{"id":"8bda2cb8-bed5-4951-a93d-06a1e244cceb","arxiv_id":"1908.05675","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For volume-preserving almost Anosov flows with a cubic neutral saddle, the paper derives Dulac map asymptotics and obtains Gaussian and stable limit laws for a class of observables.","lead":"This paper proves statistical limit laws, such as the central limit theorem, for a class of volume-preserving flows that are hyperbolic everywhere except one neutral periodic orbit. It works by deriving precise estimates for the time orbits spend near this orbit, a case earlier results could not handle.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The advertised limit laws in Corollary 1.1 are not established in this paper: the proof hands off to Theorem 2.7 of the unpublished companion [3], and the present text supplies only tail estimates, not the renewal-theoretic passage from tails to CLT/stable laws.","rationale":"I read the paper's main technical contribution as Theorem 1.1: precise regular variation of the Dulac map for cubic neutral saddles with mixed quadratic terms. The proof of Theorem 1.1 is lengthy and appears to contain real work, and the supporting estimates in Propositions 2.1 and 4.1 are consistent with the stated conclusions. The volume-preserving special case beta_2 = 2 is also coherent: Proposition 4.2 gives a return-time tail of order t^{-2}, which is the boundary case for the renewal theorem. I therefore do not think the Dulac analysis itself is the weak point. The weak point is the final step: the abstract promises limit laws, but the proof of Corollary 1.1 contains no renewal theorem. It relies entirely on 'a direct application of Theorem 2.7 in [3]', an unpublished preprint of the same group. The present paper verifies, at most, single-step tail behaviour of tau and bar v. A tail exponent alone does not distinguish a Gaussian CLT from a non-Gaussian one, and it does not produce the covariance or the non-coboundary condition. The paper explicitly says that the condition in [3] involving psi_0 is 'only important for the results on the shape of the pressure function' and that 'For us, only the tail of bar v matters'. That assertion is not proved here; it is a claim about the robustness of the renewal theorem in [3]. Consequently, the central statistical conclusions are conditional on an external result that the reader cannot check from this manuscript. This is not a matter of novelty or consensus, and I am not accusing the authors of any impropriety; it is simply that the argument is incomplete as written. The reader's weakest_assumption identified exactly this dependency, and I agree. A conditional verdict is appropriate, with the condition being that Theorem 2.7 of [3] either be made available and verified against the hypotheses in this paper, or that a self-contained proof of Corollary 1.1 be supplied.","tokens_in":18349,"tokens_out":17527,"duration_ms":179972,"concrete_test":"Obtain the full statement and proof of Theorem 2.7 of [3], or require a rewritten proof of Corollary 1.1 that is self-contained, and verify each hypothesis for the induced map F: the return-time tail Leb({tau > t}) as in (40), the regular variation of Leb({bar v > t}) with index -4/(2-rho), aperiodicity of F and of the renewal measure, and the condition that bar v is not a coboundary in the Banach space B of [3]. The concern lands if any hypothesis fails or is not verified; in that case the limit-law conclusions of Corollary 1.1 are unsupported without further argument.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central promise is Corollary 1.1, which asserts non-standard CLT, standard CLT, and stable laws for volume-preserving almost Anosov flows. The proof in Section 5 is a single paragraph saying that Corollary 1.1 is a 'direct application' of Theorem 2.7 in [3], an unpublished preprint. This is the exact step that converts the tail estimates into distributional limit theorems. What the present paper actually proves is: (i) the Dulac map asymptotics (Theorem 1.1); (ii) the tail of the return time, Leb({tau > t}) ~ C* t^{-beta_2} (Proposition 4.2); and (iii) a tail estimate for the induced observable, Leb({bar v > t}) ~ C t^{-4/(2-rho)} (Section 5, using Proposition 4.1). None of these alone yields a CLT or a stable law: the ergodic sum of bar v over many renewals requires