{"id":"99648931-e5a6-438c-a90a-067864b5cb41","arxiv_id":"2601.08208","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Introduces an intrinsic critical set for surface diffeomorphisms and shows, under 'far-from-homotheties', that it is empty exactly when the invariant set has a dominated splitting.","lead":"Defines a 'critical set' for surface diffeomorphisms: points that have a tangent direction whose derivative along the projective bundle never contracts, in either time direction, as a 2D analog of critical points in 1D dynamics. Proves that, under a non-degeneracy condition, an empty critical set is equivalent to the existence of a dominated splitting, and gives structural theorems for mildly dissipative disk diffeomorphisms.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem A rests on Lemma 21, whose one-line proof does not establish the existence of H^± points under far-from-homotheties; the subsequent transfer from y∈α(x) to a vector at x is also not justified.","rationale":"The central claim is Theorem A: under far-from-homotheties, empty critical set is equivalent to dominated splitting. The proof's most fragile link is Lemma 21, which supplies the backward-expanding direction needed to launch the Pliss argument. The one-line proof is not convincing: Definition 2 is a uniform-in-time pointwise condition and does not obviously imply the existence of an ergodic measure with distinct Lyapunov exponents; and even if such a measure exists, Oseledets only gives asymptotic growth, whereas H^± demands the stronger all-n inequality. The subsequent transfer from y∈α(x) to a vector at x is also not valid as written, because G^{m^-}(u) is based at f^{m^-}(y), not at x. These are genuine proof gaps in the central argument. However, the overall strategy is plausible, the construction is local and likely repairable, and I have no counterexample to the theorem itself. The reader's CONDITIONAL verdict is therefore appropriate. I do not see a reason to reject or accept outright; the gaps are specific and testable, but they do not by themselves disprove the claim. The reader identified far-from-homotheties as the weakest assumption; my concern is more specifically Lemma 21 and the transfer inside the proof of Theorem A, so agreement is partial rather than full.","tokens_in":30860,"tokens_out":19494,"duration_ms":195719,"concrete_test":"Analytical check: give a complete proof of Lemma 21. Specifically: (i) prove from Definition 2 that there exists an f-invariant probability measure μ with distinct Lyapunov exponents λ_+(μ) > λ_-(μ), or exhibit a C¹ cocycle over a compact invariant set that is far from homotheties but for which every invariant measure is conformal; (ii) prove the passage from Oseledets asymptotics to a point/direction with |g^n(v)| ≥ 1 for every n ≥ 0, e.g. by an explicit Pliss-lemma argument and an orbit shift. In the same pass, correct the transfer step: write the exact projective-cocycle identity expressing a vector at x in terms of u at y∈α(x) and verify (1) with the correct orbit segment and exponent. If either step requires an additional hypothesis, or if the corrected transfer yields a weaker exponent (e.g. (1-3δ)^k), then Theorem A's proof has a real hole that must be repaired before the claim is","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem A depends on Lemma 21, which claims that under far-from-homotheties both H^-(f,Λ) and H^+(f,Λ) are nonempty. This is the starting point of the Pliss construction: without a backward-expanding vector u at some y∈α(x), inequality (1) has no source and the whole critical-point construction collapses. The proof of Lemma 21 is one sentence: 'direct application of Oseldets's theorem: by hypothesis, there is an ergodic measure not supported on neither the orbit of a sink nor a repeller; therefore, it has two Lyapunov exponents one non-negative and the other non-positive.' This contains two unproved steps: (a) Definition 2 is a pointwise, uniform-in-time non-conformality condition and does not by itself produce an ergodic measure with distinct Lyapunov exponents; (b) even with such a measure, Oseledets gives asymptotic behavior |g^n(v)| ~ e^{λn} for μ-a.e. point, not the all-n inequality |g^n(v)| ≥ 1 required by H^±. Passing from asymptotic to all-n is a Pliss-type statement that must be proved and may fail if the direction repeatedly returns with temporary dips below 1. There is also a concrete transfer error in the use of Lemma 21: from y∈α(x) the text chooses m^- with f^{-m^-}(x) close to x and then sets v:=G^{m^-}(u)∈T_xM. But G^{m^-}(u) lies in T_{f^{m^-}(y)}M; no choice of m^- makes f^{-m^-}(x) close to x from y∈α(x), nor does it make f^{m^-}(y)=x. Thus the backward estimate (1) is not derived as written. These are internal proof gaps, not known counterexamples, but they are load-bearing for the central equivalence.