A note on rounding-error analysis of Cholesky factorization (Q1822196): Difference between revisions

The computed Cholesky factor \(L\) of a given symmetric positive definite \(n\times n\) matrix \(A\) fulfills \(LL^ T=A+E\). \textit{J. H. Wilkinson} [Tr. Mezhdunarod. Kongr. Mat., Moskva 1966, 629--640 (1968; Zbl 0197.13301)] has given a bound for \(\| E\|_ 2\). In this paper a sharper bound is given for \(\| E\|_ F\). In the first part of the paper, the author gives an exact analysis of the Cholesky factorization. Part two deals with the numerical process.