{"id":"df0d1787-9282-4cbe-a0f7-66d387240ba6","arxiv_id":"2412.20658","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For contact Hamiltonian systems satisfying a monotonicity condition, the omega-limit set of every positive globally minimizing orbit is contained in the Mane set of semi-static orbits.","lead":"This paper proves that in a class of contact Hamiltonian systems, every globally minimizing orbit eventually approaches the set of semi-static orbits, extending a classical result from standard Hamiltonian dynamics. The proof works through the large-time convergence of viscosity solutions of the associated Hamilton-Jacobi equation.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unproved strict inequality in the momentum-matching step leaves Theorem 1.1's conclusion (omega-limit set subset of the Mane set) unsupported.","rationale":"This is the single most load-bearing concern because the theorem's distinctive content -- omega-limit set contained in the Mane set, not merely in the graph of u_- -- depends on the equality \\bar p = \\partial L/\\partial v. The only place in the proof that enforces this equality is the chain culminating in the strict '<'. The prior step identifying \\bar u(t)=u_-(\\bar x(t)) is also essential, but the convergence of T^-_t h_{x0,u0}(\\cdot,1) to u_- is a standard type of result, and the paper's proof, though it has gaps, can likely be repaired. The strict inequality, by contrast, is asserted without any supporting argument and appears to conflict with the global minimizing property just proved for (\\bar x,\\bar u). If equality holds in the dynamic programming inequality, the proof yields no contradiction and the p-matching remains unproved. Thus the central claim is not rigorously established as written. This does not mean the theorem is false; it means a missing lemma is required. I therefore keep the reader's conditional verdict. The paper's overall strategy is sound, the pendulum example is illustrative, and the use of prior implicit-action results is not circular; the concern is specifically the unproved strict inequality in the final momentum-matching step.","tokens_in":11502,"tokens_out":15317,"duration_ms":132937,"concrete_test":"Perform a direct verification of the strict inequality in a concrete example. For the dissipative pendulum H(x,p,u)=\\frac12 p^2 -1+\\cos x + u from Example 3.6, take a point (\\bar x,\\bar p,\\bar u) with \\bar p \\ne \\partial L/\\partial v and compute the two implicit actions h_{a,b}(\\tilde x(1),2) and h_{\\bar x,h_{a,b}(\\bar x,1)}(\\tilde x(1),1) numerically along the relevant flows. If they are equal, the strict '<' fails and the contradiction in the proof is invalid. Alternatively, analytically derive the inequality from the Euler-Lagrange equations; if the derivation requires an additional hypothesis (e.g., that \\bar x(-1),\\bar x(0),\\tilde x(1) are not conjugate), state it. Absent such a derivation, the step remains an unproved assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I identify the momentum-matching step in the proof of Theorem 1.1 (Section 3.1, Step 2) as the most load-bearing unsupported point. To prove \\bar p = \\partial L/\\partial v(\\bar x,\\dot{\\bar x},\\bar u), the authors assert the strict inequality h_{\\bar x(-1),u_-(\\bar x(-1))}(\\tilde x(1),2) < h_{\\bar x, h_{\\bar x(-1),u_-(\\bar x(-1))}(\\bar x,1)}(\\tilde x(1),1), where \\tilde x is the contact flow from (\\bar x,\\partial L/\\partial v,\\bar u). No justification is given for this strict '<'. The dynamic programming principle for implicit action functions yields only a non-strict inequality h_{a,b}(y,2) \\le h_{x,h_{a,b}(x,1)}(y,1) for every intermediate x; strictness would require showing that \\bar x is not an optimal intermediate point. But the preceding argument established that (\\bar x(\\cdot),\\bar u(\\cdot)) is globally minimizing, which ordinarily forces equality at its intermediate points. If equality holds, the chain degenerates and no contradiction follows; the conclusion \\bar p = \\partial L/\\partial v is not established. Since the Mane set \\tilde N_{u_-} is defined by u_-=u_+ and p=du_\\pm, the theorem's central claim -- \\omega(\\cdot)\\subset \\tilde N_{u_-} -- rests directly on this unproved strict inequality. The convergence-theorem gaps noted by the reader are real but may be repairable from known semigroup results; this step is intrinsic to the geometric conclusion.