{"id":"a860fc92-3087-4c1f-8f3c-5269de1c1af0","arxiv_id":"2502.06984","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every link in a 3-manifold with a one-sided Heegaard splitting is isotopic to a non-orientable plat closure of a surface braid, with explicit examples in lens spaces and trivial circle bundles.","lead":"The authors define a new way to draw knots and links in a large class of 3-dimensional spaces, using a twisted surface called a non-orientable surface. They prove that every link in such a space can be represented as a 'plat' closure of a surface braid, which may make knots in these spaces easier to study and compare.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.1 is not established: the proof's central sliding assertion is unproven, and Lemma 3.1 is false.","rationale":"The reader's weakest_assumption identifies exactly the soft spot: the proof of Theorem 3.1 is a sequence of unformalized sliding operations, with no rigorous lemma establishing that an arbitrary link can be put in the required normal form. The false relative-homology computation in Lemma 3.1 compounds the concern: if the authors intended the number of capping curves to be tied to the number of generators, the correct count Z/2 would invalidate that intention, although the phrase 'some collection' in Definition 3.1 may leave enough flexibility to save the statement. I therefore do not move the verdict: conditional acceptance remains appropriate until the sliding step is proved and Lemma 3.1 is corrected. My weighting of the geometric gap in Theorem 3.1 is slightly heavier than the reader's, but the two concerns point in the same direction.","tokens_in":12143,"tokens_out":25867,"duration_ms":256077,"concrete_test":"Perform the decisive Fig. 8 slide in the smallest nontrivial case U_2 (Klein bottle), using the explicit model of N(U_2) in Section 3: start with a boundary-parallel arc, and explicitly isotope it (allowing endpoints to move on ∂N) so that the N(U_2)-part is a capping curve and the Σ_{g-1}×I-part is boundary-parallel, with the remaining arcs braided. If the slide can be constructed, the proof's key mechanism is sound; if it cannot, for example because the endpoint pairing forces an arc that is not a fibre, then Theorem 3.1 is unsupported. This same model also gives the correct Mayer-Vietoris computation H1(N(U_2),∂N(U_2)) ≅ Z/2, contradicting Lemma 3.1.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim, Theorem 3.1, depends entirely on the assertion that any properly embedded arc in N(U_g) can be slid, with endpoints allowed to move on ∂N(U_g), so that its intersection with N(U_g) is a collection of mutually unlinked capping curves and its intersection with Σ_{g-1}×I is boundary-parallel (Fig. 8), and that maxima/minima can be similarly resolved (Figs. 9-10). This is not proved, and it is not a standard consequence of general position: a boundary-parallel arc is trivial in H1(N(U_g),∂N(U_g)), while a capping curve is defined to be a generator, so the claimed isotopy would have to change the relative homology class by moving endpoints through Σ_{g-1}×I; no such construction is given. The proof also does not explain why a general link component in N(U_g) can be made boundary-parallel before sliding. Lemma 3.1, which is meant to describe the capping-curve generators, is false: H1(N(U_g),∂N(U_g)) is Z/2 for every g, not (Z/2)^g, because the tube cores do not represent independent relative classes. The theorem may be true, but as written the main existence proof is incomplete.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines a non-orientable plat closure for links in closed orientable 3-manifolds that admit a one-sided Heegaard splitting M = H_{g-1} ∪ (Σ_{g-1}×I) ∪ N(U_g), with U_g a closed non-orientable surface of genus g. Theorem 3.1 asserts that every link in such a splittable manifold is isotopic to the non-orientable plat closure of a surface braid in Σ_{g-1}×I. Sections 4 and 5 construct explicit splittings for lens spaces L(2k,1), L(4a+4,2a+1), and trivial circle bundles Σ_g×S^1 via Bredon–Wood and End embeddings. The proof of Theorem 3.1 is a figure-based isotopy sketch; Lemma 3.1, which is used to describe capping curves, is false as stated.","tokens_in":12400,"tokens_out":8907,"duration_ms":82005,"significance":"This is a potentially useful extension of the RP^3 plat construction of Mishra–Narayanan to the class of splittable manifolds introduced by Rubinstein. A rigorous version of Theorem 3.1 would give a uniform braid-plat normal form for links in a large family of orientable 3-manifolds and would connect with surface braid groups and skein modules. The explicit examples in Sections 4 and 5 are concrete and checkable, and the paper is honest in describing its scope. The central claim is not yet supported by a complete proof, so the paper cannot be accepted in its present form.","major_comments":[{"comment":"The