{"id":"ae2888e1-b61c-4eb0-9d7d-bd057bd07f54","arxiv_id":"2608.04862","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper claims the C^1-norm threshold (1−√λ)^2 is sharp for invariant graph regularity in a dissipative toy model, but the proof only bounds the maximum of a key derivative, not the minimum.","lead":"A math paper studies a simple family of dissipative twist maps and claims to locate the precise perturbation size where smooth invariant graphs can start to develop corners. If correct this would sharpen a classical persistence theorem, but a key part of the proof appears to omit a necessary bound.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2's construction only bounds sup G' above and never ensures inf G' ≥ 2√λ; in fact the Step 3 template forces inf G' < 2√λ, so the resulting φ has C1 norm strictly larger than the claimed threshold (1−√λ)^2.","rationale":"The reader's verdict REJECT is supported and strengthened. The reader identified the missing lower bound on G' as the weakest assumption; the stress-test shows that the lower bound is not merely unproved but is actively violated by the construction in Section 3. For any decreasing template ψ with ψ(0) = a > 1, ψ(1) = b < 1 and ∫ψ = 1, the strict inequality ∫_0^{u0} ψ > u0 at ψ(u0) = √λ forces G' < 2√λ at some point, making |φ'| exceed (1−√λ)^2. This directly contradicts Theorem 2's assertion ∥φλ∥_{C1} = (1−√λ)^2, so the threshold sharpness claim is not established. The rest of the paper, especially the persistence result Theorem 1 and the accompanying normal hyperbolicity analysis, appears plausible and may be correct; the problem is confined to the sharpness construction. The concern is internal consistency rather than a disagreement with prior consensus. A corrected construction might evade the lower-bound failure by using a non-monotone template or by modifying the graph-transform argument, but the present manuscript does not provide one. No ad hominem is intended; the critique is strictly on the mathematical argument.","tokens_in":20611,"tokens_out":9434,"duration_ms":94291,"concrete_test":"Take λ = 1/4 and the linear template ψ(t) = a + (b − a)t from Step 3, with a, b the roots of z + λ/z = Mλ. Compute u0 = ψ^{-1}(√λ) and v0 = ∫_0^{u0} ψ(t) dt; since ψ(0) = a > 1 and ∫_0^1 ψ = 1, verify v0 > u0 and ψ(v0) + λ/ψ(u0) < 2√λ. Then evaluate φ'(x) = G'(x) − (1 + λ) at the corresponding point and verify |φ'| > (1−√λ)^2 = 1/4. If this computation succeeds, it disproves the norm claim in Theorem 2 for the paper's own construction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The sharpness claim rests on the derived Herman–Mather formula (A'): φ'(x) = G'(x) − (1 + λ), where G = g + λg^{-1}. For a zero-mean φ, ∥φ∥_{C1} = sup |φ'|, so the equality ∥φλ∥_{C1} = (1−√λ)^2 requires the range of G' to lie in [2√λ, Mλ], with Mλ = 1 + λ + (1−√λ)^2. Section 3, Step 6 proves only the upper bound sup G' = Mλ; it never establishes the lower bound inf G' ≥ 2√λ. Worse, the lower bound is false for the construction given. Let ψ be the decreasing template of Step 3, with ψ(0) = a > 1, ψ(1) = b < 1, and ∫_0^1 ψ = 1. Choose u0 with ψ(u0) = √λ. Lemma 3.3 gives ∫_0^{u0} ψ ≥ u0, and since ψ > 1 on a neighborhood of 0 and is strictly decreasing, the inequality is strict: v0 := ∫_0^{u0} ψ > u0. Then ψ(v0) < ψ(u0) = √λ. For y with g'(y) = ψ(u0), the image point g(y) has normalized coordinate v0, so G'(g(y)) = g'(g(y)) + λ/g'(y) = ψ(v0) + λ/√λ < √λ + √λ = 2√λ. Consequently φ'(x) < 2√λ − (1 + λ) = −(1−√λ)^2, so |φ'(x)| > (1−√λ)^2 and ∥φλ∥_{C1} > (1−√λ)^2. This is an internal inconsistency in the proof of Theorem 2, not merely a missing detail: the very template used to establish the upper bound forces the lower bound to fail. The Gevrey refinement of Section 3.2 does not address the issue, because it preserves the same monotone template structure. Thus the constructed perturbation does not have the claimed C1 norm, and the sharpness conclusion of the paper is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the family of dissipative twist