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the asymptotic dynamics of globally minimizing orbits of a contact Hamiltonian system on T*M × R. Under assumptions (H1)-(H3), plus (A1) (strict positivity of ∂H/∂u on the zero-energy set) and (B) (existence of a viscosity solution of H(x,du,u)=0), Theorem 1.1 asserts that the ω-limit set of every positive globally minimizing orbit is contained in the Mañé set \\tilde N_{u-}, which is itself contained in the graph Λ_{u-} of the unique stationary viscosity solution u-. Theorem 1.2 asserts that from every (x0,u0) there is an initial momentum p0 whose forward orbit is a positive global minimizer, again with ω-limit in \\tilde N_{u-}. The proof proceeds by proving convergence of the semigroups T^-_t to u- (Theorem 2.1), proving boundedness of global minimizers (Lemma 3.5), and then identifying any ω-limit point as a semi-static orbit. The main technical step is the momentum-matching argument in Section 3.1.","tokens_in":11735,"tokens_out":13019,"duration_ms":124476,"significance":"If established, this is a natural extension of classical Mañé theory for Tonelli Hamiltonians to contact Hamiltonian systems under a monotonicity condition on ∂H/∂u. The strategy is genuinely different from the classical action-potential route: it uses the large-time convergence of viscosity solutions of the contact Hamilton-Jacobi equation and identifies a distinguished solution u- for every positive global minimizer. The manuscript is transparent about its assumptions and relies on prior published results (the implicit variational principle and semigroup theory) rather than on fitted or ad hoc parameters. The main geometric claim is plausible, but the proof as written contains a load-bearing gap in the momentum-matching step.","major_comments":[{"comment":"The displayed chain containing the strict inequality u_-(\\tilde x(1)) = T^-_2 u_-(\\tilde x(1)) ≤ h_{\\bar x(-1),u_-(\\bar x(-1))}(\\tilde x(1),2) < h_{\\bar x,h_{\\bar x(-1),u_-(\\bar x(-1))}(\\bar x,1)}(\\tilde x(1),1) is not justified. The implicit variational principle gives only a non-strict inequality for an arbitrary intermediate point. Strictness would require a proof that \\bar x is not an optimal intermediate point for the endpoint \\tilde x(1). The preceding argument establishes equality at the intermediate point only for the endpoint \\bar x(1), since (\\bar x(·),\\bar u(·)) is globally minimizing; it does not apply to \\tilde x(1). If equality holds, the chain degenerates and no contradiction follows. This step is load-bearing: it is the only argument that \\bar p = ∂L/∂v(\\bar x,\\dot{\\bar x},\\bar u), and hence that the ω-limit point lies in \\tilde N_{u-}. A repair might use convexity and regularity of minimizers to rule out a corner at \\bar x, but that argument is absent.","section":"Section 3.1 (Step 2, proof of Theorem 1.1)"},{"comment":"The assertion 'as {p_n(0)} is bounded' is unsupported. Since n is unbounded, Proposition 3.4, which requires t ≤ T, does not apply directly to the whole minimizer γ_n. The boundedness can presumably be obtained by restricting γ_n to [0,1] and applying Proposition 3.4 with T=1, but this argument is not given. Without a uniform bound on p_n(0), the existence of the convergent subsequence p_{n_k}(0) → p0 is not established, so Theorem 1.2 is incomplete as written.","section":"Section 3.2 (proof of Theorem 1.2)"},{"comment":"In case (3), the symbol u+ is used before it is defined in that case, and the claim that {T^+_t φ} is bounded on M × R+ is not derived. The text shows only that u+ ≤ T^+_{ntc+s}φ(x) ≤ max_{s∈[0,tc]} T^+_s φ(x); the latter upper bound requires an explicit use of