{"id":"40e06b06-2c6a-4c30-91bb-3747182bae7b","arxiv_id":"2506.06110","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The parametrised Whitehead torsion of a family of nearby Lagrangians in a cotangent bundle factors through the stable h-cobordism space of a point, yielding triviality on π0 and π1.","lead":"The paper shows that a topological invariant of families of exact Lagrangian submanifolds, called parametrised Whitehead torsion, factors through a much simpler space and therefore vanishes on the first two homotopy levels. This gives new evidence for the strong nearby Lagrangian conjecture and restricts which diffeomorphisms can arise as monodromies of Lagrangian loops on tori.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5.7, the parametrised twisted generating-function step, is only sketched via a doubling/restriction argument; the fiberwise critical-locus assertion is not justified, and the main factorization depends on it.","rationale":"The reader's conditional verdict is appropriate, and I do not move it. I agree with the reader that Theorem 5.7 is the load-bearing external premise. I want to sharpen the point: the paragraph in §5.1 does not merely omit details; it asserts a property, namely that restricting the doubled generating-function data back to M × B behaves as the theorem requires, but this is not a formal consequence of the closed-generating-function theorem. Restriction of a Morse-Bott function to a slice can create new critical points unless one additionally controls the derivative in the sliced direction. This is exactly the place where the proof could break. The rest of the geometric construction in Sections 2–4 and 5.3–5.5 is detailed, and I found no independent internal inconsistency; the torus application in Section 6 follows if Theorem 1.2 and the quoted Hatcher–Hsiang–Sharpe computation hold. Because the gap is a verification problem rather than a known counterexample, and because the authors explicitly rely on an unpublished preprint, a conditional verdict rather than rejection is the right outcome.","tokens_in":33007,"tokens_out":29713,"duration_ms":341440,"concrete_test":"Work the omitted step in the simplest nontrivial case: B = [0,1], M = S^1, and L_b the graph of d(b·h) in T^*S^1. Form the double Legendrian L'' ⊂ J^1(S^1 × S^1), apply the untwisting construction of [ACGK20, Lemma 3.21], and compute the slice critical set Crit(K_b) for b ∈ (0,1). If ∂_{x,v}K = 0 occurs at a point with ∂_bK ≠ 0, Theorem 5.7(2) is false as stated. If the example verifies, repeat for a family with a corner at ∂B and check that the even trivial negative eigenbundle survives; a complete parametrised version of Theorem 5.5 would settle the general case.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 1.2 reduces the problem to Theorem 4.7, whose construction is conditional on Theorem 5.7. The proof of Theorem 5.7 is not carried out in this paper: §5.1 says the parametrised version of [ACGK20]/[AAGCK] 'can be deduced directly' by doubling B and applying the closed-manifold theorem to L'' ⊂ J^1(M × (B ∪_∂B B)), then 'restrict everything to M × B'. The missing content is the slice restriction step. A global function K on the total space with critical locus L'' does not automatically give slices K_b whose critical locus is exactly L_b: the critical equations for K_b are ∂_{x,v}K = 0 without ∂_bK = 0, so the slice can have spurious critical points. The same issue affects the assertion that the negative eigenbundle is a trivial even-rank bundle over the family. Without Theorem 5.7(1)–(3), the bundles ∂^-A_{L,b} and sections s_{L,b} in §5.3–5.4 are unavailable, and Theorem 4.7 is unsupported. This is an external dependency on the preprint [AAGCK] in addition to [ACGK20], and the boundary-doubling trick is not written out.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the parametrised Whitehead torsion w: L(M) -> H(M) for families of closed exact Lagrangian submanifolds in a cotangent bundle. The main theorem, Theorem 1.2, states that for any finite CW-complex B and any map B -> L(M), the composite with w factors, up to homotopy, through the stable h-cobordism space of a point H(pt) via the product map p: H(pt) -> H(M). From this the authors deduce that w vanishes on pi_0 and pi_1, that w is weakly nullhomotopic when chi(M)=0, and, in the case of high-dimensional tori, strong restrictions on Lagrangian monodromies. The proof combines: (i) a theory of families of h-cobordisms and parametrised Whitehead torsion, (ii) a structural theorem (Theorem 3.1) comparing products of h-cobordisms with sums of disk bundles, and (iii) a twisted-generating-function construction (Theorem 5.7) producing a family of difference functions of tube type whose critical loci are the given Lagrangians. The paper is clearly written and the h-cobordism sections are careful, but the parametrised generating-function step is only sketched and is load-bearing for the main theorem.","tokens_in":33227,"tokens_out":17334,"duration_ms":192685,"significance":"If correct, the paper gives the first general constraint on the topology of the trivial path component of the space of nearby