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an intrinsically defined 'critical set' Crit(f,Λ) for surface diffeomorphisms, consisting of points with a tangent direction whose projective derivative norm |g^n(v)| is bounded below for all integer times. The main result, Theorem A, states that if the restriction f|Λ is 'far from homotheties', then Crit(f,Λ) is empty if and only if Λ admits a one-dimensional dominated splitting. The paper also states sufficient conditions for being far from homotheties (Theorems B and C), several properties of the critical set (Propositions 4 and 5), a structural result for mildly dissipative diffeomorphisms (Theorem D), and a finiteness theorem for non-trivial homoclinic classes of Misiurewicz diffeomorphisms (Theorem E). The overall goal is to provide a two-dimensional analogue of the role played by critical points in one-dimensional dynamics.","tokens_in":31318,"tokens_out":9129,"duration_ms":86161,"significance":"If Theorem A is correct, it gives a clean, intrinsic criterion for dominated splitting and is a meaningful simplification of the earlier notion from [PRH]. The proposed critical set is natural and the applications to tangencies, Hénon-type examples, and Misiurewicz diffeomorphisms indicate that the notion could be a useful organizing tool. The paper is ambitious and likely to be influential. However, the current version contains serious proof gaps in the central argument, and the supporting Theorem B is presented in a way that cannot be checked. The strengths are the conceptual framework, the explicit conjecture-style theorems, and the variety of examples; but the technical execution is not yet at the standard of a publishable paper.","major_comments":[{"comment":"The proof of Lemma 21 is a single sentence invoking Oseledets' theorem. Definition 2 is a uniform, pointwise non-conformality condition; it does not by itself produce an ergodic measure with one non-negative and one non-positive Lyapunov exponent. Even if such a measure exists, Oseledets gives only asymptotic growth along a set of full measure, whereas H^± requires |g^{-n}(v)| ≥ 1 for all n. A Pliss-type argument is needed and is not supplied. Since Lemma 21 is the starting point of the construction in Theorem A, this is a load-bearing gap.","section":"§4, Lemma 21"},{"comment":"The transfer from y ∈ α(x) to a vector at x is not justified. The text says 'Choosing m^- sufficiently large in such a way that f^{-m^-}(x) is sufficiently close to x' — this is generally impossible. Later v is defined as G^{m^-}(u) ∈ T_xM, but u lies in T_yM and G^{m^-}(u) lies in T_{f^{m^-}(y)}M; unless f^{-m^-}(x)=y, which is not arranged, the vector is not at x. Thus inequality (1) is not derived as written. This step is essential for the Pliss argument.","section":"§4, proof of Theorem A"},{"comment":"The statement of Theorem B involves no constants b or λ, but the proof uses both without definition or introduction. The notation 'T^1_λ M' is unexplained, and the line '∫ log(g)dν_+ = -∫ log(g)dν_- = λ_+(μ)-λ_-(μ)' is not compatible with the theorem's hypothesis as stated. The proof also does not state how the exponent gap γ controls the δ in Definition 2. A complete, self-contained proof with all constants defined is required.","section":"§5, Theorem B"},{"comment":"Claim 11 is stated for sequences of 'positive integers', but it is applied to positive real numbers |g|. As a statement about positive reals the claim is false: for a = (1/2, 3, 1/2, 3), the product is 9/4 ≥ 1, but no index K satisfies both backward and forward partial products ≥ 1 for all intermediate steps. Thus the second proof of Lemma 10 collapses. Since Lemma 10 is used to connect tangencies with critical points, this needs to be fixed or removed; the first proof via Theorem A is at least conditional on Theorem A.","section":"§2, Claim 11 and Lemma 10"},{"comment":"The proof of Proposition 4 does not establish the first claimed property, namely uniqueness of the critical direction under the far-from-homotheties hypothesis. The text proves a statement about periodic critical points and then gives unrelated estimates involving the 'most contractive direction'. No argument is given that two distinct critical directions would contradict Definition 2. Since Proposition 4 is a stated result of the paper, it needs a complete proof or should be reformulated.","section":"§6, Proposition 4"}],"minor_comments":[{"comment":"Please specify δ > 0. Also, in the proof of Lemma 22 the inequalities obtained are non-strict while Definition 2 uses strict inequalities; the passage needs clarification or an epsilon adjustment.","section":"§1, Definition 2"},{"comment":"There is a typo in the statement: 'for any∈Λ' should be 'for any x∈Λ'. In the proof, 'm > m0' should be 'm ≥ m0' if equation (3) is meant to hold for all sufficiently large m.","section":"§4, Lemma 22"},{"comment":"The notation G^m.v, G^{m^+}.v, and the super/subscript placement in m_0^+, m^+ is confusing. Please use a consistent convention, e.g. G^m(v).","section":"§4, proof of Theorem A"},{"comment":"The phrase 'for any invariant measure, the difference of the Lyapunov exponents of any regular points' is imprecise. It should say 'for every ergodic invariant measure, the two Lyapunov exponents differ by at least γ'. Also define 'regular points' or avoid the term.","section":"§5, Theorem B"},{"comment":"Lemma 23 and Lemma 24 overlap in content; Lemma 24 is a quantitative version of Lemma 23. Please unify and avoid duplication.","section":"§6, Lemmas 23 and 24"},{"comment":"The manuscript contains numerous typos ('unitarean', 'coonected', 'Moebious', 'proporties', etc.). A careful proofreading pass is needed before resubmission.