T^+_{tc}φ ≤ φ and monotonicity, and u+ should be defined (e.g. as lim_{t→∞} T^+_t u-). Since Theorem 2.1 supplies the limit u- used in (3.5), this gap affects the proof of the main theorem, though it appears repairable.","section":"Section 2, proof of Theorem 2.1, case (3)"}],"minor_comments":[{"comment":"The displayed equality |u(t)| = |T^-_{t-1}h_{x0,u0}(x(t),1) - T^-_{t-1}h_{x0,u0}(x(0),1)| + |h_{x0,u0}(x(0),t)| is false; the triangle inequality gives only ≤. The subsequent argument uses the ≤ direction, so the proof is easily repaired, but the written equality is incorrect.","section":"Lemma 3.5"},{"comment":"There are numerous typographical and formatting issues with the diacritics in 'Ma˜n´e' and some inconsistent reference identifiers (e.g. [15] arXiv number 2017.11554 versus 2107.11554, and 'X. Su' in [12] versus 'X. Shu' in [16]); these should be corrected.","section":"Throughout"},{"comment":"The claimed 'if and only if' characterization of the stable manifold W^s(0,0,0) is not proved in the paper and should either be justified or explicitly stated as a heuristic illustration.","section":"Example 3.6"}],"recommendation":"major_revision","confidential_remarks":"The paper is heavily built on the authors' earlier work (refs [11,13,14]), but those are published results used as lemmas rather than hidden assumptions. The main concern is the strict inequality in Section 3.1; if the authors can supply the missing argument, the paper is publishable. The example in Section 3.6 is illustrative but not a substitute for a proof of the claimed if-and-only-if statement."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper proves a contact analogue of Mané's omega-limit theorem: under strict monotonicity (A1) and existence of a stationary solution, every positive globally minimizing orbit has omega-limit contained in the Mané set, classified by the unique stationary solution u-. That is a real extension, and the method differs from the classical action-potential route by working through large-time convergence of the Lax-Oleinik semigroup. Theorem 1.2, giving a globally minimizing lift from any (x0,u0), and the pendulum example are useful additions. Section 2's convergence theorem is largely assembled from prior semigroup results, but the proof here is self-contained enough and mostly sound.\n\nNow the soft spots, in order.\n\nLemma 3.5 contains a false equality: |A - B + B| is not equal to |A-B| + |B|. It is clearly meant as ≤, and the subsequent estimate uses ≤, so this is a typo, not a real gap.\n\nThe load-bearing issue is in Step 2 of Theorem 1.1. To prove p̄ = ∂L/∂v, the authors assert a strict inequality in the chain\n\nu-(x̃(1)) ≤ h_{...}(x̃(1),2) < h_{x̄,h_{...}(x̄,1)}(x̃(1),1) = ...\n\nNo justification is given. Dynamic programming for implicit action gives only ≤ for an intermediate point, and the fact that (x̄,ū) is globally minimizing would normally force equality at optimal intermediates. If equality holds, the chain is a string of equalities and no contradiction follows. That is a genuine gap in the written proof of the main theorem.\n\nThat said, the paper contains a likely bypass: it cites [6, Prop. 3.1], which says a globally minimizing curve automatically carries momentum p = ∂L/∂v. Since (x̄,ū) was already shown globally minimizing in (3.4), that proposition may supply the missing conclusion directly. The authors do not invoke it at the critical moment. A referee should ask them to justify the strict inequality or replace the step with the cited result.\n\nTheorem 2.1, case (3), has a compressed boundedness argument that works after unpacking the semigroup inequalities, but it needs rewording. No fitted parameters, no data, and the heavy self-citation is to published prior results used as lemmas; that is not circular.