Invariant tori for a class of affined Anosov mappings with quasi-periodic forces

Hang Zhao; Xiao Ma; Xinyu Bai; Zeng Lian

REVIEW 3 major objections 4 minor 15 references

Invariant tori for a class of affined Anosov mappings with quasi-periodic forces

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For affine Anosov maps with quasi-periodic forcing, one integer m—fixed by the winding numbers of h—describes every invariant torus, and the degree-m tori count grows at the topological entropy rate.

desk verdict Plausible and interesting theorem, but the written proof breaks at two false claims; uniqueness of the degree m is not established as written. read the letter →

arxiv 2411.14844 v1 pith:JFQ7SAOR submitted 2024-11-22 math.DS

Xinyu Bai , Zeng Lian , Xiao Ma , Hang Zhao This is my paper · ORCID

classification math.DS MSC 37D2037C35

keywords affineAnosovmappingquasi-periodicforceinvarianttorusrandomperiodicpointtopologicalentropyskewproductwindingnumberhyperbolictoralautomorphism

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper studies skew-product maps on $\mathbb{T}\times\mathbb{T}^2$ given by $(\omega,x)\mapsto(\omega+\alpha, Ax+h(\omega))$, where $\alpha$ is irrational, $A$ is a hyperbolic matrix with integer entries, and $h$ is continuous. It proves that there is a unique positive integer $m$, determined only by the winding numbers of the two components of $h$, with the following property: every invariant torus of the system has degree $m$, and the number of degree-$m$ tori invariant up to time $n$ is finite and grows exponentially with rate equal to the topological entropy. This transplants the classical Bowen principle—periodic orbits counted at the entropy rate—to a setting without genuine periodic orbits, where tori take over the counting role. The integer $m$ is the smallest one that clears the denominators of a cohomological equation relating the degree of the torus to the degree of the forcing.

What carries the argument

The load-bearing construction is the induced system $\phi_m(\omega,x)=(\omega+\alpha/m,\,Ax+h(m\omega))$, together with the finite-to-one projection $K_m(\omega,x)=(m\omega,x)$ that yields $h_{\mathrm{top}}(\phi)=h_{\mathrm{top}}(\phi_m)$. An invariant torus of degree $m$ of $\phi$ is equivalent to a continuous random periodic point of $\phi_m$; the space of such points is an affine space modelled on the random periodic points of the homogeneous system $(\omega,x)\mapsto(\omega+\alpha,Ax)$. The degree condition on $m$ comes from comparing winding numbers in the lifted equation $g(\omega+\alpha)=Ag(\omega)+h(\omega)$, which forces $\deg(g)=m(I-A)^{-1}\deg(h)$ to be an integer vector.

What would settle it

For a concrete system, e.g., $A=\begin{pmatrix}2&1\\1&1\end{pmatrix}$, $\alpha=\sqrt{2}-1$, $h(\omega)=(\cos 2\pi\omega,0)$, one can solve the period-$n$ cohomological equation for steps $\alpha$ and $\alpha/2$ by Fourier series: a solution for step $\alpha/2$ will not satisfy the step-$\alpha$ equation pointwise, so $A^n_1\not\subset A^n_2$ as function sets, which isolates the gap in the uniqueness proof; the entropy formula itself can be tested by enumerating degree-1 tori for small $n$ and checking whether $(1/n)\log\sharp$ approaches $\log\lambda$.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.2. Let $m$ be the smallest positive integer satisfying $m(I-A)^{-1}(\deg h_1,\deg h_2)^T\in\mathbb{Z}^2$. Then $G(\phi)=\cup_n G(\phi;n,m)$, each $\sharp G(\phi;n,m)$ is finite, and $\lim_{n\to\infty}(1/n)\log\sharp G(\phi;n,m)=h_{\mathrm{top}}(\phi)$. In other words, the degree of every invariant torus is forced by a Diophantine integrality condition on the forcing, and the tori of that degree reproduce the exponential complexity of the system. The argument first produces one invariant torus of degree $m$ by solving a cohomological equation with Fourier series, then shows that all other invariant tori are affine translates of the homogeneous system's tori, which are exactly graphs of periodic points of $A$.

Load-bearing premise

The uniqueness of the degree $m$ in Theorem 1.2 rests on the assertion in Section 4 that every random periodic point of the induced system with step $\alpha/(lm)$ is already one for step $\alpha/m$; as stated this inclusion is false, although the intended embedding of solution spaces would still preserve the cardinality comparison.