asserted isomorphism H_1(N(U_g), ∂N(U_g)) ≅ (Z/2)^g is false for g>1. Since N(U_g) deformation retracts to U_g, H_1(N(U_g)) ≅ Z^{g-1} ⊕ Z/2. Lefschetz duality for the orientable 3-manifold N(U_g) gives H_1(N(U_g),∂N(U_g)) ≅ H^2(N(U_g)) ≅ Ext(H_1(N(U_g)),Z) = Z/2. The proof in the paragraph around Figure 6 double-counts the unique nonzero relative class. This matters because Definition 3.1 and the main construction rely on the stated description of capping curves in N(U_g).","section":"§3, Lemma 3.1"},{"comment":"The main isotopy assertion is not proved. The text states that every boundary-parallel arc in N(U_g) can be slid so that its intersection with N(U_g) is a collection of mutually unlinked capping curves and its intersection with Σ_{g-1}×I is boundary-parallel, but no argument or reference is given. This is not a standard general-position fact: a boundary-parallel arc is trivial in H_1(N(U_g),∂N(U_g)), whereas a capping curve is described as a generator, so the claimed sliding must move endpoints through Σ_{g-1}×I in a controlled way. The analogous maxima/minima slides in Figures 9 and 10 are also only schematic. Since this step is exactly what converts an arbitrary link into plat form, the proof of Theorem 3.1 is incomplete.","section":"§3, proof of Theorem 3.1, paragraphs after Figures 8–10"},{"comment":"The proof assumes that an arbitrary link can be considered as a union of boundary-parallel arcs in H and N(U_g) and of curves in Σ_{g-1}×I. General position gives only neatly embedded arcs; a separate argument is needed to show that every link component can be isotoped to this form before the sliding steps. This missing hypothesis is load-bearing because the subsequent figures all start from boundary-parallel arcs.","section":"§3, proof of Theorem 3.1, first paragraph"},{"comment":"The definition of non-orientable plat closure is too loose to make Theorem 3.1 precise. The braid group B_{g-1,n} is not defined, the number of ends of β and the number and placement of capping curves are not specified, and 'mutually unlinked' capping curves are explained only informally by saying that two capping curves are fibres in some parametrization of N(U_g). A rigorous statement of Theorem 3.1 needs a precise description of how the 2n ends of the braid are joined to a specified collection of capping curves in H_{g-1} and N(U_g).","section":"§3, Definition 3.1 and surrounding notation"}],"minor_comments":[{"comment":"There are several typographical errors, e.g. 'One might what kinds' (Section 2), 'Asconsequenceallclosed' (after Theorem 2.2), and 'the resulting manifold is an handlebody' (Section 4.1).","section":"Throughout"},{"comment":"The reference [DL15] is listed twice in the bibliography and one of the two entries should be removed.","section":"References"},{"comment":"The notation Z/2Z is typeset inconsistently, and the passage from Rubinstein's open-handlebody conclusion to the compact-handlebody definition of splittability should be made explicit.","section":"Abstract and Theorem 2.1"},{"comment":"Since these figures carry the main proof, they need more precise captions and a formal statement of the sliding lemma they illustrate; in their current form they are too schematic to verify the isotopy.","section":"Figures 8–10"}],"recommendation":"major_revision","confidential_remarks":"To the editor: this is a promising research announcement, but the main theorem is not yet proven: Lemma 3.1 is incorrect and the proof of Theorem 3.1 relies on an unproved sliding assertion. The examples in Sections 4 and 5 are largely independent of Lemma 3.1 and could appear separately if the main theorem is moved to a subsequent paper with a complete proof."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main thing you should know: this paper gives a clean generalization of the RP^3 plat closure to all splittable 3-manifolds, and the explicit splittings of L(2k,1), L(4a+4,2a+1), and Sigma_g x S^1 are genuinely useful. But the proof of Theorem 3.1 is not complete, and Lemma 3.1 is false. I'd send it to a serious referee, but with the expectation of major revision.\n\nThe false lemma is easy to state: Lemma 3.1 claims H_1(N(U_g), boundary N(U_g)) is (Z/2)^g. The correct group is Z/2. By Lefschetz duality this is H^2(N(U_g)) ~= H^2(U_g) = Z/2, since the twisted I-bundle retracts to U_g and H^2 of a closed nonorientable surface is Z/2. The paper's own model shows why: the tube cores are homologous to curves on the boundary, so they die in the relative group; only the class coming from the double cover survives. The error is not directly load-bearing for the main theorem, but it has to be fixed.