\n\nVerdict: the result is plausible and probably true, the strategy is right, and the gaps are repairable. It deserves a serious referee; I would send it out with a request for revision rather than desk-reject.","headline":"A plausible and worthwhile contact analogue of Mané's omega-limit theorem, with one unproved strict inequality in the key geometric step; the repair is likely in the paper's own cited tools.","tokens_in":12321,"tokens_out":6940,"would_cite":false,"duration_ms":66244,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37J55","35F21","35D40"],"pacs":[],"model":"deepseek-v4-flash","headline":"Under strict monotonicity of the contact Hamiltonian, every bounded globally minimizing orbit is asymptotic to a semi-static orbit on the graph of a stationary viscosity solution.","keywords":["contact Hamiltonian systems","globally minimizing orbits","Mañé set","semi-static orbits","viscosity solutions","Hamilton–Jacobi equations","ω-limit set","large-time behavior"],"falsifier":"On the dissipative pendulum $H=\\frac12 p^2-1+\\cos x+u$, Theorem 1.2 predicts that for each $(x_0,u_0)$ the selected orbit converges to $(0,0,0)$ only for initial points on the stable manifold of that fixed point. A direct check: for a grid of $(x_0,u_0)$, approximate $p_0$ by the limit of momenta of minimizers of $h_{x_0,u_0}(x,n)$ as $n\\to\\infty$, and test whether $(x_0,p_0,u_0)$ lies on the stable manifold; a single initial condition with $u_0\\neq0$ whose selected $p_0$ misses the manifold would falsify the selection claim. More generally, any bounded globally minimizing orbit whose $\\omega$-limit set contains a point with $u$-coordinate different from $u_-(x)$ for every stationary solution $u_-$ would falsify Theorem 1.1.","tokens_in":11215,"feed_emoji":"🌀","tokens_out":14833,"duration_ms":121657,"temperature":0.7,"pith_summary":"The paper establishes that the long-time behavior of globally minimizing orbits in contact Hamiltonian systems is controlled by the stationary viscosity solutions of the associated Hamilton–Jacobi equation. Under assumption (A1), that $\\partial H/\\partial u>0$ on the zero level set $\\{H=0\\}$, and assumption (B), the existence of a stationary solution, Theorem 1.1 shows that the $\\omega$-limit set of every positive globally minimizing orbit is contained in the Mañé set $\\widetilde{N}_{u_-}$ of semi-static orbits, which is itself contained in the graph $\\Lambda_{u_-}$ of the unique stationary solution $u_-$. This transfers a classical Lagrangian-dynamics result—$\\omega$-limit sets of globally minimizing orbits lie in static sets—to the contact setting. The proof connects the implicit action functions defining minimizers with the large-time limit of viscosity solutions of the evolution Hamilton–Jacobi equation. If correct, the asymptotic dynamics is organized by a single function $u_-$, and from every initial point $(x_0,u_0)$ a momentum can be selected so that the forward orbit is globally minimizing and converges toward that graph.","feed_headline":"Bounded contact-Hamiltonian minimizers end on semi-static orbits","feed_subtitle":"ω-limit sets of all minimizers lie on the graph of a stationary viscosity solution, extending classical Lagrangian dynamics.","key_machinery":"The load-bearing mechanism is the pair of evolution semigroups $T^-_t$ and $T^+_t$ for the contact Hamilton–Jacobi equation, together with the implicit action functions $h_{x_0,u_0}(x,t)$ that define globally minimizing curves. The identity $h_{x_0,u_0}(x,t+1)=T^-_t h_{x_0,u_0}(x,1)$ is what lets the authors pass from large-time convergence of viscosity solutions—Theorem 2.1, which gives $T^-_t\\varphi\\to u_-$ uniformly for every continuous $\\varphi$ under (A1) and (B)—to statements about minimizers. Uniform Lipschitz and compactness estimates on $\\{T^-_t h_{x_0,u_0}(\\cdot,1)\\}$ justify passing to limits, and the comparison principles for the semigroups supply the inequalities that force the limiting orbit to be semi-static and its momentum to match the graph.","core_discovery":"Theorem 1.1 is the central claim: for each positive globally minimizing orbit $(x(t),p(t),u(t))$ of the contact Hamiltonian flow, under assumptions (A1) and (B) there exists a viscosity solution $u_-\\in S^-$ of $H(x,\\partial_x u,u)=0$ such that $\\omega(x(0),p(0),u(0))\\subset \\widetilde{N}_{u_-}\\subset\\Lambda_{u_-}$, where $\\widetilde{N}_{u_-}$ is the Mañé set of semi-static orbits labeled by $u_-$ and $\\Lambda_{u_-}=\\mathrm{cl}\\{(x,\\mathrm{d}u_-(x),u_-(x))\\}$ is the graph of the stationary solution. Lemma 3.5 proves every positive globally minimizing orbit is bounded, so the $\\omega$-limit set is nonempty. The authors then show that any $\\omega$-limit point $(\\bar{x},\\bar{p},\\bar{u})$ generates a globally minimizing curve whose $u$-component obeys $\\bar{u}(t)=u_-(\\bar{x}(t))$ for all $t$, and that this curve is semi-static. A strict inequality argument forces the momentum component to agree with $\\partial L/\\partial v(\\bar{x},\\dot{\\bar{x}},\\bar{u})$, placing the point on $\\Lambda_{u_-}$. Theorem 1.2 adds that for every $(x_0,u_0)\\in M\\times\\mathbb{R}$ one can choose an initial momentum $p_0$ such that the forward orbit is globally minimizing and satisfies the same $\\omega$-limit conclusion.","pith_inferences":["The proof suggests a general selection rule: the momentum $p_0$ produced by Theorem 1.2 is the point where the stable manifold of the semi-static set meets the fiber $T^*_{x_0}M$; this is visible in the pendulum example and could be checked numerically for other contact Hamiltonians with a unique stationary solution.","The convergence argument only uses the semigroup limit on the specific initial data $h_{x_0,u_0}(\\cdot,1)$, so the conclusion may survive without global strict monotonicity (A1), provided the relevant semigroup still converges; a localized version of Theorem 2.1 would settle this.","The identity $h_{x_0,u_0}(x,t+1)=T^-_t h_{x_0,u_0}(x,1)$ also offers a computational route to approximating semi-static orbits: evolve the implicit action data under the semigroup until stabilization, then read off the graph $u_-$ and the selected orbit."],"forward_implications":["Every positive globally minimizing orbit is bounded, and its $\\omega$-limit set lies in $\\widetilde{N}_{u_-}\\subset\\Lambda_{u_-}$; under (A1)+(B) the long-time dynamics is organized by the single stationary solution $u_-$.","From each point $(x_0,u_0)\\in M\\times\\mathbb{R}$ one can select an initial momentum $p_0$ so that the forward orbit is globally minimizing and its $\\omega$-limit is contained in the Mañé set of semi-static orbits (Theorem 1.2).","If strict monotonicity (A1) is relaxed to nonnegativity (A2), the same dynamical conclusion holds, except that $u_-$ is no longer unique and the limiting stationary solution depends on the initial data.","Because $S^-$ has a single element under (A1)+(B), the Mañé set decomposition collapses to one component, $\\widetilde{N}=\\widetilde{N}_{u_-}$.","For the dissipative pendulum $H=\\frac12 p^2-1+\\cos x+u$, the selected orbit has $\\omega$-limit $\\{(0,0,0)\\}$ precisely when its initial point lies on the stable manifold of that hyperbolic fixed point."],"supporting_citations":[{"why":"The classical result that ω-limit sets of globally minimizing orbits lie in static curves, which the paper extends to contact systems.","marker":"[1]"},{"why":"The companion Lagrangian result and Mañé action-potential method that form the baseline for conclusion (⋆).","marker":"[2]"},{"why":"Defines the Mañé set, semi-static orbits, and the minimizers of implicit actions used in Theorem 1.2.","marker":"[6]"},{"why":"Gives the sufficient condition for assumption (B), the existence of a stationary viscosity solution.","marker":"[10]"},{"why":"Introduces the semigroups T^-_t and T^+_t, the evolution equation's well-posedness, and the compactness lemma bounding minimizers.","marker":"[11]"},{"why":"Provides the convergence result without strict monotonicity, used as Proposition 2.2 for the (A2) version.","marker":"[12]"},{"why":"Defines the implicit action functions and the variational notion of globally minimizing curves the proof is built on.","marker":"[13]"},{"why":"Supplies the decomposition of the Mañé set into components labeled by stationary solutions, used in the final inclusion.","marker":"[14]"},{"why":"Comparison