Editorial extensions

If this is right

Because the growth rate equals $h_{\mathrm{top}}(\phi)=\log\lambda$, where $\lambda$ is the expanding eigenvalue of $A$, the forcing $h$ changes which tori exist but not how fast they proliferate.
All invariant tori for every iterate $n$ share the same degree $m$, so there are no tori of other degrees to account for.
The number of degree-$m$ tori up to time $n$ equals the number of periodic points of $A$ of period at most $n$, giving an exact combinatorial dictionary between tori and periodic orbits of the linear part.
The degree $m$ is computable directly from the winding numbers of $h_1$ and $h_2$, so one can read off the allowed torus degree before solving any equations.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural testbed is to replace $\mathbb{T}^2$ with $\mathbb{T}^d$: the same cohomological integrality condition should define a degree vector, and the entropy identity should persist whenever the cohomological equation has a continuous solution, which would show that the mechanism is the affine structure rather than the dimension.
When $\deg(h)=0$ the theorem gives $m=1$, so invariant tori reduce to graphs of continuous random fixed points of $\phi$ itself, reproducing classical random periodic point counting in that case.
The set inclusion (4.5) used for uniqueness is not literally correct: a solution of the step-$\alpha/m$ system embeds into the step-$\alpha/(lm)$ system through $g(\omega)\mapsto g(l\omega)$, but the two solution spaces are not identical as function sets; the cardinality identity (4.6) still goes through, so the theorem's conclusions are recoverable, but the written proof needs this repair.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear [Colleague],

You should know that this paper has a genuinely interesting theorem but the written proof is not reliable. Theorem 1.2—that in affinely forced Anosov skew products on T×T² the invariant tori of a single degree m grow at the topological entropy rate—is the first result of this kind for systems without periodic orbits, and it deserves attention. But two load-bearing claims in the proof are false, and the uniqueness part is not established.

What is good: the idea of counting invariant tori of fixed degree instead of periodic orbits is new and natural. Lemma 3.2, showing that the random periodic points of the forced system form an affine space over the homogeneous system, is clean and correct. The Fourier construction in Lemma 2.3 for the C² case is elegant. There are no fitted parameters and the argument is self-contained except for Claim 2.4, deferred to Lemma 7.3 of [6], a published paper by one of the authors; that is a technical regularity lemma, not the main result, so I do not see circularity.

Soft spots:

1. Lemma 2.5 claims that ||r||_{C0} < 1 implies deg(r) = 0. That is false. A map like r(ω) = (ω + 1/2 mod 1, 0) has C0 norm ≤ 1/2 and degree 1. The proof of Proposition 2.1 can be repaired by choosing the C² approximation with the same degree as h, so deg r = 0 holds by construction, but the lemma as stated is wrong.

2. Equation (4.5) is also false. A random periodic point of ϕ_m of period n does not generally satisfy the ϕ_{lm} relation for n > 1; the forcing terms accumulate as h(mω + jα) under ϕ_m but as h(lmω + jα/l) under ϕ_{lm}. The map g(ω) ↦ g(lω) sends fixed points of ϕ_m to fixed points of ϕ_{lm}, but it does not send period-n solutions to period-n solutions for n ≥ 2. So the inclusion A^n_m ⊆ A^n_{lm} fails, and the equality A^n_m = A^n_{lm} used to conclude that all tori have degree m does not follow. A plausible repair is to show from the Fourier support that every solution of ϕ_{lm} has period 1/l, so the induced torus has degree m, but that argument is absent.

The central theorem is plausible and likely true. But the written proof breaks exactly where the uniqueness of m is derived. This paper deserves a serious referee, but not acceptance as is; it needs major revision. I would send it out with a request to fix the two points above.

Best, [You]

Referee Report

3 major / 4 minor

Summary. The paper studies the skew-product system φ(ω,x)=(ω+α, Ax+h(ω)) on T×T², where A∈GL(2,Z) is hyperbolic and h is continuous. It introduces φ^n-invariant tori of degree m as multi-valued graphs and proves Theorem 1.2: for m the smallest positive integer with m(I-A)^{-1}(deg h_1, deg h_2)^T∈Z², every invariant torus has degree m; for each n there are finitely many invariant tori of degree m of order at most n; and the exponential growth rate of their number equals the topological entropy of φ. The proof proceeds by an existence result via C² approximation and a linear cohomological equation, a classification of invariant tori of the homogeneous system, a coset description of the random periodic points, and an entropy comparison through a finite-to-one factor map.

