{"id":"079e7233-1185-4136-9fb4-bd2abb78e386","arxiv_id":"2501.11725","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Generic Reeb flows on 3-manifolds have homoclinic connections on every hyperbolic orbit, can be given embedded Birkhoff sections with prescribed boundary and Legendrian content, and every Legendrian knot has infinitely many Reeb chords.","lead":"This paper proves that for generic Reeb flows on closed 3-manifolds, every hyperbolic periodic orbit has transverse homoclinic connections, and that Birkhoff sections can be made embedded and forced to contain prescribed periodic orbits and Legendrian links. It also shows that any Reeb flow with a Birkhoff section has infinitely many Reeb chords for every Legendrian knot, except for a rigid two-orbit case.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"In Proposition 3.2 the prong count is computed as |b|+2|a|, but the geometric intersection number is |b−2a|; for same-sign (a1,nb0) the new section can still be 1-pronged, so elimination of 1-prongs is unproved.","rationale":"The reader identified Proposition 3.2 as the weakest assumption, and I agree that the elimination of 1-pronged points is the most load-bearing step in the proof of Theorem 1.4. My stress-test narrows this to a concrete computational issue: the number of prongs of the collapsed point is a geometric intersection number on the boundary torus, and the paper's formula n|b0|+2a1 is not that number. In the standard meridian/longitude basis used in Remark 3.1, the stable manifold of a negative hyperbolic orbit has class (1,2), so a trace of class (a,b) meets it in |b−2a| points. The paper's formula |b|+2|a| coincides only when a or b vanishes; for same-sign nonzero components it overestimates the true count and can mask a remaining 1-prong. Since the construction in Proposition 3.2 only passes from S_n to S_{n+1}, a single bad orbit with ℓ=1 for both values of n would break the argument as written. The issue is local and concrete, and it does not require questioning the broader framework: the rest of the proof, including the adaptation of Le Calvez–Sambarino, appears coherent once a section without 1-prongs is available. I therefore keep the verdict at CONDITIONAL: the paper should be accepted only if the authors supply a correct prong-counting estimate, for instance by choosing n large enough to force |nb0−2a1|≥2 for all finitely many boundary orbits, or by proving an alternative lower bound. No other concern I examined—the use of Proposition 2.11, the covering argument in Proposition 3.16, or the chord-counting in Theorem 1.10—reaches the same level of direct impact on the central theorem.","tokens_in":42349,"tokens_out":30496,"duration_ms":298757,"concrete_test":"Take a negative hyperbolic orbit γ and a model ∂-strong section with boundary trace (1,1) on ∂Mγ, with the stable manifold curve (1,2) as in Remark 3.1. Compute the transverse intersection number: |det((1,2),(1,1))| = |1·1−2·1| = 1, showing the collapsed point is 1-pronged, while the paper's formula in Proposition 3.2 would give |1|+2|1|=3≥2. Also check the specific parameters a1,b0,n that arise from the construction in Prop 3.2: if b0=1 and a1=n or a1=(n±1)/? there can be cancellations. More generally, recompute the prong counts in all three cases of the proof using ℓ=|det((1,2),(a,b))|; if any admissible class (a1,nb0) or (na0+a1,nb0) has |nb0−2a1|=1 or |nb0−2(na0+a1)|=1, the claim that S_n or S_{n+1} works fails.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section 3.1 defines ℓ(p_i) for a negative hyperbolic boundary orbit as the number of segments of S intersecting its stable (resp. unstable) manifold, which equals the geometric intersection number of the boundary trace with the curve P(E^s)⊂∂Mγ, of class (1,2) (Remark 3.1). For a trace of class (a,b) this number is |det((1,2),(a,b))| = |b−2a|. In Proposition 3.2, for γ∈L′∩∂S_n the trace of S_n is claimed to have homology (a1,nb0) with a1>0 and the number of intersections with the stable/unstable manifolds is asserted to be n|b0|+2a1. This is not the geometric intersection number; it is the ℓ^1 sum |b|+2|a|, which is generally an upper bound, not a lower bound. Taking a1=1, b0=1, n=1 gives true ℓ = |1−2| = 1, so the collapsed point is again 1-pronged, although the paper's formula yields 3. Thus the conclusion that S_n or S_{n+1} has no 1-pronged points is not established. The same error propagates in the estimates for the other cases (γ∈K, γ∈L\\L′). Since Proposition 3.2 is the only step removing 1-pronged points, and the subsequent Le Calvez–Sambarino adaptation assumes no 1-pronged points, Theorem 1.4 rests on this gap.