principle for viscosity solutions invoked in Proposition A.5 and used in the proof of Theorem 2.1.","marker":"[17]"},{"why":"The comparison principle for Hamiltonians depending on the unknown function, also used through Proposition A.5.","marker":"[18]"}],"fun_headline_variants":["Contact minimizers' ω-limits sit on semi-static orbits","Globally minimizing contact orbits end on semi-static sets","Contact minimizers converge to semi-static orbits in limit","Bounded minimizers in contact flows end on semi-static orbits","Semi-static orbits attract all contact minimizers"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on assumption (A1), that $\\partial H/\\partial u$ is strictly positive on the zero level set $\\{H=0\\}$, because this monotonicity is what makes the stationary solution unique and makes every viscosity solution of the evolution equation converge to it uniformly; without it, the identification of $\\omega$-limit points with $u_-(x)$ can fail.","fun_headline_variants_meta":{"raw":{"variants":["Contact minimizers' ω-limits sit on semi-static orbits","Globally minimizing contact orbits end on semi-static sets","Contact minimizers converge to semi-static orbits in limit","Bounded minimizers in contact flows end on semi-static orbits","Semi-static orbits attract all contact minimizers"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000775,"raw_usage":{"total_tokens":3401,"prompt_tokens":892,"completion_tokens":2509,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":508,"completion_tokens_details":{"reasoning_tokens":2429}},"tokens_in":508,"tokens_out":2509,"duration_ms":17639,"temperature":1.0,"reasoning_tokens":2429,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T23:15:20.061641+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On the dissipative pendulum $H=\\frac12 p^2-1+\\cos x+u$, Theorem 1.2 predicts that for each $(x_0,u_0)$ the selected orbit converges to $(0,0,0)$ only for initial points on the stable manifold of that fixed point. A direct check: for a grid of $(x_0,u_0)$, approximate $p_0$ by the limit of momenta of minimizers of $h_{x_0,u_0}(x,n)$ as $n\\to\\infty$, and test whether $(x_0,p_0,u_0)$ lies on the stable manifold; a single initial condition with $u_0\\neq0$ whose selected $p_0$ misses the manifold would falsify the selection claim. More generally, any bounded globally minimizing orbit whose $\\omega$-limit set contains a point with $u$-coordinate different from $u_-(x)$ for every stationary solution $u_-$ would falsify Theorem 1.1.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The classical result that ω-limit sets of globally minimizing orbits lie in static curves, which the paper extends to contact systems."},{"cited_title":"Contreras, J","cited_arxiv_id":null,"evidence_quote":"The companion Lagrangian result and Mañé action-potential method that form the baseline for conclusion (⋆)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Mañé set, semi-static orbits, and the minimizers of implicit actions used in Theorem 1.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the semigroups T^-_t and T^+_t, the evolution equation's well-posedness, and the compactness lemma bounding minimizers."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the convergence result without strict monotonicity, used as Proposition 2.2 for the (A2) version."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the implicit action functions and the variational notion of globally minimizing curves the proof is built on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the decomposition of the Mañé set into components labeled by stationary solutions, used in the final inclusion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Comparison principle for viscosity solutions invoked in Proposition A.5 and used in the proof of Theorem 2.1."},{"cited_title":"Ishii, K","cited_arxiv_id":null,"evidence_quote":"The comparison principle for Hamiltonians depending on the unknown function, also used through Proposition A.5."}],"review_version":1}