a renewal theorem for the induced Poincare map F, with the precise conditions and conclusions supplied by [3]. The paper does not verify those conditions, and it does not reproduce the renewal argument. If Theorem 2.7 of [3] requires hypotheses that are not checked here—aperiodicity of the renewal distribution, a specific renewal mass function, regularity of the transfer operator on a suitable Banach space, or non-coboundary of bar v in that space—then Corollary 1.1 does not follow from the material in this manuscript. The concern is not that [3] is outside the consensus; it is that the central claim depends on an inaccessible external result at its most load-bearing juncture.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies volume-preserving almost Anosov flows with a neutral periodic orbit of cubic saddle type, focusing on the two-dimensional horizontal flow (3). Its main technical result, Theorem 1.1, gives Dulac-time and Dulac-map asymptotics for vector fields with mixed quadratic terms under the first-integral condition (4). The paper then claims, in Corollary 1.1, a non-standard CLT, a Gaussian CLT, and stable laws for observables of the volume-preserving flow, based on the tail estimates and an external renewal theorem.","tokens_in":18737,"tokens_out":6541,"duration_ms":60382,"significance":"If the advertised limit laws are correct, the paper addresses a natural gap in the literature: previous results excluded mixed terms a1=b1=0, and the manuscript explains why those terms cannot be removed by coordinate changes. The proof of Theorem 1.1 is detailed and does not fit parameters to conclusions; the constants are constructed in the course of the proof, and the flow-box volume computation at the end of Section 3 provides a useful independent check. The main weakness is that Corollary 1.1, which is the statistical payoff promised in the abstract, is not proved in the manuscript but delegated to the unpublished companion paper [3].","major_comments":[{"comment":"Corollary 1.1 is the announced limit-law result, but its proof consists of a single paragraph invoking \"a direct application of Theorem 2.7 in [3]\". That theorem is not reproduced, and none of its hypotheses are verified for the induced Poincaré map F: aperiodicity of the renewal distribution, the renewal mass function, the Banach space B, transfer-operator regularity, or the non-coboundary condition. The material proved here (Theorem 1.1, Proposition 4.2, and the tail estimate for bar v) does not by itself imply a CLT or a stable law; some renewal theorem is needed. As the manuscript stands, the central claim of the abstract is therefore not established in this paper. I request either a self-contained proof of the renewal step for F, or a complete statement, with all hypotheses, of the external theorem and a demonstration that those hypotheses hold in the present setting.","section":"Section 5 (proof of Corollary 1.1)"},{"comment":"The corollary states for rho=0 that the variance sigma^2>0 \"unless integral of v is a coboundary\", and for rho>0 it asserts a Gaussian CLT with no exception. In the proof, however, the CLT is made conditional on the variance being positive, and the sentence \"this follows from bar v not being a coboundary, and this we assumed explicitly\" introduces a non-coboundary assumption that is not a hypothesis of the corollary. The statement should either add this assumption or prove that it holds for the class of observables v described.","section":"Corollary 1.1 and its proof"}],"minor_comments":[{"comment":"There are small typographical issues: \"3-three manifolds\" should be \"three-manifolds\", and the displayed condition \"c2_1 < 4c0c2\" should be \"c1^2 < 4c0c2\".","section":"Abstract and Introduction"},{"comment":"The sentence \"Then there constants xi0(eta), omega0(eta)\" is missing \"are\". In addition, the error term O(tilde T^{-beta*}, T^{-1/2} log T) mixes tilde T and T; it would be clearer to write O(max{tilde T^{-beta*}, tilde T^{-1/2} log tilde T}) and state that the notation is interpreted as a maximum.","section":"Theorem 1.1"},{"comment":"The proposition statement omits the hypothesis c1^2<4c0c2, although the proof uses the positivity of c0+c1 M+c2 M^2, which is guaranteed by that condition. Theorem 1.1 includes this condition, but the proposition as stated is too broad.","section":"Proposition 