{"id":"d71b64e4-22b2-4ca4-95be-3e3802e9a024","arxiv_id":"2509.04890","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Equivariant cohomology isomorphism plus fixed point data determines the equivariant homeomorphism (or symplectomorphism) type of a semi-free Hamiltonian circle manifold of dimension six, under stated geometric assumptions.","lead":"Semi-free Hamiltonian circle actions on six-dimensional symplectic manifolds are shown to be determined, up to equivariant homeomorphism or symplectomorphism, by their equivariant cohomology together with data on fixed points. The result corrects a known theorem of Gonzales and extends the authors' previous work to the topological category.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proof of Theorem 1.6 never returns from the modified pseudo momentum map to the original one; the asserted intertwinement near the critical level is unproven and the induction in Theorem 1.5 depends on it.","rationale":"The reader's condition (2) in the rationale flags exactly the passage from the modified pseudo momentum map μ' back to the original μ, so there is partial agreement. However, the reader's stated weakest assumption was the rigidity assumption in Setting 1.1, which is a different hypothesis; I find the unproven return from μ' to μ more directly load-bearing because it is needed for both the symplectic and topological versions and is internal to the proof. The paper contains substantial independent support: careful local model computations, precise statements of canonical classes, and a mostly coherent induction. The concern is not that the conclusion is false, but that a central proof step is omitted. If the concrete test succeeds, the theorem may be repaired by a collar adjustment; if it fails, the proof of Theorem 1.5 does not establish the claimed rigidity. This does not change the reader's CONDITIONAL verdict, since the issue is serious enough to prevent ACCEPT but does not, at this stage, amount to a demonstrated counterexample.","tokens_in":26257,"tokens_out":12023,"duration_ms":112100,"concrete_test":"Re-derive the non-extremal part of Theorem 1.6 without the renaming step: start with the original momentum map μ, form the modified map μ' from Lemma 3.3, and let g' be the extension constructed for μ'. Then check whether g = φ_2 ∘ g' ∘ φ_1^{-1} maps μ_1^{-1}(t) to μ_2^{-1}(t) for all t in a right-neighbourhood of λ, where φ_i is the gradient-flow identification between μ_i-levels and μ'_i-levels. If this fails, write down the local model V = C_{-1} ⊕ C_1 with μ(v) = -|z_1|^2 + |z_2|^2 and the explicit μ'_{ε,r} from Lemma 3.3, and exhibit the level set that is not preserved. This would settle whether the induction in Theorem 1.5 is valid.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the non-extremal case of Theorem 1.6 (Section 4), the proof modifies the momentum maps μ_i to pseudo momentum maps μ'_i via Lemma 3.3, with μ'_i = μ_i outside μ_i^{-1}([λ-r, λ+r]) and μ'_i - μ_i constant near the fixed point set. The proof then establishes the extension statement for the modified maps and says 'To simplify notation, we rename μ'_i into μ_i.' However, the original theorem requires an extension g: μ_1^{-1}((-∞, λ+δ]) → μ_2^{-1}((-∞, λ+δ]) that agrees with f below λ-r and, for non-maximal λ, intertwines the original pseudo momentum maps μ_1 and μ_2 near level λ+δ. A homeomorphism constructed for μ'_1 and μ'_2 intertwines the modified level sets, but μ_i and μ'_i have different level sets inside the tubular neighbourhood U of the fixed set. The proof never returns from μ' to μ, and the sentence in Theorem 1.6 saying that 'it may be assumed that g intertwines the pseudo momentum maps μ_1 and μ_2 near level λ+δ' is asserted without proof. Because Theorem 1.5 is proved by repeatedly applying Theorem 1.6 across all critical levels, this missing intertwinement is load-bearing: without it the induction cannot continue past a non-extremal critical level. This gap affects both the symplectic and the topological versions, since both use the same modification argument.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper investigates equivariant cohomological rigidity for semi-free Hamiltonian S^1-manifolds of dimension six. Its main result, Theorem 1.5, asserts that if two such manifolds (in a symplectic setting, Setting 1.1, or a topological setting, Setting 1.3) have μ-isomorphic equivariant cohomologies together with compatible fixed-point data and an isomorphism of neighbourhoods of the minima, then they are equivariantly homeomorphic (or equivariantly symplectomorphic) with the prescribed cohomology isomorphism. The proof is organized around an extension theorem, Theorem 1.6, which extends an isomorphism defined below a critical level across that level. The central technical tools are the construction of canonical classes associated to fixed-point components (Theorem 1.10), a modification of the momentum map to a pseudo momentum map with isolated critical levels (Lemma 3.3), and applications