maps F^φ_{λ,α}(x,y)=(x+α1+λy+φ(x), α2+λy+φ(x)) with 1-periodic zero-mean C^1 perturbations φ. Theorem 1 asserts that if ∥φ∥_{C^1} < (1−√λ)^2, then the map admits a unique C^1 invariant graph that is 1-normally hyperbolic, and the proof uses a graph transform, a cone condition, and a normal-hyperbolicity computation. Theorem 2 claims that this threshold is sharp: for every rational β there exist α and a C∞ zero-mean perturbation φλ with ∥φλ∥_{C^1} = (1−√λ)^2 such that the map has a unique Lipschitz invariant graph with non-differentiable points whose restriction has rotation number β. The construction of Theorem 2 goes through the Herman–Mather formula (A), building a non-differentiable circle map g with break points and setting G(x)=g(x)+λg^{-1}(x). The paper also discusses inclusions among sets of perturbations admitting C^1 graphs, normally hyperbolic graphs, and persistent graphs.","tokens_in":21142,"tokens_out":8485,"duration_ms":76695,"significance":"If Theorem 2 were established, the paper would give a sharp quantitative threshold for the persistence of C^1 invariant graphs in dissipative twist maps, complementing the qualitative normally hyperbolic invariant manifold theorem and extending the conservative Herman–Mather constructions to the conformally symplectic setting. The proof of Theorem 1 is mostly standard and appears internally sound. However, the construction underlying Theorem 2 contains a load-bearing error: the equality ∥φλ∥_{C^1} = (1−√λ)^2 requires both an upper and a lower bound on G′, but the paper proves only the upper bound, and the lower bound is false for the proposed template. The main sharpness result is therefore not established.","major_comments":[{"comment":"Proposition 3.1 proves only sup_x G′(x) = Mλ (Eq. (13)). The subsequent construction defines φ through φ′ = G′ − (1+λ) with zero mean, so ∥φ∥_{C^1} = sup_x |φ′(x)| = sup_x |G′(x) − (1+λ)|. To obtain ∥φλ∥_{C^1} = (1−√λ)^2, one needs G′(x) ∈ [2√λ, Mλ] for all x, since 2√λ − (1+λ) = −(1−√λ)^2 and Mλ − (1+λ) = (1−√λ)^2. The proof never establishes inf G′ ≥ 2√λ, and the stated sup bound alone is insufficient; it only controls positive excursions of φ′. Theorem 2's equality claim therefore does not follow from the proposition as proved.","section":"Section 3, Proposition 3.1 and Step 6"},{"comment":"The lower bound inf G′ ≥ 2√λ fails for the constructed template. Let ψ be the strictly decreasing function of Step 3 with ψ(0)=a>1, ψ(1)=b<1, and ∫_0^1 ψ = 1. Choose u0∈(0,1) with ψ(u0)=√λ. Lemma 3.3 gives v0 := ∫_0^{u0} ψ ≥ u0, and the inequality is strict for u0∈(0,1). Since ψ is strictly decreasing, ψ(v0) < ψ(u0) = √λ. For y with g′(y)=ψ(u0), the normalized coordinate of g(y) is v0, so G′(g(y)) = ψ(v0) + λ/ψ(u0) < √λ + λ/√λ = 2√λ. Hence φ′(g(y)) < 2√λ − (1+λ) = −(1−√λ)^2, and therefore ∥φ∥_{C^1} > (1−√λ)^2. The construction used to attain the sup bound necessarily violates the needed lower bound, so the claimed sharpness is contradicted by the paper's own template.","section":"Section 3, Steps 3–6"},{"comment":"The Gevrey-s refinement preserves the strictly decreasing template structure and the Step 6 argument unchanged; it introduces no lower bound on G′. Consequently the failure described in the previous comment persists for the Gevrey (and hence C∞) version of the construction. The paper contains no alternative argument that would repair the missing inf G′ ≥ 2√λ, so the error is not a local omission but a structural obstruction in the proof of Theorem 2.","section":"Section 3.2"}],"minor_comments":[{"comment":"The piecewise definition of p* contains a duplicated interval '1/8 ≤ x ≤ 1/4' and the line '1 + 8(1−b′)(x−3/4)' is assigned to 1/8 ≤ x ≤ 1/4 instead of the intended interval, making the formula difficult to parse.","section":"Section 4.1, Proposition 4.1"},{"comment":"The text attributes the dissipative Herman–Mather formula (A) to [SW26], which shares an author with the present paper, but then derives the formula inline from the invariance equation; the precise contribution imported from [SW26] should be