scientific article; zbMATH DE number 3958571

From MaRDI portal
Publication:3727249





[For the entire collection see Zbl 0591.00036.] One approach to computing the eigenvalues of a real matrix is to compute the roots of its characteristic polynomial. Since the eigenvalues are very sensitive to very small perturbation of its coefficients, an interesting alternative is to compute the characteristic polynomial exactly. It was suggested by \textit{E.Krishnamurthy}, \textit{T. Mahadeva Rao} and \textit{K. Subramanian} [Proc. Indian Acad. Sci., Sect. A 81, 58-79 (1975; Zbl 0324.10004)] that the problems of overflow and truncation error associated with floating-point arithmetic can be overcome using p-adic arithmetic. This is similar to modular arithmetic in which arithmetic operations are defined on coded representations of rational numbers and correspond to the rational arithmetic operations. We show how such an arithmetic system can be combined with Danilevskii's method of reducing a matrix to Frobenius standard form to obtain an exact algorithm for computing the characteristic polynomial of an integer matrix.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3727249)