of isotopy theorems in four-manifolds in both the topological (Sunukjian) and symplectic (Lalonde–Pinsonnault) settings.","tokens_in":26442,"tokens_out":12807,"duration_ms":114614,"significance":"If the proof is completed, this would be a substantial and definitive result: it removes the restrictive distribution assumption on non-extremal fixed surfaces from the authors' previous work, extends the rigidity statement to the equivariant topological category, and provides the kind of fixed-point-data-to-isomorphism statement needed in applications such as Cho's classification of monotone semi-free Hamiltonian S^1-manifolds. The paper is also valuable for introducing canonical classes as a bridge between equivariant cohomology and geometric extension data, and for carefully formulating the rigidity hypotheses. The topological version, relying on complement-fundamental-group conditions and Sunukjian's isotopy theorem, is a meaningful new contribution. However, the proof as written has a load-bearing gap in the passage from modified pseudo momentum maps back to the original maps, and the verification of the main theorem is therefore incomplete in its current form.","major_comments":[{"comment":"The proof never returns from the modified pseudo momentum maps μ'_i to the original maps μ_i. After stating 'It remains to show that the theorem holds with μ'_i replacing μ_i. To simplify notation, we rename μ'_i into μ_i,' the construction yields an extension g intertwining μ'_1 and μ'_2, but the theorem's conclusion requires g to intertwine the original μ_1 and μ_2 near level λ+δ. The maps μ_i and μ'_i differ by constants on neighbourhoods of the fixed-point set, so their level sets differ there, and the proof does not explain how to correct g so that it preserves the original level sets. The sentence in Theorem 1.6 asserting that 'it may be assumed that g intertwines the pseudo momentum maps μ_1 and μ_2 near level λ+δ' is not justified. This gap is load-bearing because Theorem 1.5 is proved by repeatedly applying Theorem 1.6 across critical levels; without the intertwinement near the new level, the induction cannot proceed past a non-extremal critical level. The issue affects both the symplectic and the topological versions, since both use the same modification argument.","section":"§4, Case I (non-extremal λ), proof of Theorem 1.6"},{"comment":"The isomorphism η is only assumed to be an isomorphism of algebras H^*_{S^1}(M_2)→H^*_{S^1}(M_1), but the proof of Lemma 4.5 uses that η restricts to the identity on H^*({pt.}×CP^∞) in order to conclude c_1 = ±η(c_2) with the correct sign. An arbitrary ring isomorphism of equivariant cohomology rings need not preserve the base ring H^*_{S^1}(pt); for example, the automorphism t↦-t can extend to an algebra automorphism. Since the desired conclusion f^* = η implies that η is base-preserving (f^* is induced by an equivariant map and hence is the identity on the base), Definition 1.4 should explicitly require η and η' to be isomorphisms of H^*_{S^1}(pt)-algebras, or the proof must justify that the μ-isomorphism data forces η to be base-preserving. Without this, the sign argument and the identification η(c_2)=c_1 are not fully supported.","section":"Definition 1.4 and Lemma 4.5"},{"comment":"The proofs of the main results rely at essential junctures on statements from the unpublished companion preprint [KW25]. Examples include [KW25, Theorem 1.9] and [KW25, Lemma 5.11] in the proof of Equation (1.6) for maximal λ, and [KW25, Lemma 5.20] in the symplectic non-extremal case. Because these results are load-bearing for the extension theorem, the manuscript's claim to give a definitive statement is conditional on the correctness and availability of [KW25]. The authors should state precisely which results from [KW25] are used, indicate whether they are proved there or in this paper, and ideally make the present paper self-contained for the statements on which Theorem 1.6 directly depends.","section":"Throughout; dependence on [KW25]"}],"minor_comments":[{"comment":"The displayed definition of μ'_{ε,r} contains a typo: '(−|v_1|^2 + |v_1|^2)' should read '(−|v_1|^2 + |v_2|^2)'.","section":"Lemma 3.3, proof"},{"comment":"The statement begins 'Let M be a semi-free Hamiltonian S^1-manifold of dimension 6 and μ:M→S^1 a proper, pseudo momentum map'; the codomain of μ should be R, not S^1.","section":"Lemma 3.4"},{"comment":"The proof jumps from the assumed isomorphism of neighbourhoods of the minima to applying Equation (1.6) on the whole sublevel set (M_1)_{≤λ_1-r}. It is not explained how the neighbourhood isomorphism is extended over the regular interval below the first critical value; in the symplectic version this appears to require a Moser-type argument using the rigidity assumption of Definition 2.4, and this extension should be stated explicitly.","section":"Proof of Theorem 1.5"},{"comment":"The proof refers several times to 'fig. 1', but no figure appears in the manuscript text. Please ensure the figure is