2.1"},{"comment":"The terms Clog delta log delta and the later conditions \"Clog is only nonzero if a2/b2=1\" are notationally awkward; the constants C1, C2, C3, C4, Clog should be defined more explicitly, and the notation for the analogous constants in the reversed-role estimate should be introduced consistently.","section":"Equation (31)"},{"comment":"The comment \"If rho >= 2, this asymptotic formula should be interpreted as Leb(bar v>t)=0 for t large\" concerns a parameter range that is outside the statement of Corollary 1.1 (which only treats rho in (-2,1)); either remove the remark or explain its role.","section":"Section 5"}],"recommendation":"major_revision","confidential_remarks":"The main issue for me is that the paper's advertised statistical result, Corollary 1.1, depends on the unpublished companion [3] at its most load-bearing juncture. I would ask the editor to require either a published or otherwise accessible version of [3] containing the renewal theorem, or an appendix in which the renewal argument is carried out and its hypotheses verified. The Dulac-map part of the paper, especially Theorem 1.1 and Section 3, appears to be a solid standalone contribution; if the limit-law corollary cannot be fully supported, the authors could publish the Dulac estimates alone and mark the limit laws as conditional."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nIf you pick this paper up for the Dulac-map asymptotics, you are on solid ground. The genuinely new thing is Theorem 1.1: exact regularly varying asymptotics for the Dulac time and map at a cubic saddle with mixed quadratic terms, under condition (4). Previous work by the same group assumed a1=b1=0; here the first integral is found explicitly, the derivation is detailed, and the volume-preserving case (5) reduces everything to a one-parameter family. The flow-box check at the end of Section 3 gives the leading coefficient the right interpretation. That part is worth citing.\n\nWhat is not on solid ground is Corollary 1.1, the abstract's \"limit laws\" promise. The proof is one paragraph: \"direct application of Theorem 2.7 in [3],\" where [3] is an unpublished preprint. The present paper supplies tail estimates for the return time τ and for the induced observable v̄, but those tails alone do not produce a CLT, a non-standard CLT, or stable laws without a renewal theorem for the induced map F. The paper never verifies the hypotheses of [3, Thm 2.7]—aperiodicity, transfer-operator regularity, non-coboundary in the right Banach space. So as written, the limit laws are conditional on an inaccessible black box. Also, the step from radial weights to general observables is asserted rather than proved: Proposition 4.1 computes ∫ r^ρ dt, and Section 5 says \"Since Proposition 4.1 applies to v,\" but v is only defined as having homogeneous integral of order ρ, not necessarily radial. That gap is smaller but real.\n\nOn the positive side, the main theorem's proof is independent and parameter-free, so I see no circularity in the Dulac estimate. The citation pattern is acceptable under the caveat that the companion result is unpublished; the self-citation becomes a problem only because the load-bearing step is inaccessible to the reader.\n\nWho is this for? People working on Dulac maps, almost Anosov flows, and tail estimates for non-uniformly hyperbolic flows. They should read Theorem 1.1 and treat Corollary 1.1 as provisional until [3] appears.\n\nRecommendation: yes, send to peer review. The technical core deserves a referee. The referee should be told to treat Corollary 1.1 as unproved as stated, and the authors should either make the renewal argument self-contained or weaken the corollary accordingly. That is a fixable situation, not a rejection.","headline":"The Dulac-map estimates with mixed terms are a genuine, carefully worked extension; the advertised limit laws are outsourced to an unpublished companion and should not be cited as proved here.","tokens_in":19209,"tokens_out":2552,"would_cite":true,"duration_ms":24665,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37C10","37D20","37D25","60F05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Dulac maps with mixed terms: limit laws for almost Anosov flows","keywords":["Dulac map","almost Anosov flows","limit laws","stable laws","Central Limit Theorem","non-uniform