stated more clearly to avoid ambiguity about the novelty.","section":"Section 1.2"},{"comment":"The proof that Φ(H)=(0,1) uses the connectedness of H and the facts that Φ cannot attain 0 or 1, but the latter assertion is not justified; a short argument from φ(0)=0 and φ(1)=1 would make the proof complete.","section":"Section 3, Lemma 3.2"},{"comment":"The constants K1, K2, K3 are introduced, but K1 is never used after Lemma 2.1; the presentation would be clearer if the constants were collected in one place or if K1 were referenced explicitly in the comparison preceding the contraction argument.","section":"Section 2, Lemmas 2.1–2.3"}],"recommendation":"reject","confidential_remarks":"The paper's central sharpness claim is unsupported and in fact contradicted by the construction in Section 3; the error is at the core of Theorem 2 and cannot be fixed by a minor adjustment within the current framework, since the monotone template forces the lower bound to fail. The quantitative persistence statement in Theorem 1 and the normal-hyperbolicity criteria in Section 2 may be of independent value, but the manuscript as it stands does not establish the advertised sharp threshold. I recommend rejection, though a substantially revised version that constructs a perturbation with both bounds on G′ could be reconsidered."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper asks a good question—what is the exact C^1-norm threshold for persistence of invariant graphs in dissipative twist maps—and Theorem 1 gives a clean quantitative persistence statement with a standard graph-transform proof. The connection to the Herman–Mather formula is apt and the construction strategy for building non-differentiable graphs with prescribed perturbation size is creative. If Theorem 2 were correct, it would be a sharp, subfield-relevant result.\n\nBut the central claim is not supported. Proposition 3.1 constructs G(x)=g(x)+λg^{-1}(x) and proves only sup G' = M_λ. The sharpness claim requires sup|φ'| = (1−√λ)^2, which via the derived Herman–Mather formula needs G'(x) ∈ [2√λ, M_λ] everywhere. The proof never establishes the lower bound. Worse, the stress-test is right that the lower bound is false for the given template. Pick u0 with ψ(u0)=√λ. Lemma 3.3 gives v0=∫_0^{u0}ψ > u0, so ψ(v0) < ψ(u0)=√λ, and at the image point G' = ψ(v0)+λ/ψ(u0) < 2√λ. Thus φ' = G' − (1+λ) < −(1−√λ)^2, so ∥φ∥_{C^1} > (1−√λ)^2. The constructed φ does not have the claimed norm, and the sharpness conclusion collapses.\n\nThere are also smaller issues. In Lemma 2.1, K1 should be 2/√λ −1, not 2√λ −1; as written, K1 is negative for λ<1/4 and the claim K2 < min{K1,K3} fails. The Lipschitz estimate in Lemma 2.2 also seems numerically off for λ=0.25, where the displayed ratio exceeds K2. These are likely fixable, but they add to the impression that the details have not been fully checked.\n\nThe paper is not acceptable in its current form. The flaw in Theorem 2 is load-bearing, not cosmetic. Still, the question is worth asking and Theorem 1 may survive repair, so the paper deserves a serious referee—if the authors can fix the construction, the result would be worth publishing. For now, I would not cite it.\n\nRecommendation: send to peer review, but expect major revision or rejection unless the construction is corrected.","headline":"The sharpness claim for the C^1 threshold is not established: the construction in Theorem 2 fails to control inf G', and the prescribed template actually forces G' below 2√λ, making the perturbation's C^1 norm larger than claimed.","tokens_in":21635,"tokens_out":7770,"would_cite":false,"duration_ms":71397,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37J40","37E40"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a family of dissipative twist maps, the sharp $C^1$ perturbation threshold for a smooth invariant graph is exactly $(1-\\sqrt\\lambda)^2$.","keywords":["Dissipative twist maps","Normally hyperbolic invariant manifolds","Invariant graphs","Herman-Mather formula","Sharp threshold","Circle maps","Cone