{"id":"a8ce6d9a-5a1e-47ab-aaa5-6e02ded4ef46","arxiv_id":"2412.15769","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Compact edges of a G2 multi-moment graph lift from the base via cohomological formulas, and in the toric case the loop closes exactly when the cohomological condition [ω]∪[F]=0 holds.","lead":"These mathematicians show how to compute the compact part of a symmetry graph for seven-dimensional G2-manifolds from cohomology classes on a six-dimensional base. The result gives a concrete computational tool for G2-manifolds built as circle bundles over symplectic and toric Calabi-Yau spaces.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the lifting formulas (3.7)-(3.10) and Theorem 4.3 are internally consistent; the issues in Example 5.1 are localized typos, not flaws in the central claim.","rationale":"The reader's CONDITIONAL verdict rests on a sign inconsistency in Example 5.1 and on the Hamiltonian condition (3.2). My independent check of Sections 3 and 4 finds the main derivation coherent: the definition of the lifted generators, the formula for dν3 along an edge, the chain identities relating σX1, σX2 and E, and the computation of Σ tj(fj - fE) = ([ω]∪[F]).E are mutually consistent. The sign issue in Example 5.1 is real but purely typographical: the computed λ(B) and ν3(D) are the ones obtained from (3.8) with [F]=mω+-nω- once [F]∩E_AB=-n is used; the printed s_AB=n and D=(q+,p+) are local errors. The weakest assumption (3.2) is necessary for any lifted T^3-action, but for the toric Calabi-Yau manifolds considered, b1(B)=0 and F is integral, so the potentials λ_i exist; the paper's examples also implicitly satisfy this. I therefore do not see a load-bearing threat to the central claim, and the reader's conditional verdict should stand pending typo corrections rather than escalate.","tokens_in":16596,"tokens_out":32804,"duration_ms":306936,"concrete_test":"Recompute Example 5.1 literally: take F = mω+ - nω-, evaluate on E_AB = {p+} × CP^1- to get -n, apply (3.8) with (r1,r2)=(1,0), and check whether λ(B)=n(0,1) is recovered only if the printed s_AB=n is corrected to -n. If the printed value is used verbatim, the example contradicts (3.8), confirming that the issue is a sign typo rather than a failure of the lifting theorem.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I find no load-bearing flaw in the central cohomological lifting argument. Equations (3.7)-(3.10) follow from the Hamiltonian condition (3.2): the sign in dν3 = -λ2 dµ1 + λ1 dµ2, the vanishing of ω(X1,X2) and ψ+(X1,X2,·) along an edge, and the chain-level identities r1[E] = [σX2], r2[E] = -[σX1] all check out. Theorem 4.3's reduction of the height change to ([ω]∪[F]).E is consistent with (4.5)-(4.7). The premise (3.2) is genuinely load-bearing, but in the toric Calabi-Yau setting b1(B)=0 and F has integral periods, so the required λ_i exist globally; this is an assumption, not a gap. The concrete defect I find is in Example 5.1: with the stated [F]=m[ω+]-n[ω-], one has [F]∩E_AB=-n, not n as printed, and D should be (q+,q-), not (q+,p+). These are typographical errors confined to the illustration; the displayed µ(B), λ(B), and ν3(D) are actually the values consistent with (3.7)-(3.10) after correcting s_AB to -n.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies seven-dimensional manifolds with closed G2-structures obtained as circle bundles over six-dimensional symplectic SU(3)-manifolds carrying a two-torus symmetry. Under a multi-Hamiltonian hypothesis, the authors derive cohomological formulas for how the compact part of the base multi-moment graph lifts to the G2 multi-moment graph: along an edge with stabilizer generated by X = r1X1+r2X2, the moment-map differences are given by equations (3.7)-(3.10), involving the classes [omega], [F], and the homology class of the corresponding surface [E]. In the toric Calabi-Yau case, the paper specializes to a fan computation and proves in Theorem 4.3 that the lifted polygon closes exactly when [omega] union [F] = 0, matching a condition appearing in the Foscolo-Haskins-Nordström construction. The paper closes with three explicit examples illustrating the lifting procedure.","tokens_in":16862,"tokens_out":24262,"duration_ms":205727,"significance":"The main formulas provide a practical, cohomological method for computing the compact part of G2 multi-moment graphs without solving for the metric or the connection. This is a useful tool for the growing literature on circle-bundle constructions of G2-manifolds from Calabi-Yau three-folds, and the identification of the Fenchel-type closing condition with the cohomological condition [omega] union [F] = 0 is a clean and convincing result. The derivation of Section 3 is explicit and checkable, and the fan computation in Section 4 is concrete and parameter-free in the sense that the lift is determined directly by the classes [omega] and [F]. The paper does not provide machine-checked proofs or code, but the hand derivations are sufficiently