hyperbolicity","cubic saddle","first integral"],"falsifier":"Compute the induced Poincaré map for a concrete volume-preserving almost Anosov flow with $\\gamma\\in(-4,4)$ and measure the tail probability $\\mathrm{Leb}(\\{\\tau>t\\})$ for return times; if it does not behave like $C t^{-2}$ with a nonzero constant, the tail estimate behind Corollary 1.1 fails. Alternatively, test the stable-law prediction directly: for an observable with homogeneous order $\\rho=-1$, the normalised integrals should converge to a stable distribution of index $4/3$.","tokens_in":18143,"feed_emoji":"📈","tokens_out":4704,"duration_ms":44024,"temperature":0.7,"pith_summary":"This paper establishes precise asymptotic formulas for the Dulac map of a planar vector field with a neutral cubic saddle, and uses them to derive statistical limit laws for volume-preserving almost Anosov flows in three dimensions. The novelty is that the vector field may contain quadratic mixed terms $xy$; earlier results had assumed those terms vanish. Under one co-dimension one parameter condition, the paper finds an explicit first integral that reduces the problem to a one-dimensional equation, yielding regular variation of the Dulac map and of return-time tails. From this it claims a central limit theorem with non-standard $\\sqrt{t\\log t}$ scaling for observables that vanish at the neutral orbit, a Gaussian CLT for stronger observables, and stable laws in the intermediate range. These limit laws matter because they describe the statistical behavior of natural non-uniformly hyperbolic flows where uniform hyperbolicity breaks down along a single periodic orbit.","feed_headline":"Dulac maps with mixed terms: limit laws for almost Anosov flows","feed_subtitle":"A one-parameter condition gives CLT, non-standard scaling, and stable laws for flows with a neutral cubic saddle.","key_machinery":"The load-bearing object is an explicit first integral $L(x,y)=x^u y^v(\\frac{a_0}{v}x^2+\\frac{a_1}{v+1}xy+\\frac{b_2}{u}y^2)$, whose existence follows from condition (4). Because $L$ is constant along the flow of the unperturbed quadratic-cubic vector field, switching to the slope variable $M=y/x$ turns the planar system into the one-dimensional ODE (16). Integrating that ODE yields the Dulac time and the asymptotic expressions for $\\xi$ and $\\omega$ in Theorem 1.1; the same estimates are then shown to be stable under $O(4)$ perturbations by comparing the perturbed first integral with the unperturbed one. The return-time tail estimate $\\mathrm{Leb}(\\{\\tau>t\\}) \\sim C t^{-\\beta_2}$ follows by integrating these estimates over unstable leaves of the induced Poincaré map.","core_discovery":"The paper's central claim is Theorem 1.1: for a $C^3$ vector field of local form (3) with parameters satisfying (2) and (4), the coordinate functions of the Dulac map are regularly varying, specifically $\\xi(\\eta,T)=\\xi_0(\\eta)T^{-\\beta_2}(1+O(T^{-\\beta_*},T^{-1/2}\\log T))$ and $\\omega(\\eta,T)=\\omega_0(\\eta)T^{-\\beta_0}(1+O(T^{-\\beta_*},T^{-1/2}\\log T))$, so the Dulac map itself satisfies $\\omega=D(\\xi)=\\omega_0 \\xi_0^{-\\beta_0/\\beta_2} \\xi^{\\beta_0/\\beta_2}(1+o(1))$. The paper then states Corollary 1.1: for a volume-preserving almost Anosov flow with a neutral cubic saddle, $C^1$ observables whose integrals along the neutral orbit are homogeneous of order $\\rho$ satisfy a CLT with $\\sqrt{t\\log t}$ scaling when $\\rho=0$, a Gaussian CLT when $\\rho>0$, and stable laws of order $4/(2-\\rho)$ when $-2<\\rho<0$.","pith_inferences":["If the renewal machinery in the companion preprint [3] applies as claimed, the same method should also yield mixing rates or almost-sure invariant principles for these flows; the paper does not develop those consequences.","The parameter condition (4) is co-dimension one and follows automatically from volume preservation, but for non-volume-preserving flows the limit laws will likely depend on whether $\\beta_0$ and $\\beta_2$ fall on either side of $1$, so the boundary cases $\\beta_i=2$ could be probed for logarithmic corrections.","A concrete numerical check: simulate the induced Poincaré map for a volume-preserving almost Anosov flow with parameter $\\gamma\\in(-4,4)$ and measure the distribution of normalised integrals; for an observable with $\\rho=-1$ the histogram should converge to a stable law