HSL_MA77
Out-of-core Sparse Symmetric Solver
HSL_MA77 is designed to efficiently solve symmetric sparse linear systems so large they cannot be solved in memory. By holding the data and work arrays in files on disk, the memory requirements are limited, enabling HSL_MA77 to solve much larger problems than is possible with a conventional sparse direct solver.
Note: For unsymmetric systems, HSL_MA78 may be used.
Functionality
- Implements a multifrontal algorithm to solve sparse symmetric systems Ax=b using the Cholesky factorization A = PL(PL)T or the symmetric indefinite factorization A = (PL)D(PL)T.
- Supports input as an assembled matrix, or by elements for finite-element calculations.
- Flexible out-of-core working for large matrices.
- Out-of-core scaling.
- Preserve factors on disk between runs.
- Can also solve the related systems.
- PLx = b, Dx = b, (PL)Tx = b, and D(PL)Tx = b.
Performance
- Solves much larger problems than conventional sparse direct solvers.
- Factorization time competitive with that of well-known in-core codes.
- The overhead of working out-of-core is minimsed through the use of our specially-designed virtual memory system (HSL_OF01).
- Exploits multicore machines using optional OpenMP for linear algebra kernels.
Numerical results are presented in [1] and [2].
Availability
We offer both in-house usage and incorporation licences for HSL_MA77. It is freely available for academic use. Please see our website, or email us at hsl@stfc.ac.uk for further details. Development was funded by EPSRC grant EP/E053351/1.[1] J. K. Reid and J. A. Scott, An out-of-core sparse Cholesky solver, Technical Report RAL-TR-2006-013.
[2] J. K. Reid and J. A. Scott, An efficient out-of-core sparse symmetric indefinite direct solver, Technical Report RAL-TR-2008-024.