2103.04610
Classification of Backward Filtrations and Factor Filtrations: Examples from Cellular Automata
Paul Lanthier, Thierry de la Rue
correctmedium confidence
- Category
- Not specified
- Journal tier
- Strong Field
- Processed
- Sep 28, 2025, 12:56 AM
- arXiv Links
- Abstract ↗PDF ↗
Audit review
The paper proves: (i) the factor filtration built from the deterministic automaton τ is not dynamically standard, and (ii) for sufficiently large noise ε, the factor filtration of the PCA τε is dynamically standard. Both claims are established rigorously via dynamical I-cosiness/entropy arguments and an explicit immersion into a product-type noise filtration, respectively. By contrast, the model’s Part (i) incorrectly concludes that the underlying static filtration F^{τ−1} is not standard (the paper shows it is of product type), and its I-cosiness argument conflates static and dynamical notions. Part (ii) of the model gives a plausible alternative (Dobrushin) route to dynamic standardness at high noise, consistent in spirit with the paper but not the method used there.
Referee report (LaTeX)
\textbf{Recommendation:} minor revisions
\textbf{Journal Tier:} strong field
\textbf{Justification:}
The paper cleanly separates the static (Vershik) and dynamical classifications of filtrations arising from cellular automata, exhibiting a natural example where the two diverge. The deterministic case uses an elegant entropy argument to rule out dynamical I-cosiness, while the noisy case builds a nontrivial immersion into a product-type factor by envelope-PCA and percolation ideas. The exposition is generally clear; I recommend minor revisions to streamline the link between dynamical I-cosiness and the immersion construction and to highlight quantitative regimes (e.g., any explicit noise thresholds when available).