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves three main results for C^∞-generic contact forms on closed 3-manifolds: (1) every hyperbolic periodic Reeb orbit admits a transverse homoclinic orbit in each branch of its stable and unstable manifolds (Theorem 1.4, implying Theorem 1.1); (2) one can find an embedded ∂-strong Birkhoff section containing any prescribed finite collection of periodic orbits in its boundary and a C^0-small Legendrian isotopic copy of any prescribed Legendrian link in its interior (Theorem 1.7); and (3) if a Reeb flow admits a Birkhoff section, then every Legendrian knot has infinitely many Reeb chords, with an explicit finite exception when the manifold is a lens space or the sphere with exactly two periodic orbits (Theorem 1.10). The proof of Theorem 1.4 follows the Le Calvez–Sambarino approach for surface diffeomorphisms, adapted to the first-return map on a Birkhoff section, using the authors' previous work [3] to eliminate degenerate points. The proof of Theorem 1.7 uses Fried pairs of pants, and the proof of Theorem 1.10 uses the flux-zero property of the first return map together with the Hutchings–Taubes chord theorem.","tokens_in":42702,"tokens_out":11089,"duration_ms":127647,"significance":"If the results are correct, they are substantial: the generic existence of homoclinic connections for every hyperbolic periodic orbit is a strong improvement over earlier partial results, the Birkhoff-section statement answers a natural analogue of Giroux's open-book question in a generic setting, and the Reeb-chord dichotomy is a new and meaningful finiteness result. The paper is detailed and makes honest references to external published results, including Irie's equidistribution theorem, Contreras–Mazzucchelli's Birkhoff-section theorem, the authors' own [3], and the Le Calvez–Sambarino machinery. The main weakness is that the paper is explicitly non-self-contained and, more importantly, one of the technical propositions used to eliminate 1-pronged points contains a computational error that is load-bearing for Theorem 1.4. The error appears repairable, but the proof as written is incomplete.","major_comments":[{"comment":"The proof of Proposition 3.2 uses an incorrect count of intersections between the trace of the Birkhoff section and the stable/unstable manifolds of a negative hyperbolic boundary orbit. For a trace of class (a,b) and a stable/unstable curve of class (1,2), the geometric intersection number is |det((1,2),(a,b))| = |b−2a|, not |b|+2|a|. In the case γ ∈ L′∩∂S_n, the paper claims the class (a_1, n b_0) intersects the stable/unstable curve in n|b_0|+2a_1 points, but the actual number is |n b_0 − 2a_1|. Taking a_1=1, b_0=1, n=1 gives 1, so the corresponding point would still be 1-pronged, contradicting the claimed lower bound. The same issue appears in the estimates for γ ∈ L\\L′ and γ ∈ K. Since Proposition 3.2 is the only step that removes 1-pronged points before the Le Calvez–Sambarino adaptation, Theorem 1.4 is not proved as written. The gap appears repairable by choosing n large enough so that |n b_0 − 2a_1| ≥ 2; the authors should replace the invalid lower bound with the correct determinant formula and adjust the choice of n accordingly.","section":"Section 3.1, Proposition 3.2"},{"comment":"The same intersection-number error occurs in the geodesic-flow replacement for Proposition 3.2. After adding a Birkhoff annulus of class (a_1,0) to a boundary component of class (0,b_0), the paper asserts that the resulting trace class (a_1,b_0) meets the stable/unstable curve of class (1,2) in |b_0|+2a_1 points. The correct count is |b_0−2a_1|, which