Some Remarks on the Generalised Bareiss and Levinson Algorithms
The Bareiss (or Schur) and Levinson algorithms are the most popular algorithms for solving linear systems with dense n × n Toeplitz coefficient matrix in O(n 2) arithmetic operations. Both algorithms have been generalised to solve linear systems whose n × n coefficient matrices A are not necessarily Toeplitz (in O(n 3) operations). We show in this paper that the generalised Levinson algorithm is a direct consequence of the generalised Bareiss algorithm, thereby considerably simplifying its presentation in comparison to previous work.
Unable to display preview. Download preview PDF.
- Delosme J.-M. and Ipsen I.C.F., “From Bareiss’ Algorithm to the Stable Computation of Partial Correlations”, Journal of Computational and Applied Mathematics, To appear, 1989.Google Scholar
- Delsarte P., Genin Y. and Kamp Y., “Generalized Schur Positivity Test and Levinson Recursion”, Proc. Europ. Conf. Circuit Theory and Design, pp. 321–3, 1983.Google Scholar
- Golub G.H. and van Loan C.F., “Matrix Computations”, The Johns Hopkins Press, 1983.Google Scholar
- Ipsen I.C.F., “Systolic Algorithms for the Parallel Solution of Dense Symmetric Positive-Definite Toeplitz Systems”, Numerical Algorithms for Modern Parallel Computer Architectures“, pp. 85–108, Schultz M.H., Springer Verlag, 1988.Google Scholar