallel sparse Cholesky factorization algorithms for shared-memory multiprocessor systems

This paper considers sparse Cholesky factorization on a multiprocessor system that possesses a globally shared memory and develops a parallel version of a serial blocked left-looking factorization algorithm that reduces the volume of traffic to and from main memory.

Software for Sparse Gaussian Elimination with Limited Core Storage.

A variant of Gaussian elimination is presented for solving sparse symmetric systems of linear equations on computers with limited core storage, without the use of auxiliary storage such as disk or tape, thus trading an increase in work for a decrease in storage.

A combined unifrontal/multifrontal method for unsymmetric sparse matrices

A combined unifrontal/multifrontal algorithm is considered to enable general fill-in reduction orderings to be applied without the data movement of previous multifrontal approaches.