On the Maximal Eigenvector of a Positive Matrix

From MaRDI portal





A classical theorem of Perron states that an \(n\)-square matrix \(A=(a_{i j})\) has a simple maximal positive eigenvalue \(\rho\) and a positive eigenvector \(x=(x_1, \ldots, x_n)\), the maximal eigenvector, corresponding to \(p\). Let \(\gamma=\max\limits_{i, j} \frac{x_i}{x_j}\). \textit{A. Ostrowski} [J. Lond. Math. Soc. 27, 253--256 (1952; Zbl 0046.01301)] showed that (1) \(\gamma \le \max\limits_{i, j} a_{i, j} / \min\limits_{i, j} a_{i, j}\). In the present paper the bound in (1) is improved. It is shown that (2) \( \gamma \le \max\limits_{j, s, t} (a_{s j}/a_{t j})\) and that equality can hold in (2) if and only if the \(p\)-th row of \(A\) is a multiple of the \(q\)-th row, for some pair of indices \(p\) and \(q\) satisfying \(a_{ph}/a_{q h}=\max\limits_{j, s, t}(a_{s, j}/a_{t j}))\).












This page was built for publication: On the Maximal Eigenvector of a Positive Matrix

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5623835)