of index $4/3$."],"forward_implications":["For volume-preserving almost Anosov flows satisfying the volume-preserving parameter conditions, the neutral saddle produces a return-time tail with exponent $-2$, placing the induced system exactly at the threshold between finite and infinite variance.","Observables whose integral along the neutral orbit vanishes to order $\\rho=0$ have fluctuations of order $\\sqrt{t\\log t}$, not the usual $\\sqrt{t}$.","Positive-order observables satisfy the ordinary Gaussian CLT with $\\sqrt{t}$ scaling.","Intermediate observables with $-2<\\rho<0$ obey stable laws of order $4/(2-\\rho)\\in(1,2)$, so the heavy-tailed regime is fully described by the single parameter $\\rho$.","The strip-measure estimates behind Theorem 1.1 also constrain the induced return map and can support further limit theorems beyond the three cases stated in Corollary 1.1."],"supporting_citations":[{"why":"States Theorem 2.7, the renewal-theoretic result from which Corollary 1.1 is derived as a direct application.","marker":"[3]"},{"why":"Supplies the almost Anosov set-up, the Markov partition, and previous Dulac-map estimates in the absence of mixed terms, which this paper extends.","marker":"[2]"},{"why":"Introduces the Dulac map as the return map near a saddle, the object whose asymptotic behaviour is proved here.","marker":"[5]"},{"why":"Provides the regularity theorem for integral-curve foliations used when comparing the perturbed and unperturbed first integrals in Section 3.","marker":"[16]"}],"fun_headline_variants":["One neutral saddle, many limit laws","Almost Anosov flows: from CLT to stable laws","Neutral cubic saddle: one flow, many limit laws","From CLT to stable laws: almost Anosov flows with a neutral saddle","Limit laws for almost Anosov flows: the neutral cubic saddle case"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The limit-law conclusions rest on an unpublished renewal theorem from the paper's companion preprint [3]; if that theorem does not apply to an induced map whose return-time tail decays like $C t^{-2}$, the CLT and stable-law statements do not follow from this paper alone.","fun_headline_variants_meta":{"raw":{"variants":["One neutral saddle, many limit laws","Almost Anosov flows: from CLT to stable laws","Neutral cubic saddle: one flow, many limit laws","From CLT to stable laws: almost Anosov flows with a neutral saddle","Limit laws for almost Anosov flows: the neutral cubic saddle case"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001407,"raw_usage":{"total_tokens":5631,"prompt_tokens":833,"completion_tokens":4798,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":449,"completion_tokens_details":{"reasoning_tokens":4712}},"tokens_in":449,"tokens_out":4798,"duration_ms":32205,"temperature":1.0,"reasoning_tokens":4712,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:12:01.850765+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the induced Poincaré map for a concrete volume-preserving almost Anosov flow with $\\gamma\\in(-4,4)$ and measure the tail probability $\\mathrm{Leb}(\\{\\tau>t\\})$ for return times; if it does not behave like $C t^{-2}$ with a nonzero constant, the tail estimate behind Corollary 1.1 fails. Alternatively, test the stable-law prediction directly: for an observable with homogeneous order $\\rho=-1$, the normalised integrals should converge to a stable distribution of index $4/3$.","supporting_citations":[{"cited_title":"Bruin, D","cited_arxiv_id":null,"evidence_quote":"States Theorem 2.7, the renewal-theoretic result from which Corollary 1.1 is derived as a direct application."},{"cited_title":"Bruin, D","cited_arxiv_id":null,"evidence_quote":"Supplies the almost Anosov set-up, the Markov partition, and previous Dulac-map estimates in the absence of mixed terms, which this paper extends."},{"cited_title":"Dulac, Sur les cycles limites, Bull","cited_arxiv_id":null,"evidence_quote":"Introduces the Dulac map as the return map near a saddle, the object whose asymptotic behaviour is proved here."},{"cited_title":"Teschl, Ordinary Differential Equations and Dynamical Systems, Graduate Studies in Mathematics, 140, Amer","cited_arxiv_id":null,"evidence_quote":"Provides the regularity theorem for integral-curve foliations used when comparing the perturbed and unperturbed first integrals in Section 3."}],"review_version":1}