detailed to be verified independently. The main caveat is that one of the illustrative examples contains sign inconsistencies that need to be corrected; these do not appear to affect the central theorems.","major_comments":[{"comment":"The example as printed is algebraically inconsistent. With [F] = m[omega+] - n[omega-] and the stated [omega] = k(m[omega+] + n[omega-]), one computes [omega] union [F] = +/-2kmn [omega+ union omega-], which is nonzero for kmn != 0; hence the stated implication '[omega] union [F] = 0 implies [omega] = k(m[omega+] + n[omega-])' is false. The correct consequence would be [omega] = k(m[omega+] - n[omega-]), or else the sign of the n-term in [F] must be changed. Moreover, the subsequent values lambda(B) = n(0,1) and nu3(D) = -kmn are consistent with equations (3.8) and (3.10) only if s_AB = -n and the orientation of E_AB is chosen accordingly; the printed value s_AB = n gives lambda(B) = -n(0,1) and then nu3(D) = +kmn. This is an illustrative example rather than a step in the proof of Theorem 4.3, but it should be repaired so that it actually demonstrates the formulas.","section":"Section 4.2, Theorem 4.3"},{"comment":"The proof computes nu3(q_{m+1}) - nu3(q_1) for the loop around a single interior ray E and shows that it equals ([omega] union [F]).E. To justify the global 'if and only if [omega] union [F] = 0', the paper should state explicitly that the classes E = A_j cap E, as E runs over interior rays and A_j over adjacent boundary rays, generate H_2(B;Z) in the present toric setting; otherwise the conclusion is only established for the particular loop under consideration. This generation statement is standard for smooth toric varieties obtained by triangulating a polygon, but it is not written down and is needed for the theorem as stated.","section":"Section 4.2, Theorem 4.3"}],"minor_comments":[{"comment":"The fixed point D is printed as (q+,p+); it should be (q+,q-) to match the later use of E_BD as {p+} x CP(1)-? Actually E_BD is described as CP(1)+ x {q-}, so its endpoints are B = (p+,q-) and D = (q+,q-).","section":"Section 5, Example 5.1"},{"comment":"The printed value lambda(q6) = (-f1+f2, -f1-f2) is inconsistent with equation (4.6) applied to the edge from q5 to q6; using t6 = w1 and u6-uE = (-1,0) gives lambda(q6) = (-f1+f2, f1-f2). The second component does not affect the printed value of nu3(q6), but it should be corrected.","section":"Section 5, Example 5.2"},{"comment":"The standing hypothesis that the functions lambda_i solving (3.2) exist globally should be stated explicitly before equations (3.7)-(3.10) are used, even though it is introduced as a necessary and sufficient condition just above. In the toric Calabi-Yau setting this is guaranteed by b1(B) = 0, but the main formulas of Section 3 apply only under this Hamiltonian assumption on F.","section":"Section 3"},{"comment":"There are several typographical errors: the abstract reads 'may obtained' instead of 'may be obtained', equation (2.1) has 'permuations' instead of 'permutations', and the introduction contains 'sympletic' instead of 'symplectic'. These should be corrected in a final pass.","section":"Throughout"}],"recommendation":"minor_revision","confidential_remarks":"The main lifting argument and the fan computation appear sound. The sign problems in Example 5.1 are real but localized; they do not undermine Theorem 4.3. The missing generation statement in the proof of Theorem 4.3 is a small gap in exposition. I would be happy to see the paper accepted after these local corrections."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper is a genuine contribution. The compact-edge lifting formulas (3.7)–(3.10) and the fan-based closure criterion in Theorem 4.3 are new and the derivations are careful. I went through the path-and-surface arguments leading to (3.7) and (3.8), and the computation of dν3 above an edge; they are consistent. The reduction of loop closure to [ω]∪[F]=0 in the toric setting is clean and the intersection-theoretic proof checks out. The link to the Foscolo–Haskins–Nordström condition is a real payoff, not a formal restatement.\n\nThe one real defect is in Example 5.1. With [F] = m[ω+] − n[ω−], the intersection with E_AB is −n, not n as printed. The displayed λ(B) = n(0,1) and ν3(B)=0 are what come out of (3.8) and (3.10) with s_AB = −n, so only the stated value of s_AB is wrong. Similarly, the point D should be (q+, q−), not (q+, p+). These are typos in the illustration and do not affect the main results, but they should be fixed.\n\nThe Hamiltonian condition (3.2) is load-bearing, but in the toric Calabi-Yau setting with b1(B)=0 it is automatically satisfied for any invariant integral curvature form, so this is an assumption that is met in the intended examples. An explicit remark to that effect would help. Minor typos elsewhere (e.g. 'may obtained' in the abstract) do not matter.