Significance. If established, the theorem gives a clean, explicitly computable degree for invariant tori of quasi-periodically forced affine Anosov maps and relates their counting function to topological entropy, complementing the density and uncountability results of Huang–Lian–Lu. The strategy of passing to induced systems, using a finite-to-one factor to compare entropies, and counting through the homogeneous system is natural and, in its main lines, economical. The statement is also concretely testable: the degree m is determined by the homology class of h, and the growth rate is a genuine prediction rather than a fitted quantity. However, several load-bearing proof steps are currently incorrect or incomplete, so the result should not be regarded as established in this version.

major comments (3)

[Section 4, Eq. (4.5)] The assertion immediately before (4.5), that every random periodic point of the induced system φ_m is automatically a random periodic point of φ_lm for every l, is false. A relation for the rotation increment α/m cannot be substituted into a relation for the increment α/(lm); the forced terms h(mω+jα) and h(lmω+jα) differ. Consequently the inclusion A_m^n ⊆ A_lm^n in (4.5) is invalid, and the derivation of A_m^n = A_lm^n collapses. This is exactly the step that proves uniqueness of the degree m in Theorem 1.2(1). The argument can likely be repaired by using the embedding T_l(g)(ω)=g(lω) from solutions of φ_m into solutions of φ_lm and then applying the cardinality identity (4.6), but as written the proof of part (1) of Theorem 1.2 is not valid.
[Lemma 2.5] The proof infers deg(r)=(0,0) from ‖r‖_{C^0}<1. This inference is false: with the standard flat metric on T, the map r(ω)=(ω mod 1, 0) has C^0 norm at most 1/2<1 but degree (1,0). The lemma as stated is therefore false. In the application in Proposition 2.1, one can choose the C² approximation ~h so close to h that deg(h−~h)=0, and then state Lemma 2.5 under the hypothesis deg(r)=0, but the current statement and proof are not valid. Since Proposition 2.1 supplies the invariant torus whose degree is claimed to be m, this is a load-bearing issue for the existence part.
[Lemma 3.1] The proof of Lemma 3.1 is incomplete. To prove that a nonconstant invariant torus of the homogeneous system cannot exist, the text asserts that (Im(˜g)\setminus{˜g(0)}) intersects W^s(0) or W^u(0) merely because those manifolds are dense. Density only forces intersection with nonempty open sets, and Im(˜g)\setminus{˜g(0)} need not contain an open set; moreover the claim that removing one point from a continuous image of T leaves a path-connected set is not true for general Peano continua. Since Lemma 3.1 gives the correspondence between invariant tori of the homogeneous system and periodic points of A, this gap is central to the counting identity (4.4). A correct proof can be obtained, for instance, by taking Fourier coefficients in g(ω+β)=A g(ω) and using that A has no eigenvalues on the unit circle for k≠0.

minor comments (4)

[Lemma 2.2, proof] In the displayed computation after (2.2), the expression g(ω+nω) should be g(ω+nα), and the equals sign before it should include the mod-1 reduction on the base coordinate.
[Lemma 2.3, final paragraph] The reference 'By Lemma 2.4' near the end of the proof should be 'By Lemma 2.2'; the numbering appears to be a typo.
[Definition 1.1 and Section 1.2] The phrase 'collection of all φ^k(k≤n)-invariant tori' defining G(ϕ;n,m) is ambiguous; it should say explicitly that a torus belongs to G(ϕ;n,m) if it is φ^k-invariant for at least one k with 1≤k≤n.
[Proposition 2.1] The notation m'∈N∩(0,m) in Lemma 2.3 is nonstandard but understandable; consider writing 1≤m'<m.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the unique degree m is computed from the forcing winding numbers, and the theorem is proved by explicit construction and counting; the only self-citation is a technical regularity lemma, and the gap at (4.5) is a correctness issue, not a circular reduction.