can be 1: for a_1=1 and b_0=1, the claimed lower bound gives 3, but the actual number of intersections is 1. The same problem occurs in the estimate for the orbit σ_+, where the asserted count 1+2a_0 should be |±1−2a_0|. Therefore the proof of Proposition 4.1, and with it the proof of Theorem 1.2 as presented, needs a corrected divisibility argument. As with Proposition 3.2, a suitable choice of the added annuli or of the iteration count may repair the argument, but the current text does not establish the absence of 1-pronged points.","section":"Section 4, Proposition 4.1"}],"minor_comments":[{"comment":"The paper contains several typographical errors that should be corrected in a final revision, including 'address te following' in Section 4, 'at thxe expense' in Appendix A, and 'Propostion 4.1' in the proof of Proposition 4.1.","section":"General"},{"comment":"The paper relies on Proposition 2.11 and Theorem 2.10 of [3] without restating them. Since these are load-bearing inputs, a short statement of their exact content would improve readability and make the logical dependence clearer.","section":"Section 2"},{"comment":"The tetrachotomy for the four local configurations of an elliptic boundary orbit is hard to follow in prose. A table or a small figure showing the signs of q_i, α_i − q_i/p_i, and the resulting boundary orientation would help the reader verify the subsequent inequality.","section":"Section 5.2, Step 1"}],"recommendation":"major_revision","confidential_remarks":"The main concern is the intersection-number formula in Proposition 3.2 and Proposition 4.1. I believe this is a genuine proof gap, but it appears local and repairable: one can choose the integer n (or the added annuli in the geodesic case) to avoid the finitely many values where the true determinant |nb_0−2a_1| equals 0 or 1. I therefore recommend major revision rather than rejection. The referee should ask the authors to correct the count, re-verify the subsequent steps that assume absence of 1-pronged points, and also check the analogous estimates for the other boundary-orbit cases."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper has the right ambitions and a lot of good work in it, but the central proof currently has a hole. The headline results—generic transverse homoclinics for every hyperbolic Reeb orbit, Birkhoff sections with prescribed boundary and Legendrian interior, and the Reeb chord dichotomy—are exactly what people want in this area. The adaptation of Le Calvez–Sambarino to the degenerate setting is intricate, and the chord arguments in Section 6 look independent and sound. But the stress-test note is correct: Proposition 3.2 counts prongs with the wrong formula. On the blow-up torus, the stable/unstable manifold of a negative hyperbolic orbit has homology class (1,2). A trace curve of class (a,b) intersects it in |b−2a| points, not |b|+2|a|. Taking a1=1, b0=1, n=1 gives |1−2|=1, meaning the collapsed point is still 1-pronged; the paper's formula would give 3. So the proof that 1-prongs are eliminated is not established. This is load-bearing: the entire surface-dynamics machinery of Section 3 assumes no 1-pronged points, so Theorem 1.4 is unproved as written. The Birkhoff section improvements in Section 5 rely on Theorem 1.4 for hyperbolic orbits, so they inherit the gap. The Reeb chord theorem in Section 6 does not use Proposition 3.2 and looks like a solid contribution on its own. The paper is transparent about importing technical results from [3], which is fine, but the prong count issue needs to be fixed or replaced. I would send this to a careful referee because the scope is important and the chord part alone is worth reviewing; but the main homoclinic theorem should not be cited until the gap is closed.","headline":"Impressive scope but a real gap in Proposition 3.2: the generic homoclinic theorem is not proven as written.","tokens_in":43223,"tokens_out":6815,"would_cite":false,"duration_ms":70148,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37C29","37E30","53D10","53D35","37D40"],"pacs":[],"model":"deepseek-v4-flash","headline":"Generic Reeb flows force homoclinic