Lagrangians, beyond the pi_0 vanishing of Abouzaid-Kragh. The statement is strong and falsifiable: it forces w to vanish on pi_0 and pi_1 and makes the image of w divisible by the Euler characteristic, which is a concrete obstruction to exotic Lagrangian monodromies. The paper also contains an attractive and apparently self-contained result, Theorem 3.1, on comparing product h-cobordisms with sums of disk bundles, which may be of independent interest. The main risk is that the central new analytic input, the parametrised difference-function theorem, is not proved in the manuscript and is outsourced to a sketch plus an unpublished preprint. The geometric parts of the paper are detailed, with explicit constructions and proofs for the h-cobordism facts; however, the main factorisation theorem inherits its validity from Theorem 5.7, whose proof is currently incomplete.","major_comments":[{"comment":"The parametrised version of the twisted generating-function theorem is not proved. The text says that one can double B, apply Theorem 5.5 to the resulting closed manifold, and then 'restrict everything to M × B'. This restriction step is not justified and is in fact delicate. If K is a global Morse-Bott difference function on M × B × R^l with critical locus L', then a point of the slice K_b is critical for the slice only when the x- and v-derivatives vanish; criticality of the global K additionally requires the b-derivative to vanish. Thus the slice K_b can acquire spurious critical points that do not lie on L_b. Consequently the identification Crit(K_b) ≅ L_b in Theorem 5.7(2), and the triviality of the negative eigenbundle in Theorem 5.7(3), do not follow from the argument given. This is not a cosmetic gap: Theorem 4.7 and then Theorem 1.2 depend directly on Theorem 5.7(1)-(3) through the disk-bundle construction of §5.3-5.4. Please either supply a complete parametrised proof, or state precisely which theorem in [AAGCK] is being invoked and verify that the slice restriction preserves the critical locus and the even-rank trivial negative eigenbundle.","section":"§5.1, Theorem 5.7"},{"comment":"The proof of the triviality of the negative eigenbundle asserts: 'Since g_b is a homotopy equivalence, there are families of vector bundles E'_b and E''_b on M such that E'_b ⊕ E''_b is trivial and g_b^* E'_b ⊕ E_b is trivial of rank 2k.' This is not automatic from the given hypotheses. A family of homotopy equivalences over B does not, without further argument, admit a continuous family of homotopy inverses; moreover the stable complements E'_b and E''_b must be chosen compatibly over B. The sentence also changes the fibre dimension l by stabilising E''_b; this should be reconciled with the later inequality l ≫ n + 2k used in Lemma 5.8 and Corollary 2.18, so that the disk-bundle construction in §5.3 is well-defined for the chosen l.","section":"§5.1, proof of Theorem 5.7(3)"}],"minor_comments":[{"comment":"The word 'euivalently' is a typo for 'equivalently'.","section":"§1.1, Remark 1.7"},{"comment":"A family of Lagrangians over a manifold with boundary is not explicitly assumed to be collared, i.e. locally constant near the boundary; such a collar condition is needed for the doubling argument in §5.1 unless it is proved separately that every family can be isotoped to be locally constant near ∂B.","section":"§4.1, Definition 4.1"},{"comment":"The text says 'for l ≫ 0 large enough' when applying Corollary 2.18 to obtain the family V_b; it would be helpful to state the required lower bound in terms of dim(B), n and k, since the validity of the connectivity estimate depends on those dimensions.","section":"§5.4, Corollary 2.18 application"},{"comment":"The notation eG(M) is used before it is introduced; the sentence containing 'π2T^n ∼= 0, W h_2(π1T^n) ∼= 0' would benefit from a reference for Wh_2 and for the vanishing of the relevant k-invariant.","section":"§6.1"}],"recommendation":"major_revision","confidential_remarks":"The decisive issue is the unproved parametrised generating-function step, Theorem 5.7, which is load-bearing for the main result and is delegated to a sketch plus the unpublished preprint [AAGCK]. I would ask the editor to insist on a complete proof or a precise citable statement of the parametrised theorem, together with a verification of the slice-restriction point. If that can be supplied, the paper is likely to be a strong contribution; as it stands, the main theorem is conditional on a substantial external input."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is the first paper to constrain the parametrised Whitehead torsion of families of nearby Lagrangians in a general way. Theorem 1.2 factors w through H(pt), which gives π0 and π1 vanishing, and Theorem 1.5 turns that into divisibility of the image by χ(M). The torus monodromy corollary is concrete and new. The h-cobordism machinery in Sections 2–3 is well executed, and Theorem 3.1 is a solid result on its own. The paper is honest about what remains open (Remarks 1.3 and 1.8).