Limit theorems for a strongly irreducible product of independent random matrices under optimal moment assumptions
Let $ν$ be a probability distribution over the semi-group of square matrices of size $d \ge 2$ over a locally compact field $\mathbb{K}$, \textit{e.g.} $\mathbb{R}$. We consider the random walk $\overlineγ_n := γ_0\cdotsγ_{n-1}$ for $(γ_k)_{k \in \mathbb{N}}$ independent of law $ν$. Let $s_1 \ge s_2 \ge \dots \ge s_d$ be the singular values given by the Cartan projection. Under a contraction assumption on $ν$, we show that $(\log\frac{s_1}{s_2}(\overlineγ_n))_{n \in\mathbb{N}}$, escapes to infinity linearly and satisfies exponential large deviations inequalities below its escape rate. This extends the notion of simplicity of the top Lyapunov exponent. We also show that the image of a generic line by $\overlineγ_n$ as well as its eigenspace of maximal eigenvalue both converge to the same random line $\ell^\infty$ at an exponential speed. If we moreover assume that $ν$ is supported on the group of invertible matrices and that the push-forward distribution $N_*ν$ is $\mathrm{L}^p$ for $N: g \mapsto\log\|g\|\|g^{-1}\|$ and for some $p > 0$, then we show that $- \log\mathrm{d}(\ell^\infty, H)$ is uniformly $\mathrm{L}^p$ for all proper subspace $H \subset \mathbb{R}^d$. For $p = 1$, we moreover show that the rescaled logarithm of each coefficient of $\overlineγ_n$ almost surely converges to the top Lyapunov exponent. To prove these results, we do not rely on the existence of the stationary measure nor on the existence of the Lyapunov exponents. Instead we describe an effective way to group the i.i.d. factors into i.i.d. random words that are somehow aligned in the Cartan decomposition. We moreover have an explicit control over the moments.