orbits on every hyperbolic orbit","keywords":["Reeb flows","homoclinic orbits","Birkhoff sections","Reeb chords","Legendrian knots","contact 3-manifolds","surface dynamics","equidistribution of periodic orbits"],"falsifier":"One concrete check: for a $C^\\infty$-generic Reeb flow on a closed contact 3-manifold, construct the Birkhoff section from the paper's reference [3], collapse its boundary, and compute the first return map on the resulting surface; if some hyperbolic periodic point had stable and unstable branches that never intersect, Theorem 1.4 would fail. Another check targets Theorem 1.10: a $\\partial$-strong Birkhoff section whose boundary has three or more components, together with a Legendrian knot having only finitely many geometrically distinct Reeb chords, would disprove the claimed dichotomy.","tokens_in":42151,"feed_emoji":"🌀","tokens_out":16593,"duration_ms":162669,"temperature":0.7,"pith_summary":"The paper proves three interlocking results about Reeb vector fields on closed 3-manifolds, valid for a $C^\\infty$-generic set of contact forms. First, every hyperbolic periodic orbit has a transverse homoclinic connection in each branch of its stable and unstable manifolds, obtained by adapting a surface-dynamics theorem from the literature through a Birkhoff section. Second, those homoclinic connections allow the construction of embedded Birkhoff sections that contain any prescribed finite collection of periodic orbits in their boundary and any prescribed Legendrian link, up to a $C^0$-small Legendrian isotopy, in their interior. Third, if a Reeb flow admits a Birkhoff section, then every Legendrian knot has infinitely many Reeb chords; the only flows with finitely many geometrically distinct chords are the sphere and lens spaces with exactly two periodic orbits. The paper also gives a new proof of the analogous generic result for geodesic flows on closed surfaces.","feed_headline":"Generic Reeb flows force homoclinic orbits on every hyperbolic orbit","feed_subtitle":"Transverse homoclinics on every hyperbolic orbit yield prescribed Birkhoff sections and infinitely many Reeb chords.","key_machinery":"The proof's main mechanism is a transfer from flow to surface dynamics. Collapsing the boundary components of a $\\partial$-strong Birkhoff section $S$ to points produces a closed surface, and the first return map becomes a homeomorphism whose periodic points are hyperbolic or elliptic inside, and degenerate sectorial points (with $\\ell$ prongs) at the collapsed boundary. The paper adapts a surface-dynamics theorem from the literature ([23]) to this setting, first using the equidistribution condition (G2) to eliminate one-pronged sectorial points (Proposition 3.2), then proving every equivalence class of sectorial periodic points is homoclinic through Lefschetz-index and genus estimates (Propositions 3.10–3.16). A second mechanism is pair-of-pants surgery: homoclinic connections give embedded three-holed-sphere sections (Appendix A) whose addition changes the boundary, forces embeddedness, and slides the section past a Legendrian knot. A third mechanism is the flux-zero property of the first return map, which turns the area-preserving nature of the flow into forced intersections among iterates of a loop cut out by a Legendrian knot, yielding infinitely many Reeb chords.","core_discovery":"The central claim (Theorem 1.4) is that any Reeb vector field on a closed 3-manifold that is strongly non-degenerate (periodic orbits are only hyperbolic or elliptic, with transverse intersections of stable and unstable manifolds), whose periodic orbits are equidistributed with respect to the invariant volume (condition G2), and which satisfies the elliptic-orbit condition (G3), has a transverse homoclinic orbit on every branch of every hyperbolic periodic orbit. Because these three conditions hold $C^\\infty$-generically for a fixed co-oriented contact structure, the paper concludes that a generic Reeb flow has transverse homoclinics on every hyperbolic orbit (Theorem 1.1). The same homoclinic structure powers Theorem 1.7: given any finite set of periodic orbits $\\Gamma$ and any