included and that the references are consistent with the displayed illustration.","section":"Proof of Theorem 1.10"}],"recommendation":"major_revision","confidential_remarks":"The paper's overall strategy is sound and the results, if correctly established, are significant. The main obstacle is the missing justification for the return from modified pseudo momentum maps to the original maps in the proof of Theorem 1.6; this is a fixable but essential gap. I would also urge the authors to resolve the dependence on the unpublished preprint [KW25] before publication, either by stating the needed results as theorems in the present paper or by making the companion preprint publicly available in final form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version: this paper is a genuine advance on a question that matters in the field, and the main theorem is probably true, but the proof as written has a gap in the place where it matters most.\n\nWhat's actually new: the authors replace the awkward distribution assumption from their earlier [KW25] (non-extremal fixed surfaces only at one level, etc.) with a global equivariant cohomology condition that is necessary for an isomorphism. That is the right kind of hypothesis. They also push the rigidity statement into the topological category, using a cyclic-fundamental-group condition instead of symplectic rigidity. The construction of canonical classes in Section 3 is careful and the evaluation lemmas (3.11, 3.12) look solid. The writing is clear and they are honest about Gonzales' error.\n\nThe soft spot is real, and it is exactly the one flagged in the stress test. In the non-extremal case of Theorem 1.6, the proof modifies the momentum/pseudo-momentum maps to mu'_i via Lemma 3.3, proves the extension statement for the modified maps, and then says 'rename mu'_i into mu_i.' The problem is that the extension g constructed for mu'_1, mu'_2 is an intertwining homeomorphism for the modified level sets, not for the original mu_1, mu_2. The theorem needs g to intertwine the original pseudo momentum maps near the new level lambda+delta, because Theorem 1.5 iterates Theorem 1.6 across critical levels and the next application requires the map to intertwine the original maps at the start level. The proof never returns from mu' to mu. The sentence in the theorem statement saying 'it may be assumed that g intertwines the pseudo momentum maps mu_1 and mu_2 near level lambda+delta' is exactly the missing step, and it is asserted without proof. This is not cosmetic: without it, the induction in Theorem 1.5 cannot continue.\n\nI want to be fair: the gap looks fixable. Since mu' and mu agree outside a tubular neighborhood of the fixed set, one can probably straighten the level sets with a fiberwise reparametrization and adjust g in a collar of the new level. But it needs to be written, and it is load-bearing.\n\nTwo more concerns, in proportion. The proof relies heavily on the unpublished preprint [KW25] for several key statements (e.g., [KW25, Theorem 1.9] in the symplectic extension and Lemma 4.6 in the topological maximal case). For a definitive paper, those dependencies need to be resolved—either by a published version or by including the statements. Second, the rigidity assumption in Setting 1.1 is strong and hard to verify, so the symplectic theorem's range of applicability is narrower than it first appears; that is not a flaw, but it should be said plainly.\n\nWho this is for: anyone working on Hamiltonian S^1-actions or equivariant cohomological rigidity. The canonical class machinery is a contribution on its own. The paper deserves a serious referee, but the referee should be asked to check the mu' to mu transition carefully and to demand a resolution of the [KW25] dependency before acceptance.","headline":"Real progress on a real problem, but the proof of the key extension theorem has a load-bearing gap at the modified-momentum-map step.","tokens_in":27071,"tokens_out":7007,"would_cite":true,"duration_ms":57971,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D35","55N91","53D20","58D19"],"pacs":[],"model":"deepseek-v4-flash","headline":"Equivariant cohomology and fixed-point data determine the isomorphism type of semi-free Hamiltonian circle actions on six-manifolds.","keywords":["semi-free Hamiltonian circle actions","equivariant cohomological rigidity","six-dimensional symplectic manifolds","fixed point data","canonical classes","Morse flow","symplectic reduction","equivariant homeomorphism"],"falsifier":"Take the pair of six-manifolds used in the paper's counterexample to Gonzales' theorem ([KW25, Example 2.1]) and compute whether their equivariant cohomology algebras admit a $\\mu$-isomorphism $(\\eta,\\eta',(\\eta_\\lambda))$ that extends the given fixed-point isomorphism and the isomorphism of neighbourhoods of the minima. The paper's claim predicts that no such compatible $\\eta$ exists, because the homology class of a fixed surface below the critical level obstructs the matching; if a compatible $\\eta$ were found and the two manifolds were still not equivariantly symplectomorphic (or homeomorphic), Theorem 1.5 would be refuted.","tokens_in":25949,"feed_emoji":"🔄","tokens_out":11753,"duration_ms":94310,"temperature":0.7,"pith_summary":"This paper claims that equivariant cohomology, together with compatible data on the fixed point set, determines the isomorphism type of a compact, connected, semi-free Hamiltonian $S^1$-manifold of dimension six (semi-free meaning all stabilizers are connected). Two such manifolds with equivariantly isomorphic cohomology algebras, matching fixed-point components, matching equivariant normal bundles, and matching neighbourhoods of the minima must be equivariantly symplectomorphic, and in a topological version equivariantly homeomorphic. The result repairs a theorem of Gonzales that the authors showed was false: the earlier correction imposed an artificial restriction on where non-extremal fixed surfaces could appear, and this paper removes it by replacing it with a global equivariant-cohomology condition that any isomorphism would have to satisfy anyway. If the paper is right, classification of these six-manifold actions reduces to checking algebraic data rather than constructing maps by hand.","feed_headline":"Cohomology plus fixed points fixes six-manifold circle actions","feed_subtitle":"Matching algebraic data now forces an equivariant homeomorphism or symplectomorphism of the whole six-manifold.","key_machinery":"The load-bearing object is the canonical class of a connected fixed-point component. For a fixed sphere or an isolated fixed point of index 2 at a critical level $\\lambda$, Definition 1.9 asks for a class $c\\in H^*_{S^1}(M)$ that vanishes on $M_{>\\lambda}$, restricts on the component $C$ to the equivariant Euler class of its positive normal bundle, and vanishes on all other fixed components at $\\lambda$; Theorem 1.10 proves such a $c$ exists and is unique, using the Tolman–Weitsman splitting of the long exact sequence in equivariant cohomology. The key property is that $c$ restricts in the reduced space $M_{\\lambda-r}$ to the Poincaré dual of the Morse-flow preimage $[C']$ of $C$, so evaluating $c$ on spheres computes intersection numbers with $C'$. Under a $\\mu$-isomorphism, $\\eta$ must send canonical classes to canonical classes, which forces the map below $\\lambda$ to preserve the subdivision of homology classes into point classes $D^{\\mathrm{pt}}$ and sphere classes $D^{\\mathrm{sph}}(k,l)$ and to preserve normal-bundle Euler classes. With that rigid homological control, the paper uses ambient-isotopy theorems for spheres in four-manifolds (Sunukjian's theorem in the topological setting, and the Lalonde–Pinsonnault result through the authors' earlier work in the symplectic setting) to isotope the map so that it matches the Morse-flow preimages, then pushes it across the critical level with the Morse flow. A pseudo momentum map—a perturbation of a momentum map that is constant near the fixed set—is the device that lets this argument isolate one fixed component at a time.","core_discovery":"The central claim is Theorem 1.5: if $(M_1,\\mu_1)$ and $(M_2,\\mu_2)$ are compact, connected, semi-free Hamiltonian $S^1$-manifolds of dimension six satisfying Setting 1.1 (symplectic version) or Setting 1.3 (topological version), and if their equivariant cohomology algebras are $\\mu$-isomorphic via compatible data $(\\eta,\\eta',(\\eta_\\lambda)_{\\lambda\\in C})$ with an isomorphism of neighbourhoods of the minima, then there is an equivariant homeomorphism (equivariant symplectomorphism) $f\\colon M_1\\to M_2$ with $f^*=\\eta$ as an isomorphism $H^*_{S^1}(M_2)\\to H^*_{S^1}(M_1)$. The workhorse is Theorem 1.6, an extension principle: an isomorphism defined on the sublevel sets right below a critical level $\\lambda$ extends across $\\lambda$ whenever the equivariant-cohomology diagram commutes and the fixed-point sets at $\\lambda$ are matched by $\\eta_\\lambda$. Iterating this from the minimum to the maximum proves the main theorem, so the paper's truth rests entirely on the claim that the algebraic compatibility is strong enough to force the geometric extension at each critical level.","pith_inferences":["Inference beyond the paper: the canonical-class comparison is likely to generalize to Hamiltonian torus actions with higher-dimensional fixed components whenever the reduced spaces admit an ambient-isotopy theorem; the equivariant-cohomology mechanism does not itself use six-dimensionality except through the four-dimensional reduced spaces.","Inference beyond the paper: the topological version suggests a practical test for the manifolds in Cho's classification: compute the fundamental groups of complements of the fixed spheres in their reduced spaces; if the cyclic condition holds automatically there, the full equivariant classification of monotone semi-free actions follows from Theorem 1.5 alone.","Inference beyond the paper: the proof indicates that the rigidity assumption