condition"],"falsifier":"Compute the infimum of $G'(x)$ for the template constructed in Section 3 at $\\lambda=0.25$. If $\\inf G'<1$, then $|\\varphi'|=|G'-(1+\\lambda)|>0.25=(1-\\sqrt{0.25})^2$ somewhere, and the constructed $\\varphi$ violates the norm condition of Theorem 2; this is a concrete check of the missing lower-bound estimate.","tokens_in":20410,"feed_emoji":"🌀","tokens_out":10264,"duration_ms":105600,"temperature":0.7,"pith_summary":"The paper studies the family of dissipative twist maps $F^\\varphi_{\\lambda,\\alpha}(x,y)=(x+\\alpha_1+\\lambda y+\\varphi(x),\\alpha_2+\\lambda y+\\varphi(x))$ and asks how large the $C^1$ norm of the perturbation $\\varphi$ may be before the unique invariant graph loses smoothness. Its answer is that the critical size is exactly $(1-\\sqrt\\lambda)^2$: whenever $\\|\\varphi\\|_{C^1}<(1-\\sqrt\\lambda)^2$, the map has a unique $C^1$ invariant graph that is $1$-normally hyperbolic, while at equality there exist $C^\\infty$ perturbations with $\\|\\varphi\\|_{C^1}=(1-\\sqrt\\lambda)^2$ whose unique invariant graph is Lipschitz but has non-differentiable points. This matters because the normally hyperbolic invariant manifold theorem gives persistence only for sufficiently small perturbations; the paper turns that qualitative smallness condition into a sharp, computable threshold for this model.","feed_headline":"Smooth invariant curves break at exactly (1−√λ)²","feed_subtitle":"Below the threshold the graph stays C¹ and normally hyperbolic; at it, C∞ perturbations create non-differentiable points.","key_machinery":"The central object is the Herman–Mather formula $g(x)+\\lambda g^{-1}(x)=(1+\\lambda)x+(1-\\lambda)\\alpha_1+\\lambda\\alpha_2+\\varphi(x)$, which expresses the invariant graph through the circle map $g$ induced on the graph, and its differentiated form $g'(x)+\\lambda/g'(g^{-1}(x))=1+\\lambda+\\varphi'(x)$. The proof of Theorem 2 constructs $g$ by prescribing $g'$ on the intervals of a partition coming from a rational rotation: a smooth decreasing template $\\psi$ with $\\psi(0)=a$, $\\psi(1)=b$, $\\int_0^1\\psi=1$, where $a>1>b>0$ are the two roots of $z+\\lambda/z=1+\\lambda+(1-\\sqrt\\lambda)^2$. The identity $G'=g'(g(y))+\\lambda/g'(y)$ then makes $G$ smooth and gives $\\sup G'=1+\\lambda+(1-\\sqrt\\lambda)^2$, yielding the equality case.","core_discovery":"On the paper's own terms, the discovery is that the classical persistence statement is quantitatively optimal: the threshold $(1-\\sqrt\\lambda)^2$ separates smooth persistence from a genuine loss of differentiability. The positive direction (Theorem 1) shows that the graph transform is a contraction on Lipschitz graphs once $\\|\\varphi\\|_{\\mathrm{Lip}}\\le(1-\\sqrt\\lambda)^2$, and that strict inequality upgrades the unique invariant graph to $C^1$ and $1$-normal hyperbolicity via cone conditions and the derived Herman–Mather formula. The negative direction (Theorem 2) constructs, for every rational rotation number, a $C^\\infty$ perturbation $\\varphi_\\lambda$ whose $C^1$ norm equals the threshold and for which the unique invariant graph has non-differentiable points, so the strict inequality in Theorem 1 cannot be relaxed. The construction is built from a circle map $g$ with prescribed derivative profile and the identity $G(x)=g(x)+\\lambda g^{-1}(x)$.","pith_inferences":["A direct numerical check of the Section 3 construction would test the missing half of the norm estimate: for $\\lambda=0.25$, if $\\inf G'$ comes out below $2\\sqrt\\lambda=1$, then $\\|\\varphi\\|_{C^1}$ exceeds $(1-\\sqrt\\lambda)^2$, so the proof would need an additional lower-bound argument to keep the claimed equality.","The ratio-symmetric function $t+\\lambda/t$ suggests that analogous sharp thresholds for conformally symplectic twist maps in higher dimensions should be controlled by the normal contraction rate $\\lambda$; a natural next step is to look for the same equality/non-differentiability transition in the dissipative standard map.","One could try replacing the rational rotation in the construction