\n\nThe bigger issue is the proof of Theorem 3.1. The sliding argument in Figures 8-10 is a figure sketch, not a proof. And as stated, it cannot be right: a boundary-parallel arc in N(U_g) is trivial in H_1(N, boundary), while the capping curves of Figure 7 represent the nonzero Z/2 class. Sliding endpoints along the boundary can add elements of H_1(boundary), but the image of H_1(boundary) in the relative group is the quotient by the Z/2, so it cannot move a trivial class to the nontrivial one. I suspect the intended argument is the standard bridge-position one: trivial arcs in N(U_g) can be pushed entirely into the collar, and arcs that represent the nonzero class are already capping curves after a boundary slide. That would prove the theorem, but it has to be written down. As it stands, the main existence result is plausible but not established.\n\nWhere credit is due: the definition of non-orientable plat closure is a natural extension of the authors' earlier RP^3 work, the one-sided splitting framework is well motivated by Rubinstein's theorem, and the concrete examples in lens spaces and Sigma_g x S^1 are carefully constructed, with the complement identified as a handlebody. The background on Bredon-Wood and End embeddings is accurate, and the citations look appropriate.\n\nThis is a paper for low-dimensional topologists working on braid and plat representations in arbitrary 3-manifolds. It deserves peer review, because the idea is good and the examples are solid, but a referee should require that the proof of Theorem 3.1 be made rigorous and Lemma 3.1 corrected before acceptance.","headline":"Useful extension of plat closures to one-sided splittings, but the main proof is a sketch and Lemma 3.1 is plainly wrong; referee it, but expect major revision.","tokens_in":12922,"tokens_out":7749,"would_cite":false,"duration_ms":67043,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57K10","20F36","57K30"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that every link in a splittable 3-manifold is isotopic to a non-orientable plat closure of a surface braid, and gives explicit splittings for lens spaces L(2k,q) and trivial circle bundles Σ×S^1.","keywords":["non-orientable plat closure","one-sided Heegaard splitting","splittable 3-manifolds","surface braid groups","lens spaces","trivial circle bundles","non-orientable surfaces","knots and links"],"falsifier":"Compute the relative homology H_1(N(U_g),∂N(U_g)) by a cellular or Mayer-Vietoris calculation for the Klein-bottle case U_2: if the group is Z/2 rather than Z/2⊕Z/2, then the capping-curve collection in Definition 3.1 is larger than the topology permits, and the capping step of Theorem 3.1 would need to be revised.","tokens_in":11966,"feed_emoji":"🔗","tokens_out":10284,"duration_ms":80756,"temperature":0.7,"pith_summary":"This paper introduces a normal form for knots and links inside a broad class of closed orientable 3-manifolds: the 'splittable' ones, which decompose as a handlebody glued to the mapping cylinder of the orientating double cover of a non-orientable surface. The paper's central claim is that every link in such a manifold is isotopic to a non-orientable plat closure of a surface braid—a braid in the product layer Σ_{g-1}×I whose strand ends are capped by unlinked arcs in the handlebody and in the twisted neighborhood of the non-orientable surface. This unifies link descriptions in lens spaces L(2k,q) and in trivial circle bundles Σ_g×$S^{1}$, and it opens the door to studying links in these manifolds through surface braid groups.","feed_headline":"Non-orientable plats tame every link in splittable manifolds","feed_subtitle":"Braids on an orientable surface, capped by unlinked arcs, describe all links in these manifolds.","key_machinery":"The central object is the non-orientable plat closure, built from a one-sided Heegaard splitting M = H_{g-1} ∪ Σ_{g-1}×I ∪ N(U_g). A braid in the surface braid group of Σ_{g-1} is closed up by two kinds of 'capping curves': unlinked boundary-parallel arcs in the handlebody H_{g-1}, and similar residual capping curves in the twisted I-bundle N(U_g) over the non-orientable surface U_g. These capping curves are the mechanism that converts a surface braid into a closed link, and the isotopy argument in Theorem 3.1 is the claim that any link can be rearranged into exactly this form by sliding arcs and critical points across the three layers.","core_discovery":"The central discovery is Theorem 3.1: every link in a splittable 3-manifold M = H_{g-1} ∪_φ C(U_g) is isotopic to the non-orientable plat closure of a braid in the surface braid group of Σ_{g-1}=∂H_{g-1}. The proof decomposes M as H_{g-1} ∪ Σ_{g-1}×I ∪ N(U_g) and then slides an arbitrary link into a standard position: monotone braided strands in the middle product layer, with all critical points pushed into capping curves in the two outer pieces. The paper also shows that the Bredon-Wood embeddings in lens spaces L(2k,q) and the standard embeddings in Σ_g×$S^{1}$ induce one-sided Heegaard splittings, so the theorem applies there and explicit non-orientable plat descriptions exist for those