Legendrian link $L$, one can build an embedded $\\partial$-strong Birkhoff section whose boundary contains $\\Gamma$ and whose interior contains a Legendrian link that is $C^0$-close to $L$. Theorem 1.10 upgrades the chord conjecture from at least one Reeb chord to infinitely many for every Legendrian knot whenever a Birkhoff section exists, with the only finite-geometric-chord exception being the two-periodic-orbit flows on the sphere or the lens spaces.","pith_inferences":["The two-or-infinitely-many dichotomy for Reeb chords is likely to hold without assuming a Birkhoff section; the paper leaves this as Question 1.14, and the flux-zero mechanism suggests the dichotomy is a purely dynamical fact.","The exponential honest-chord count (Corollary 1.13) gives a growth invariant for Legendrian knots that could be compared with symplectic-homology wrapping numbers; the Birkhoff-section return map makes this growth computable in principle.","Since equidistribution enters only through Proposition 3.2, families of contact forms without the generic equidistribution property (G2) might still have Birkhoff sections yet lack universal homoclinics; testing such families, such as perturbations of integrable Reeb flows, would isolate the role of equidistribution.","The embedded thin pair-of-pants surgery from Appendix A is a local tool that may transfer to other non-singular flows in 3-manifolds, potentially producing global surfaces of section for flows that are not Reeb."],"forward_implications":["Generic Reeb flows on closed 3-manifolds have transverse homoclinic connections on every branch of every hyperbolic periodic orbit, not merely positive entropy.","Any finite set of periodic orbits can be included in the boundary of an embedded global surface of section, and any Legendrian link can be moved by a $C^0$-small Legendrian isotopy into its interior; the same surgery works for transitive uniformly hyperbolic flows without a contact structure (the paper's Theorem 5.2).","Birkhoff sections imply infinitely many Reeb chords for every Legendrian knot; finite geometric distinctness occurs exactly in the two-periodic-orbit sphere or lens space case.","For geodesic flows on closed Riemannian surfaces, every Legendrian knot has infinitely many geometrically distinct chords, without any genericity assumption on the metric.","Under the paper's generic hypotheses, every Legendrian knot can be deformed by a $C^\\infty$-small Legendrian isotopy so that the number of honest Reeb chords grows exponentially with the action."],"supporting_citations":[{"why":"Supplies the mechanism constructing $\\partial$-strong Birkhoff sections from equidistribution, the homology-class criterion (Theorem A.1), and the surface-dynamics lemmas that the homoclinic proof adapts.","marker":"[3]"},{"why":"Gives the Birkhoff section from the strongly non-degenerate hypothesis (G1), used in Theorem 2.5.","marker":"[6]"},{"why":"Provides the equidistribution of periodic orbits (G2), the genericity property that Proposition 3.2 uses to remove one-pronged points.","marker":"[18]"},{"why":"The surface-dynamics theorem and its proof strategy (equivalence classes, genus, Lefschetz index arguments) that the flow case follows.","marker":"[23]"},{"why":"Supplies the prime-ends rotation and continuum results (Corollaries 8.7 and 8.9) used in Lemma 3.5 and Corollaries 3.6–3.7.","marker":"[21]"},{"why":"Provides the sectorial point theory and the branch-invariance result (Corollary 8.3) used to define equivalence classes.","marker":"[24]"},{"why":"Proves the chord conjecture in dimension 3, giving the initial Reeb chord whose existence Theorem 1.10 amplifies to infinitely many.","marker":"[15, 16]"},{"why":"Introduces the pair-of-pants surgery that later sections refine to embedded thin sections and use to modify Birkhoff sections.","marker":"[13]"},{"why":"Classifies flows with exactly two simple Reeb orbits as irrational pseudo-rotations, used at the end of Theorem 1.10's proof.","marker":"[11]"},{"why":"Replaces equidistribution in the geodesic case by density of closed geodesics, used in Proposition 4.1.","marker":"[17]"}],"fun_headline_variants":["Generic