2208.03575
Upper bound on the regularity of the Lyapunov exponent for random products of matrices
Jamerson Bezerra, Pedro Duarte
correcthigh confidence
- Category
- Not specified
- Journal tier
- Strong Field
- Processed
- Sep 28, 2025, 12:56 AM
- arXiv Links
- Abstract ↗PDF ↗
Audit review
The paper proves that for finitely supported measures on SL2(R) with L(µ)>0 that are not uniformly hyperbolic, the Lyapunov exponent cannot be locally α-Hölder around µ for any α>H(µ)/L(µ). In its tangency-based formulation (Theorem A), it also exhibits an analytic one-parameter family through µ witnessing the failure of α-Hölder regularity when µ has a heteroclinic tangency . The model’s solution invokes exactly this theorem and the standard “unfolding a tangency” mechanism, giving a correct argument that aligns with the paper’s core strategy. Minor overreach: the model asserts a specific cusp lower bound |E|^{H/L}·|log|E|| along a rotation family; the paper’s method builds analytic Schrödinger-type families and proves the non-α-Hölder threshold via matchings/IDS counting and Thouless’ formula, without explicitly establishing that precise cusp for an arbitrary rotation path . Overall, both are consistent; the model’s proof is essentially an application of the paper’s main result, with consistent mechanism and assumptions.
Referee report (LaTeX)
\textbf{Recommendation:} minor revisions \textbf{Journal Tier:} strong field \textbf{Justification:} The paper establishes a clear entropy/Lyapunov threshold obstructing Hölder regularity of the Lyapunov exponent for finitely supported SL2(R) products in the non-uniformly hyperbolic regime. The argument is well-structured, combining Schrödinger embeddings, tangency proliferation, matchings, and IDS via the Thouless formula. It tackles a natural and relevant question and yields an appealing quantitative boundary. A few places could better signpost the flow of estimates and where key tools (e.g., the Avalanche Principle) enter.