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"This is a promising manuscript from leading researchers, and the main idea is attractive. However, the proof of Theorem A has a genuinely incomplete lemma and an invalid vector-transfer step, and Theorem B is not checkable as written. These are load-bearing issues for the central claims. I do not think the paper should be rejected outright — the strategy is plausible and likely repairable — but the current version would not allow a reader to verify the main theorem. I would ask for a complete rewrite of the proofs of Lemma 21, Theorem A, Theorem B, and Proposition 4, as well as correction of Claim 11. I would also ask the authors to clarify the novelty relative to [PRH], since the key dichotomy and several supporting lemmas are closely related to that paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper is worth a serious referee, not because Theorem A is airtight — it isn't yet — but because the critical-set framework and the structural results are real contributions from people who know the area. The cleanest new thing is Definition 1: an intrinsic critical set that does not require dissipativity or limit-set assumptions, in contrast to [PRH]. Theorems D and E, for mildly dissipative disk diffeomorphisms, and the horseshoe-with-single-critical-orbit example (Proposition 12) go beyond prior work and look substantial. The authors are appropriately explicit that Lemma 22 is a slight variation of [PRH]'s main proposition, so the novelty claim is honest.\n\nNow the soft spots, in proportion. Lemma 21 is load-bearing and its proof is a single sentence invoking Oseledets. Far-from-homotheties is a pointwise, uniform-in-time condition; it is not obvious that it produces an ergodic measure with Lyapunov exponents of opposite signs, and even with such a measure you need an additional Pliss-type argument to get the all-n inequalities defining H^±. That step is missing. Just after that, the proof of Theorem A picks m^- such that f^{-m^-}(x) is close to x from y∈α(x); what α(x) gives you is f^{-n}(x)→y, not proximity to x. As written, the vector transfer from y to x does not follow. Both of these look fixable, but they are real gaps in the central equivalence.\n\nTheorem B's proof does not match its statement: the statement has no b, the proof uses b and λ as if it were the dissipative case. Needs a rewrite. On Claim 11, I disagree with the reader's report: saying \"positive integers\" is a typo; for positive reals the claim is true, by taking K where the cumulative product is minimal. So that concern is minor.\n\nMy overall read: the architecture — empty critical set exactly detects dominated splitting under far-from-homotheties — is likely correct, but the manuscript does not yet rigorously close it. The structural theorems D/E are long and I have not verified every step; they rest on the same framework. Who gets value: surface dynamics people working on dominated splitting, critical points, and the 1D-2D analogy. Send it to a serious referee, and make sure the referee has the full proof of Lemma 21 and a clean Theorem B. If those come back solid, it will be a useful paper.","headline":"Critical set notion worth taking seriously; Theorem A's proof has a load-bearing gap around Lemma 21 and the vector transfer, but the structural results and the overall framework justify a serious referee.","tokens_in":31783,"tokens_out":5083,"would_cite":true,"duration_ms":49009,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37D30","37D20"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper defines an intrinsic critical set for surface diffeomorphisms and proves that, under a far-from-homotheties condition, a compact invariant set admits a dominated splitting if and only if this critical set is empty.","keywords":["critical set","surface diffeomorphisms","dominated splitting","hyperbolicity","projective tangent bundle","Misiurewicz diffeomorphisms","far from homotheties","homoclinic classes"],"falsifier":"Construct a C^1 surface diffeomorphism with a compact invariant set Λ that is far from homotheties, has empty critical set, and yet admits no dominated splitting; or find an invariant set with a dominated splitting that still contains a point with a direction v satisfying |g^n(v)| ≥ 1 for all n ∈ Z. Either example would refute Theorem A.","tokens_in":30760,"feed_emoji":"🌀","tokens_out":5807,"duration_ms":47773,"temperature":0.7,"pith_summary":"This paper proposes a definition of 'critical set' for smooth surface diffeomorphisms, intended to play the same role that critical points play in one-dimensional dynamics: the set of points where hyperbolicity fails. Its central theorem states that for a compact invariant set that is 'far from homotheties' — meaning along every orbit the derivative is infinitely often far from conformal — the presence of a dominated splitting (a hyperbolic cone structure) is exactly equivalent to the critical set being empty. In other words, the critical set is a precise obstruction to domination, and thus, by earlier work the paper cites, a precise obstruction to hyperbolicity for generic C^2 surface diffeomorphisms. The paper also shows that periodic critical points are necessarily attracting in the dissipative case, and it uses the notion to define two-dimensional Misiurewicz diffeomorphisms, for which it proves a finiteness theorem for homoclinic classes.","feed_headline":"Critical set for surface maps finds exactly when domination fails","feed_subtitle":"If the new criterion holds, absent critical points guarantee a dominated