Reeb flows: homoclinics on all hyperbolic orbits","Homoclinic orbits become generic in 3D Reeb flows","Every hyperbolic orbit gets a homoclinic connection","Generic Reeb flows force homoclinics, infinite Reeb chords","3D Reeb flows: homoclinics yield Birkhoff sections, chords"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof requires that a $\\partial$-strong Birkhoff section can be modified so that no boundary orbit collapses to a one-pronged point in the induced surface return map; this modification uses the equidistribution of periodic orbits (or, for geodesic flows, the density of closed geodesics), and without it the surface-dynamics argument cannot begin.","fun_headline_variants_meta":{"raw":{"variants":["Generic Reeb flows: homoclinics on all hyperbolic orbits","Homoclinic orbits become generic in 3D Reeb flows","Every hyperbolic orbit gets a homoclinic connection","Generic Reeb flows force homoclinics, infinite Reeb chords","3D Reeb flows: homoclinics yield Birkhoff sections, chords"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000893,"raw_usage":{"total_tokens":3909,"prompt_tokens":1062,"completion_tokens":2847,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":678,"completion_tokens_details":{"reasoning_tokens":2754}},"tokens_in":678,"tokens_out":2847,"duration_ms":21546,"temperature":1.0,"reasoning_tokens":2754,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T17:56:22.392884+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"One concrete check: for a $C^\\infty$-generic Reeb flow on a closed contact 3-manifold, construct the Birkhoff section from the paper's reference [3], collapse its boundary, and compute the first return map on the resulting surface; if some hyperbolic periodic point had stable and unstable branches that never intersect, Theorem 1.4 would fail. Another check targets Theorem 1.10: a $\\partial$-strong Birkhoff section whose boundary has three or more components, together with a Legendrian knot having only finitely many geometrically distinct Reeb chords, would disprove the claimed dichotomy.","supporting_citations":[{"cited_title":"Colin, P","cited_arxiv_id":null,"evidence_quote":"Supplies the mechanism constructing $\\partial$-strong Birkhoff sections from equidistribution, the homology-class criterion (Theorem A.1), and the surface-dynamics lemmas that the homoclinic proof adapts."},{"cited_title":"Contreras and M","cited_arxiv_id":null,"evidence_quote":"Gives the Birkhoff section from the strongly non-degenerate hypothesis (G1), used in Theorem 2.5."},{"cited_title":"Irie, Equidistributed periodic orbits of C∞-generic three-dimensional Reeb flows","cited_arxiv_id":null,"evidence_quote":"Provides the equidistribution of periodic orbits (G2), the genericity property that Proposition 3.2 uses to remove one-pronged points."},{"cited_title":"Le Calvez, M","cited_arxiv_id":null,"evidence_quote":"The surface-dynamics theorem and its proof strategy (equivalence classes, genus, Lefschetz index arguments) that the flow case follows."},{"cited_title":"Koropecki, P","cited_arxiv_id":null,"evidence_quote":"Supplies the prime-ends rotation and continuum results (Corollaries 8.7 and 8.9) used in Lemma 3.5 and Corollaries 3.6–3.7."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the sectorial point theory and the branch-invariance result (Corollary 8.3) used to define equivalence classes."},{"cited_title":"Fried, Transitive Anosov flows and pseudo-Anosov maps , Topology, 22 (1983), 299–303","cited_arxiv_id":null,"evidence_quote":"Introduces the pair-of-pants surgery that later sections refine to embedded thin sections and use to modify Birkhoff sections."},{"cited_title":"Cristofaro-Gardiner, U","cited_arxiv_id":null,"evidence_quote":"Classifies flows with exactly two simple Reeb orbits as irrational pseudo-rotations, used at the end of Theorem 1.10's proof."},{"cited_title":"Irie, Dense existence of periodic Reeb orbits and ECH spectral invariants","cited_arxiv_id":null,"evidence_quote":"Replaces equidistribution in the geodesic case by density of closed geodesics, used in Proposition 4.1."}],"review_version":1}