in the symplectic version may be stronger than needed; the places it is used are to ensure that homologous exceptional spheres are ambiently isotopic and that cohomologous deformations are homotopic to isotopies, so a version that checks these properties only at critical levels, rather than on every interval of regular va"],"forward_implications":["The distribution restriction from the authors' earlier correction is gone: fixed surfaces may occur at several non-extremal levels, and the rigidity conclusion still holds without assuming those levels are simple.","Cho's classification of six-dimensional monotone semi-free Hamiltonian $S^1$-manifolds can be obtained directly from the corrected statement, without the extra fixed-point-distribution assumption that was tailored to that application.","In the topological category, the criterion is checkable from standard data: compare equivariant cohomology, fixed-point components, and neighbourhoods of the minima, and verify that complements of fixed spheres have cyclic fundamental group.","Because a genuine equivariant homeomorphism inducing $\\eta$ necessarily produces the $\\mu$-isomorphism data, the theorem's hypotheses are not incidental: no classification of these actions can use weaker equivariant-cohomology information."],"supporting_citations":[{"why":"Supplies the rigidity assumption and the classification statement that the paper corrects and extends.","marker":"[Go11]"},{"why":"Earlier corrected version by the authors; provides the counterexample to Gonzales, the symplectic extension results used in Section 4, and the local model.","marker":"[KW25]"},{"why":"Proposition 2.1 gives the splitting of the long exact sequence in equivariant cohomology that constructs canonical classes.","marker":"[TW99]"},{"why":"Theorem 6.1 supplies ambient isotopy of homologous spheres in four-manifolds with cyclic complement, used in the topological extension over critical levels.","marker":"[Su15]"},{"why":"Through the authors' Lemma 5.20, supplies the ambient symplectic isotopy of exceptional spheres used in the symplectic version.","marker":"[LP04]"},{"why":"Kirwan injectivity lets the proof identify canonical classes by their restrictions to the fixed point set.","marker":"[Ki84]"},{"why":"Provides the integral version of injectivity and cupping statements needed for semi-free actions over Z.","marker":"[TW03]"},{"why":"Isotopy extension theorem used to extend topological isotopies of submanifolds globally in the extension argument.","marker":"[EK69]"}],"fun_headline_variants":["Equivariant cohomology and fixed points decide six-manifold S^1 actions","Fixed data and cohomology force isomorphism of six-manifold S^1 actions","Semi-free circle actions on six-manifolds rigid from cohomology and fixed sets","Definitive rigidity: cohomology and fixed sets pin six-manifold circle actions","Cohomology and fixed data fix six-manifold S^1 actions, topologically too"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The symplectic version of the theorem depends on a global rigidity condition that is hard to verify: for every interval of regular values, the reduced four-manifold must have a path-connected identity component of its symplectomorphism group, and any deformation of the symplectic form that stays in the same cohomology class must be homotopic, through such deformations, to an actual isotopy; if that condition fails anywhere, the proof's extension over a critical level no longer goes through.","fun_headline_variants_meta":{"raw":{"variants":["Equivariant cohomology and fixed points decide six-manifold S^1 actions","Fixed data and cohomology force isomorphism of six-manifold S^1 actions","Semi-free circle actions on six-manifolds rigid from cohomology and fixed sets","Definitive rigidity: cohomology and fixed sets pin six-manifold circle actions","Cohomology and fixed data fix six-manifold S^1 actions, topologically too"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001104,"raw_usage":{"total_tokens":4639,"prompt_tokens":1019,"completion_tokens":3620,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":3503}},"tokens_in":635,"tokens_out":3620,"duration_ms":22527,"temperature":1.0,"reasoning_tokens":3503,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:25:33.140972+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the pair of six-manifolds used in the paper's counterexample to Gonzales' theorem ([KW25, Example 2.1]) and compute whether their equivariant cohomology algebras admit a $\\mu$-isomorphism $(\\eta,\\eta',(\\eta_\\lambda))$ that extends the given fixed-point isomorphism and the isomorphism of neighbourhoods of the minima. The paper's claim predicts that no such compatible $\\eta$ exists, because the homology class of a fixed surface below the critical level obstructs the matching; if a compatible $\\eta$ were found and the two manifolds were still not equivariantly symplectomorphic (or homeomorphic), Theorem 1.5 would be refuted.","supporting_citations":[],"review_version":2}