English

Computation of Lyapunov exponents of matrix products

Dynamical Systems 2025-01-22 v1 Probability

Abstract

For mm given square matrices A0,A1,,Am1A_0, A_1, \cdots, A_{m-1} (m2m\ge 2), one of which is assumed to be of rank 11, and for a given sequence (ωn)(\omega_n) in {0,1,,m1}N\{0,1, \cdots, m-1\}^\mathbb{N}, the following limit, if it exists, L(ω):=limn1nlogAω0Aω2Aωn1L(\omega):=\lim_{n\to \infty} \frac 1n \log \|A_{\omega_0} A_{\omega_2}\cdots A_{\omega_{n-1}}\| defines the Lyapunov exponent of the sequence of matrices (Aωn)n0(A_{\omega_n})_{n\ge 0}. It is proved that the Lyapunov exponent L(ω)L(\omega) has a closed-form expression under certain conditions. One special case arises when AjA_j's are non-negative and ω\omega is generic with respect to some shift-invariant measure; a second special case occurs when AjA_j's (for 1j<m1\le j<m) are invertible and ω\omega is a typical point with respect to some shift-ergodic measure. Substitutive sequences and characteristic sequences of B\mathcal{B}-free integers are considered as examples. An application is presented for the computation of multifractal spectrum of weighted Birkhoff averages.

Keywords

Cite

@article{arxiv.2501.11941,
  title  = {Computation of Lyapunov exponents of matrix products},
  author = {Aihua Fan and Evgeny Verbitskiy},
  journal= {arXiv preprint arXiv:2501.11941},
  year   = {2025}
}