\n\nThe soft spot is exactly what the stress-test note flags: Theorem 5.7, the parametrised twisted generating function statement, is not proved in the paper. The doubling argument leaves out the slice restriction step. A global K on M × (B ∪∂B B) with critical locus L″ does not automatically give slices K_b whose critical locus is exactly L_b; slice critical equations only impose ∂_{x,v}K = 0, so spurious critical points can appear. The same gap affects the claim that the negative eigenbundle is trivial over the family. This is load-bearing: Section 5.3’s disk bundles, hence Theorem 4.7 and Theorem 1.2, rest on it. The paper relies on [AAGCK], an unpublished preprint, and the parametrised version is not obviously stated there either. This is a dependency risk, not a circularity.\n\nThat said, I do not think the paper is wrong. The rest of the proof is careful, the reduction from Theorem 4.7 to Theorem 1.2 is clear, and the topological input (Igusa, Waldhausen, Hatcher) is standard. The parametrised generating function step is plausible and likely fixable, maybe by perturbation or by quoting a more precise statement from AAGCK.\n\nWho should read it: symplectic topologists working on the nearby Lagrangian conjecture and anyone interested in parametrised torsion. It deserves a serious referee, but the referee should be asked to check Theorem 5.7 carefully. If the authors supply a complete proof or a precise reference for the parametrised statement, this becomes a strong paper. I would accept it for peer review and require revision on that point.","headline":"Strong and likely correct extension of Abouzaid–Kragh, but the parametrised generating-function step it leans on is only sketched and depends on an unpublished paper.","tokens_in":33793,"tokens_out":4190,"would_cite":true,"duration_ms":39114,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D12","57R67","57R52"],"pacs":[],"model":"deepseek-v4-flash","headline":"For any finite family of nearby Lagrangian submanifolds, the parametrised Whitehead torsion factors through the h-cobordism space of a point, forcing vanishing on π0 and π1 and divisibility by the Euler characteristic.","keywords":["parametrised Whitehead torsion","nearby Lagrangian conjecture","twisted generating functions","h-cobordism spaces","Lagrangian monodromy","cotangent bundle","Euler characteristic","mapping class group of tori"],"falsifier":"A concrete refutation would be a family φ: B→L(M) with χ(M)=0 whose parametrised torsion w∘φ is not nullhomotopic, or a family with χ(M)≠0 whose image in π_1H(M) is not divisible by χ(M). Alternatively, a loop of exact Lagrangians in T^*T^n whose monodromy projects nontrivially to GL_n(Z)⋉(Z/2)^∞ would contradict Corollary 1.10.","tokens_in":32767,"feed_emoji":"🌀","tokens_out":5673,"duration_ms":56807,"temperature":0.7,"pith_summary":"This paper seeks to show that the parametrised Whitehead torsion of any finite family of closed exact Lagrangian submanifolds in a cotangent bundle is completely controlled by the h-cobordism space of a point. Concretely, for any map φ from a finite CW-complex B into the space L(M) of nearby Lagrangians, the torsion map w∘φ factors, up to homotopy, through the product-with-M map p: H(pt)→H(M). Because the h-cobordism space of a point has trivial π0 and π1, this forces w to vanish on π0 and π1, recovering and generalising the simple-homotopy theorem for individual nearby Lagrangians. The paper further shows that p agrees weakly with multiplication by the Euler characteristic, so when χ(M)=0 the parametrised torsion of every finite family is weakly nullhomotopic. These constraints apply to Lagrangian monodromy, ruling out all but finitely many diffeomorphism monodromies on high-dimensional tori.","feed_headline":"Nearby Lagrangian families' torsion factors through a point","feed_subtitle":"Parametrised Whitehead torsion of exact Lagrangians is controlled by H(pt), killing π0 and π1 and forcing divisibility by χ(M).","key_machinery":"The load-bearing machinery is the theory of twisted generating functions of tube type, used to produce a global difference function K_b on M × R^l for each Lagrangian L_b: a Morse-Bott function whose critical locus is diffeomorphic to L_b, whose fibrewise negative eigenbundle is trivial of even rank 2k, and whose fibrewise restrictions are almost quadratic of tube type. Flowing along the gradient of K_b converts this data into disk bundles and h-cobordisms, ultimately giving the family X_b of h-cobordisms on a disk that encodes δ. A second ingredient is the product h-cobordism theorem of Section 3, which shows that multiplying an h-cobordism by a manifold P is weakly the same as taking χ(P,∂P) copies of a disc product, thereby identifying the map p with multiplication by the Euler characteristic.","core_discovery":"The paper's central claim is Theorem 1.2: for any finite CW-complex B and any map φ: B→L(M), there is a map δ: B→H(pt) such that w∘φ is homotopic to p∘δ, where p: H(pt)→H(M) takes a stable h-cobordism on a point and multiplies it by M. The proof constructs, using parametrised twisted generating functions of tube type, a smooth family of disk bundles E_b over the Lagrangians L_b together with diffeomorphisms E_b ≅ M × (D^l ∪ X_b), where X_b is a smooth family of h-cobordisms on a disk; the family