manifolds.","pith_inferences":["If the isotopy theorem is correct and the capping-curve count can be made rigorous, a natural next step is a Markov-type equivalence for non-orientable plat closures, which would let one compare different braid words representing the same link; the paper explicitly leaves the algebraic study to a sequel.","One could use this normal form to define numerical invariants, such as a minimal braid index or a plat bridge number for links in splittable manifolds, generalizing classical invariants from S^3; the paper does not explore these.","The validity of the construction depends on the relative-homology count in Lemma 3.1: if H_1(N(U_g),∂N(U_g)) is only Z/2 rather than the claimed (Z/2)^g, then the standard collection of capping curves is smaller than stated, and the capping step in the plat definition needs a revised description.","A direct computational check of the theorem on explicit knots in L(2k,1) or in Σ×S^1, converting a given diagram into the claimed plat form, would test the sliding procedure and could reveal where a rigorous lemma is missing."],"forward_implications":["Every link in every splittable 3-manifold has a non-orientable plat presentation, so surface braid groups become a common algebraic language for links in these manifolds.","The lens spaces L(2k,q) and the trivial circle bundles Σ_g×S^1 are all covered, and the paper gives explicit splitting surfaces and handlebody complements for the families L(2k,1) and L(4a+4,2a+1).","Since any closed orientable 3-manifold that contains an embedded non-orientable surface is splittable, the representation applies well beyond the worked examples.","The non-orientable plat closure provides a direct counterpart to the orientable plat closures associated with Heegaard splittings, setting up a framework in which link invariants could be developed from the braid algebra."],"supporting_citations":[{"why":"Proves that any 3-manifold with nonzero H_2(M;Z/2) admits a one-sided Heegaard splitting, which defines the splittable class the paper works with.","marker":"[Rub78]"},{"why":"Introduces non-orientable plat closures via residual curves in RP^3; this paper generalizes that construction to all splittable manifolds.","marker":"[MN23]"},{"why":"Classifies which non-orientable surfaces embed in lens spaces and in Σ×S^1, giving the genus constraints used in the examples.","marker":"[BW69]"},{"why":"Provides the explicit Bredon-Wood embeddings and the minimal genus formula N(2k,q) used to construct one-sided splittings of lens spaces.","marker":"[End92]"},{"why":"Proves the analogous Markov statement for orientable plat closures from Heegaard splittings, the result Theorem 3.1 mirrors in the non-orientable setting.","marker":"[CG20]"}],"fun_headline_variants":["Non-orientable plats represent every link in splittable manifolds","Every link in splittable 3-manifolds is a non-orientable plat closure","One-sided plats tie down all links in splittable manifolds","Non-orientable plats: a universal diagram for splittable 3-manifolds","From knots to plats: non-orientable surfaces tame every link"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof's load-bearing premise is that any arc lying inside the thickened non-orientable surface can be slid, with its endpoints kept on the boundary, into a standard collection of mutually unlinked capping curves, and that the analogous slides of maxima and minima in the product layer never create obstructions; this is illustrated in Figures 8 to 10 but not proven, and the relative-homology count in Lemma 3.1 that fixes how many capping curves are needed appears to misreport the group as (Z/2)^g when its own generators would collapse to a single Z/2.","fun_headline_variants_meta":{"raw":{"variants":["Non-orientable plats represent every link in splittable manifolds","Every link in splittable 3-manifolds is a non-orientable plat closure","One-sided plats tie down all links in splittable manifolds","Non-orientable plats: a universal diagram for splittable 3-manifolds","From knots to plats: non-orientable surfaces tame every link"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001979,"raw_usage":{"total_tokens":7750,"prompt_tokens":991,"completion_tokens":6759,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":607,"completion_tokens_details":{"reasoning_tokens":6652}},"tokens_in":607,"tokens_out":6759,"duration_ms":44551,"temperature":1.0,"reasoning_tokens":6652,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T14:14:45.302005+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the relative homology H_1(N(U_g),∂N(U_g)) by a cellular or Mayer-Vietoris calculation for the Klein-bottle case U_2: if the group is Z/2 rather than Z/2⊕Z/2, then the capping-curve collection in Definition 3.1 is larger than the topology permits, and the capping step of Theorem 3.1 would need to be revised.","supporting_citations":[],"review_version":1}