splitting — the 2-D counterpart of 1-D hyperbolicity theorems.","key_machinery":"The central object is the projective tangent bundle map G on the unit tangent bundle, whose fiber derivative g controls how directions expand and contract. A critical point is defined exactly as a point with a direction v for which |g^n(v)| ≥ 1 for all n ∈ Z, i.e., a direction that is neither forward nor backward contracted. The far-from-homotheties condition (no orbit whose derivative stays within (1±δ)^n for all n) is the hypothesis that makes the critical set a sharp detector; the proof of Theorem A converts the failure of dominated splitting, via Pliss's lemma, into the existence of such a direction.","core_discovery":"The central claim, Theorem A, is that if f is a C^1 surface diffeomorphism and Λ is a compact invariant set on which f is far from homotheties, then Λ has a one-dimensional dominated splitting E⊕F if and only if the critical set Crit(f,Λ) is empty. A point is critical when a direction exists whose iterated projective derivative never contracts: |g^n(v)| ≥ 1 for all n ∈ Z. The 'only if' direction is straightforward — domination forces every direction to contract either forward or backward — while the 'if' direction is proved by contradiction using Pliss's lemma: if the critical set is empty, the absence of a uniformly contracting direction contradicts far-from-homotheties. The same ideas yiel","pith_inferences":["A testable extension: the far-from-homotheties assumption might be replaceable by a milder 'non-conformal almost everywhere' condition; a counterexample would be a compact invariant set with empty critical set and no dominated splitting but with some conformal segments.","The critical set may serve as a combinatorial bookkeeping device for renormalization of dissipative surface maps; a natural next step is to explore whether Misiurewicz diffeomorphisms carry SRB measures when the critical set is non-recurrent, mirroring one-dimensional Misiurewicz maps.","The notion of critical set might extend to higher-dimensional dynamics by replacing directions with subspaces; the analog of Theorem A would then be a statement about partial hyperbolicity and dominated splitting in higher rank."],"forward_implications":["If Theorem A holds, then for surface diffeomorphisms that are far from homotheties, the critical set is the complete obstruction to the existence of a dominated splitting; combined with the earlier work on dominated splitting that the paper cites, this yields a characterization of hyperbolicity for generic C^2 surface diffeomorphisms: empty critical set iff hyperbolic set.","The finiteness and semi-continuity properties of the critical set imply that small perturbations cannot create critical points far from existing ones, so the obstruction to domination is stable under C^1 perturbations when far from homotheties.","For dissipative diffeomorphisms (Theorem C), the critical set becomes a finite, robust object away from finitely many sinks, meaning the theory applies to a large class of dissipative systems.","Theorem D gives a dichotomy: the accumulation set of any unstable branch of a saddle in a mildly dissipative disk diffeomorphism is either a (semi-)attracting periodic point, a normally hyperbolic attracting arc, or contains a critical point.","Theorem E: for mildly dissipative Misiurewicz diffeomorphisms, the number of non-trivial generalized homoclinic classes is finite, and all sufficiently long-period saddles outside those classes lie on periodic normally hyperbolic arcs."],"fun_headline_variants":["Surface maps: no critical directions iff dominated splitting","Critical set criterion: empty set equals hyperbolicity for maps","When critical set vanishes, surface diffeos gain dominated splitting","New theorem ties critical set to domination in surface maps","Surface diffeos: domination iff no critical directions exist"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is the 'far from homotheties' condition — that along every orbit the derivative cocycle is infinitely often far from conformal; without it the equivalence between empty critical set and dominated splitting is not established.","fun_headline_variants_meta":{"raw":{"variants":["Surface maps: no critical directions iff dominated splitting","Critical set criterion: empty set equals hyperbolicity for maps","When critical set vanishes, surface diffeos gain dominated splitting","New theorem ties critical set to domination in surface maps","Surface diffeos: domination iff no critical directions exist"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000156,"raw_usage":{"total_tokens":957,"prompt_tokens":550,"completion_tokens":407,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":294,"completion_tokens_details":{"reasoning_tokens":327}},"tokens_in":294,"tokens_out":407,"duration_ms":5187,"temperature":1.0,"reasoning_tokens":327,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T10:52:53.149951+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a C^1 surface diffeomorphism with a compact invariant set Λ that is far from homotheties, has empty critical set, and yet admits no dominated splitting; or find an invariant set with a dominated splitting that still contains a point with a direction v satisfying |g^n(v)| ≥ 1 for all n ∈ Z. Either example would refute Theorem A.","supporting_citations":[],"review_version":1}