X_b provides the map δ. A structural result about product h-cobordisms (Theorem 3.1) then identifies p with multiplication by χ(M) up to weak homotopy, giving divisibility of the image of w and, when χ(M)=0, weak nullhomotopy of w. The paper presents these as the first general constraints on the topology of the trivial path component of L(M), and derives from them a restriction on Lagrangian monodromies on high-dimensional tori.","pith_inferences":["The paper's Remark 1.3 suggests δ should extend to a map on all of L(M), which would upgrade weak nullhomotopy to genuine nullhomotopy and exclude phantom-map behaviour; constructing that extension is a natural next step.","If the factorisation is natural enough, one might expect analogous factorisations for families of exact Lagrangians in more general symplectic manifolds, replacing the tube-type difference functions with a suitable local model.","The divisibility by χ(M) gives a testable algebraic signature: in degrees where π_*H(pt) is nontrivial, a manifold with χ(M)=1 should never support a family whose torsion class is primitive in H(M).","The torus monodromy conclusion is stated for loops, but the same argument should constrain higher-dimensional monodromy homomorphisms π_k L_0(M)→π_{k-1}Diff(M) when the mapping class groups are computable."],"forward_implications":["For any finite CW-complex B, the parametrised Whitehead torsion w∘φ of a family of nearby Lagrangians factors through H(pt), so all homotopy-theoretic constraints on H(pt) pass to w.","The map w is trivial on π0 and on π1, giving a new proof of the simple-homotopy theorem for individual nearby Lagrangians and extending it to one-parameter families.","The image of w is divisible by χ(M) in the homotopy groups of H(M), and when χ(M)=0 the parametrised torsion of every finite family is weakly nullhomotopic.","For a high-dimensional torus T^n, the monodromy of any loop of exact Lagrangians based at the zero section is isotopic to the identity through homeomorphisms, ruling out all but finitely many potential diffeomorphism monodromies.","The factorisation through H(pt) gives a general mechanism for transferring classical facts about h-cobordism spaces of a point to constraints on families of Lagrangians in cotangent bundles."],"supporting_citations":[{"why":"Proves the π0 vanishing of the parametrised Whitehead torsion that this paper extends to all homotopy groups through a factorisation.","marker":"[AK18]"},{"why":"Develops twisted generating functions and their untwisted difference functions, which are the main source of the Morse-Bott functions used here.","marker":"[ACGK20]"},{"why":"Reformulates twisted generating functions in the quadratic tube-type setting needed for the parametrised Theorem 5.7.","marker":"[AAGCK]"},{"why":"Constructs the parametrised Whitehead torsion as a space-level map from the structure space to the h-cobordism space.","marker":"[WW88]"},{"why":"Gives the injectivity properties of parametrised Whitehead torsion on pseudoisotopy quotients used to interpret what w detects.","marker":"[WW01]"},{"why":"Computes the mapping class groups of high-dimensional tori and identifies the pseudoisotopy-to-identity subgroup used in Corollary 1.10.","marker":"[Hat78]"},{"why":"Supplies the companion Hsiang-Sharpe computation of mapping class groups of tori needed for the monodromy application.","marker":"[HS76]"}],"fun_headline_variants":["Nearby Lagrangian torsion: trivial on π0, π1, divisible by χ","Lagrangian torsion factors through H(pt)","Divisibility by χ for nearby Lagrangian torsion","High-dim torus monodromy constrained by torsion","Torsion kills π0 and π1 for nearby Lagrangians"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the parametrised twisted-generating-function theorem holds for every family of nearby Lagrangians after doubling the parameter space, producing a single difference function whose negative eigenbundle is trivial of even rank; if this construction fails, the disk-bundle factorisation collapses.","fun_headline_variants_meta":{"raw":{"variants":["Nearby Lagrangian torsion: trivial on π0, π1, divisible by χ","Lagrangian torsion factors through H(pt)","Divisibility by χ for nearby Lagrangian torsion","High-dim torus monodromy constrained by torsion","Torsion kills π0 and π1 for nearby Lagrangians"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001445,"raw_usage":{"total_tokens":5814,"prompt_tokens":927,"completion_tokens":4887,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":543,"completion_tokens_details":{"reasoning_tokens":4804}},"tokens_in":543,"tokens_out":4887,"duration_ms":37309,"temperature":1.0,"reasoning_tokens":4804,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T06:00:23.052093+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete refutation would be a family φ: B→L(M) with χ(M)=0 whose parametrised torsion w∘φ is not nullhomotopic, or a family with χ(M)≠0 whose image in π_1H(M) is not divisible by χ(M). Alternatively, a loop of exact Lagrangians in T^*T^n whose monodromy projects nontrivially to GL_n(Z)⋉(Z/2)^∞ would contradict Corollary 1.10.","supporting_citations":[],"review_version":1}