by an irrational rotation; the same derivative-profile mechanism might produce critical perturbations with prescribed irrational frequency, linking this threshold phenomenon to known non-differentiable invariant curves in conservative twist maps."],"forward_implications":["The threshold $(1-\\sqrt\\lambda)^2$ cannot be enlarged: at equality there are $C^\\infty$ perturbations whose unique invariant graph has non-differentiable points, so no general persistence theorem can guarantee a $C^1$ graph beyond this size.","For every rational rotation number $\\beta$, the critical perturbation can be chosen so that the restriction of the map to the invariant graph is a circle map of frequency $\\beta$, so the breakdown is compatible with any rational frequency.","When $\\|\\varphi\\|_{C^1}<(1-\\sqrt\\lambda)^2$, the invariant graph is not only $C^1$ but $1$-normally hyperbolic, hence it persists as a normally hyperbolic invariant manifold under further perturbations.","The examples in Section 4 show that $C^1$ invariant graphs can exist without being normally hyperbolic, with perturbations whose $C^1$ norm can be made small as $\\lambda\\to1^-$.","At the threshold, Birkhoff's theorem's Lipschitz regularity is the best possible: the graph is unique and Lipschitz but not differentiable everywhere."],"supporting_citations":[{"why":"supplies the NHIM theorem, fiber-contraction estimates, and the $C^r$ section theorem used in Theorem 1.","marker":"[HPS77]"},{"why":"gives the dissipative version of the Herman–Mather formula used throughout the proofs.","marker":"[SW26]"},{"why":"provides the circle-map rotation number and conjugacy facts used to construct $g$.","marker":"[Her79]"},{"why":"supplies the area-preserving invariant-curve construction that the authors adapt to the dissipative setting.","marker":"[Her83]"},{"why":"contributes the non-differentiable invariant-curve construction idea.","marker":"[Arn11]"},{"why":"provides the geometrical proof of persistence via cone conditions used for the $C^1$ upgrade.","marker":"[BB13]"},{"why":"gives Birkhoff's graph theorem guaranteeing the Lipschitz invariant graph and its uniqueness.","marker":"[Bir20]"}],"fun_headline_variants":["Sharp threshold: below (1−√λ)² invariant curves stay smooth","At exactly (1−√λ)², invariant graphs lose differentiability","Tight C^1 norm bound for persistence of invariant graphs","Critical perturbation size: smoothness breaks exactly at (1−√λ)²","Sharp threshold for invariant manifolds: below it C^1, at it broken"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the auxiliary function $G(x)=g(x)+\\lambda g^{-1}(x)$ obeys $\\inf G'\\ge 2\\sqrt\\lambda$ as well as the proved upper bound $\\sup G'=1+\\lambda+(1-\\sqrt\\lambda)^2$; both bounds together put $\\varphi'=G'-(1+\\lambda)$ inside the announced $C^1$ ball, and the paper supplies only the upper one.","fun_headline_variants_meta":{"raw":{"variants":["Sharp threshold: below (1−√λ)² invariant curves stay smooth","At exactly (1−√λ)², invariant graphs lose differentiability","Tight C^1 norm bound for persistence of invariant graphs","Critical perturbation size: smoothness breaks exactly at (1−√λ)²","Sharp threshold for invariant manifolds: below it C^1, at it broken"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000411,"raw_usage":{"total_tokens":2122,"prompt_tokens":933,"completion_tokens":1189,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":549,"completion_tokens_details":{"reasoning_tokens":1089}},"tokens_in":549,"tokens_out":1189,"duration_ms":10802,"temperature":1.0,"reasoning_tokens":1089,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:41:07.574916+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the infimum of $G'(x)$ for the template constructed in Section 3 at $\\lambda=0.25$. If $\\inf G'<1$, then $|\\varphi'|=|G'-(1+\\lambda)|>0.25=(1-\\sqrt{0.25})^2$ somewhere, and the constructed $\\varphi$ violates the norm condition of Theorem 2; this is a concrete check of the missing lower-bound estimate.","supporting_citations":[],"review_version":1}