full rationale

The derivation is not circular. Theorem 1.2 defines m from the input data (the winding numbers of h and the hyperbolic matrix A) and then proves, rather than assumes, that invariant tori must have this degree. The existence proof (Prop. 2.1) solves a linear cohomological equation (Lemma 2.5) and a Fourier-series equation (Lemma 2.3); the homogeneity/affineness lemmas (3.1, 3.2) show invariant tori correspond to periodic points of A up to a translation by a fixed solution; Lemma 3.4 transfers entropy by conjugacy; and (4.4) equates the counts. No fitted parameter is renamed as a prediction and no stated result is used to define itself. The only self-citation is Claim 2.4, deferred to Lemma 7.3 of [6] (a published paper with overlapping author Lian); it is a parameter-free C^2 Fourier-regularity lemma used only to justify continuity of the constructed torus, not the theorem's central conclusion, so it does not constitute circularity. I separately flag, as a correctness concern, the assertion before (4.5) that every random periodic point of the induced system phi_m is automatically one of phi_lm; this inclusion is false as stated because the shifts alpha/m and alpha/(lm) differ, so the uniqueness proof for the degree m in Theorem 1.2(1) has a gap (a rescaling repair appears available). That is a mathematical error in the proof, not a circularity: the statement is not equivalent to its inputs by construction.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No free parameters, fitted constants, or new entities are introduced. The integer m is determined uniquely by the system via the degree condition m(I-A)^{-1}deg(h)∈Z²; it is not fitted to data. The only external inputs are standard theorems and the technical Lemma 7.3 from [6].

assumptions (4)

domain assumption For h∈C²(T,T²), the Fourier solution r=(r1,r2) in Lemma 2.3 is continuous and periodic (Claim 2.4); proof taken from Lemma 7.3 of [6].
Invoked in the proof of Lemma 2.3, Section 2.1; it is an unproved regularity input in this paper.
standard math The exponential growth rate of periodic points of a hyperbolic toral automorphism T_A equals h_top(T_A).
Used in Section 4, equation (4.3); this is a classical theorem of Bowen for Axiom A diffeomorphisms.
standard math Topological entropy is preserved by finite-to-one continuous factor maps and additive for direct products.
Used for the K_m finite-to-one factor and for h_top(ϕ_{m,0})=h_top(R_{α/m})+h_top(T_A) in Section 4.
domain assumption C²(T,T²) is C0-dense in C(T,T²), and degree is locally constant under C0 perturbations below the injectivity radius (the paper's threshold 1 is too large).
Used in Proposition 2.1 to approximate h by C² h̃; the as-written threshold in Lemma 2.5 is incorrect.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Invariant tori for a class of affined Anosov mappings with quasi-periodic forces." pith.science (2026). https://pith.science/paper/JFQ7SAOR

@misc{pith2026241114844,
  author       = {Pith},
  title        = {Pith review of: Invariant tori for a class of affined Anosov mappings with quasi-periodic forces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JFQ7SAOR}},
  note         = {Machine review of arXiv:2411.14844}
}

abstract

In this paper, we consider a class of affined Anosov mappings with quasi-periodic forces, and show that there is a unique positive integer $m$, which only depends on the system, such that the exponential growth rate of the cardinality of invariant tori of degree $m$ is equal to the topological entropy.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages

[6]

Dynamical complexity of Anosov systems driven by a quasi-periodic forcing

Huang, W., Lian, Z., and Lu, K.-N. Dynamical complexity of Anosov systems driven by a quasi-periodic forcing. Sci. China Math. (2023). doi:10.1007/s11425-023-2207-x

work page doi:10.1007/s11425-023-2207-x 2023
[1]

and McAndrew, M.H

Adler, R.L., Konheim, A.G. and McAndrew, M.H. Topological entropy. Trans. Amer. Math. Soc. , 114 (1965), 309--319

work page 1965
[2]

Topological entropy and axiom A, in Global Analysis, Berkeley, Calif., 1968, in: Proc

Bowen, R. Topological entropy and axiom A, in Global Analysis, Berkeley, Calif., 1968, in: Proc. Sympos. Pure Math., vols XIV-XVI 23--41, Amer. Math. Soc., Providence RI, 1968, pp. 23--41

work page 1968
[3]

Entropy for group endomorphisms and homogeneous spaces

Bowen, R. Entropy for group endomorphisms and homogeneous spaces. Trans. Amer. Math. Soc. 153(1971), 401--414

work page 1971
[4]

Periodic points and measures for Axiom A diffeomorphisms

Bowen, R. Periodic points and measures for Axiom A diffeomorphisms. Trans. Amer. Math. Soc. 154(1971), 377--397

work page 1971
[5]

Periodic orbits for hyperbolic flows

Bowen, R. Periodic orbits for hyperbolic flows. Amer. J. Math. 94(1972), 1--30

work page 1972
[7]

