Products of random matrices: dimension and growth in norm (Q988758)

Let \(X_i\) be a sequence of independent \(N \times N\) random matrices and \(\prod_n = X_n \cdots X_1\). In the paper, the question of how the speed of convergence of the quantity \(n^{-1}\log\|\prod_n\|\) (the Furstenberg-Kesten theorem [\textit{H. Furstenberg} and \textit{H. Kesten}, Ann. Math. Stat. 31, 457--469 (1960; Zbl 0137.35501)]) characterizing the stability and properties of the system \(\prod_n\) depends on the dimension \(N\) of the matrices is investigated. When requiring that the matrices \(X_i\) are rotationally invariant, that is, the distribution of matrix elements does not depend on the choice of the basis, then under another additional assumptions about the asymptotic behavior of \(n^{-1}\log\|\prod_n\|\) it is proved that the speed of the convergence of \(n^{-1}\log\|\prod_n\|\) does not decrease as the dimension \(N\) of the matrices increases.