Computation of the eigenvalues of real symmetric matrices using a processor farm
A parallel implementation of an algorithm to compute the eigenvalues of a real symmetric matrix using a MIMD machine is described. An array of Transputers configured in a ring topology is used as a processor farm for the computation. The speedup obtained by using P processors asymptotically approaches P when the size of (he problem becomes large.
KeywordsLinear algebra matrix multiplication orthogonalisation Transputers ring topology processor farm
Unable to display preview. Download preview PDF.
- Clint, M., Holt, R., Perrott, R. and Stewart, A. A comparison of two parallel algorithms for the symmetric eigenproblem, Int. Jnl. Comp. Math. 15 (1984) 291–302.Google Scholar
- Clint, M., Holt, R., Perrott, R. and Stewart, A. DAP Fortran subroutine for the eigensolution of real symmetric matrices, Comput. J. 28 (1985) 340–342.Google Scholar
- Golub, G. and van Loan, C. Matrix Computations, North Oxford Academic (1983).Google Scholar
- Modi, J. J. and Pryce, J. D. Efficient implementation of Jacobi's diagonalisation method on the DAP, Num. Math. 46 (1985) 443–454.Google Scholar
- Weston, J. S. and Clint, M. Two algorithms for the parallel computation of eigenvalues and eigenvectors of large symmetric matrices using the ICL DAP, Parallel Computing 13 (1990) 281–288Google Scholar
- Waring, L. C. and Clint, M. Improved parallel Gram-Schmidt orthogonalisation on a network of transputers. Applications of Transputers 3, Durrani, T. S., Sandham, W. A., Soraghan, J. J. and Forbes, S. M. (Editors), (1991) vol 1, 117–122.Google Scholar
- Waring, L. C. A general purpose communication shell for a network of Transputers, Microprocessing and Microprogramming 29, vol 2 (1989/90) 107–119.Google Scholar
- The Occam 2 Reference Manual, INMOS Ltd.Google Scholar