Lyapunov exponents, entropy and periodic orbits for diffeomorphisms

Katok, A. Lyapunov exponents, entropy and periodic orbits for diffeomorphisms. Inst. Hautes \' E tudes Sci. Publ. Math. No. 51 (1980), 137--173

work page 1980
[8]

Periodicity and Sharkovsky’s theorem for random dynamical systems

Klünger, M. Periodicity and Sharkovsky’s theorem for random dynamical systems. Stoch. Dyn. 1 (2001), no. 3, 299--338

work page 2001

Show all 15 references

[9]

Kolmogorov, A. N. Entropy per unit time as a metric invariant of automorphisms. Dokl. Akad. Nauk SSSR (N.S.) 124 (1959), 754--755

1959
[10]

Existence of periodic orbits and horseshoes for mappings in a separable B anach space

Lian, Z., and Ma, X. Existence of periodic orbits and horseshoes for mappings in a separable B anach space. J. Differential Equations 269 (2020), no 12, 11694--11738

2020
[11]

Lyapunov exponents, periodic orbits and horseshoes for mappings of H ilbert spaces

Lian, Z., and Young, L.-S. Lyapunov exponents, periodic orbits and horseshoes for mappings of H ilbert spaces. Ann. Henri Poincar\' e 12 (2011), no. 6, 1081--1108

2011
[12]

Lyapunov exponents, periodic orbits, and horseshoes for semiflows on H ilbert spaces

Lian, Z., and Young, L.-S. Lyapunov exponents, periodic orbits, and horseshoes for semiflows on H ilbert spaces. J. Amer. Math. Soc. 25 (2012), no. 3, 637--665

2012
[13]

Smooth ergodic theory of random dynamical systems

Liu, P.-D., and Qian, M. Smooth ergodic theory of random dynamical systems . Lecture Notes in Mathematics , 1606. Springer-Verlag, Berlin, 1995

1995
[14]

Existence of periodic orbits and horseshoes for semiflows on a separable Banach space. Calc. Var. Partial Differential Equations 61 (2022), no. 6, Paper No. 217, 37 pp. comment

2022
[15]

and Zheng, Z.-H

Zhao, H.-Z. and Zheng, Z.-H. Random periodic solutions of random dynamical systems. J. Differential Equations 246 (2009), no. 5, 2020--2038

2009

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [6]

Dynamical complexity of Anosov systems driven by a quasi-periodic forcing

Huang, W., Lian, Z., and Lu, K.-N. Dynamical complexity of Anosov systems driven by a quasi-periodic forcing. Sci. China Math. (2023). doi:10.1007/s11425-023-2207-x

work page doi:10.1007/s11425-023-2207-x 2023

[2] [1]

and McAndrew, M.H

Adler, R.L., Konheim, A.G. and McAndrew, M.H. Topological entropy. Trans. Amer. Math. Soc. , 114 (1965), 309--319

work page 1965

[3] [2]

Topological entropy and axiom A, in Global Analysis, Berkeley, Calif., 1968, in: Proc

Bowen, R. Topological entropy and axiom A, in Global Analysis, Berkeley, Calif., 1968, in: Proc. Sympos. Pure Math., vols XIV-XVI 23--41, Amer. Math. Soc., Providence RI, 1968, pp. 23--41

work page 1968

[4] [3]

Entropy for group endomorphisms and homogeneous spaces

Bowen, R. Entropy for group endomorphisms and homogeneous spaces. Trans. Amer. Math. Soc. 153(1971), 401--414

work page 1971

[5] [4]

Periodic points and measures for Axiom A diffeomorphisms

Bowen, R. Periodic points and measures for Axiom A diffeomorphisms. Trans. Amer. Math. Soc. 154(1971), 377--397

work page 1971

[6] [5]

Periodic orbits for hyperbolic flows

Bowen, R. Periodic orbits for hyperbolic flows. Amer. J. Math. 94(1972), 1--30

work page 1972

[7] [7]

Lyapunov exponents, entropy and periodic orbits for diffeomorphisms

Katok, A. Lyapunov exponents, entropy and periodic orbits for diffeomorphisms. Inst. Hautes \' E tudes Sci. Publ. Math. No. 51 (1980), 137--173

work page 1980

[8] [8]

Periodicity and Sharkovsky’s theorem for random dynamical systems

Klünger, M. Periodicity and Sharkovsky’s theorem for random dynamical systems. Stoch. Dyn. 1 (2001), no. 3, 299--338

work page 2001