\n\nThis paper is for readers working on G2 geometry, multi-moment maps, or toric webs. It deserves a serious referee. I would send it out and recommend acceptance after minor revisions.","headline":"Core lifting formulas and loop-closure criterion are correct; fix the sign typo in Example 5.1.","tokens_in":17464,"tokens_out":6278,"would_cite":true,"duration_ms":46658,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C25","14M25","53C26","53C29","53D20"],"pacs":[],"model":"deepseek-v4-flash","headline":"For circle bundles over symplectic $SU(3)$-manifolds, compact edges of the base multi-moment graph lift to $G_2$ multi-moment graph edges, with endpoint differences given by intersection products $[\\omega]\\cap[E]$ and $[F]\\cap[E]$; in the…","keywords":["G2-manifolds","multi-moment maps","multi-toric graphs","circle bundles","Calabi-Yau three-folds","toric geometry","special holonomy","cohomology"],"falsifier":"Using the star-triangulated hexagon example with integer coefficients for which $[\\omega]\\cup[F] \\neq 0$, compute $\\nu_3(q_7) - \\nu_3(q_1)$ from (4.5) and (4.6); Theorem 4.3 predicts a non-zero value, so a result of zero—or any edge whose endpoint difference contradicts (3.7)–(3.10)—would falsify the central claim.","tokens_in":16346,"feed_emoji":"📐","tokens_out":12298,"duration_ms":92951,"temperature":0.7,"pith_summary":"This paper shows that the multi-moment graph of a seven-dimensional $G_2$-manifold obtained as a circle bundle over a six-dimensional symplectic $SU(3)$-manifold can be computed from cohomology classes on the base, without solving for the lifted geometry. In a multi-Hamiltonian setup—where invariant maps whose differentials are contractions of the invariant forms with torus vector fields play the role of moment maps—each compact edge of the base graph lifts to a straight edge of the $G_2$ graph, and the difference in the lifted coordinates between the edge's endpoints is given by intersection products: $[\\omega]\\cap[E]$ for the symplectic side and $[F]\\cap[E]$ for the curvature side, with the third coordinate determined by these together. In the toric Calabi-Yau case, the paper proves the lifted polygon closes exactly when $[\\omega]\\cup[F]=0$, the same cohomological condition appearing in existing constructions of complete non-compact $G_2$-manifolds. This matters because it reduces a hard lifting problem to intersection-theoretic bookkeeping and shows that a known existence condition is intrinsic to the graph geometry.","feed_headline":"Cohomology decides how toric graphs lift to G2","feed_subtitle":"The symplectic and curvature classes on the six-dimensional base fix the lifted graph's edge data.","key_machinery":"The load-bearing object is the multi-moment graph: for a closed $G_2$-structure with a $T^3$-action, the image under the multi-moment map $\\nu: M \\to \\Lambda^2 \\mathrm{Lie}(T^3)^* \\cong \\mathbb{R}^3$ of the points with non-trivial stabiliser, an embedded trivalent graph whose edges are straight lines with rational tangents and whose vertices satisfy a zero-tension condition. On the six-dimensional base, the analogous graph is defined from the symplectic moment map $\\mu$ of the $T^2$-action on the symplectic $SU(3)$-structure. The lifting mechanism is the relation between the $G_2$ three-form and the $SU(3)$ data of the base, together with the Hamiltonian condition $d\\lambda_i = -F(X_i,\\cdot)$ on the circle bundle's curvature; combining these, formulas (3.7)–(3.10) express endpoint differences as the intersection products $[\\omega]\\cap[E]$ and $[F]\\cap[E]$, with $E$ the surface generated by the edge and the complementary circle action.","core_discovery":"The paper's central claim is that the compact part of the $G_2$ multi-moment graph of a circle bundle $M \\to B$ is determined, edge by edge, by cohomological data of the base. If $B$ carries a symplectic $SU(3)$-structure with an effective multi-Hamiltonian $T^2$-action, and the curvature $F$ of the circle bundle satisfies the Hamiltonian condition (3.2), then each compact edge $e$ of the base graph, whose stabiliser is generated by $X = r_1 X_1 + r_2 X_2$, lifts to a corresponding edge of the $G_2$ graph. Writing $E$ for the surface swept out by the edge under the complementary circle action, the endpoint differences are given by $\\mu(c) - \\mu(b) = [\\omega]\\cap[E](-r_2,r_1)$, $\\lambda(c) - \\lambda(b) = -[F]\\cap[E](-r_2,r_1)$, and $\\nu_3(c) - \\nu_3(b) = t(r_1\\lambda_1 + r_2\\lambda_2)$ with $t = [\\omega]\\cap[E]$. In the toric Calabi-Yau case, Theorem 4.3 states that the lifted polygon closes if and only if $[\\omega]\\cup[F]=0$.","pith_inferences":["The same formulas should apply to any multi-Hamiltonian $T^2$-action on a symplectic $SU(3)$-manifold with a curvature satisfying (3.2), so the lifting procedure is not restricted to toric or Calabi-Yau bases.","Graph closure could serve as a necessary cohomological test for the existence of torsion-free $G_2$-structures in this circle-bundle class: a base whose lifted graph fails to close cannot admit the corresponding $G_2$ metric.","The non-planar examples suggest that varying $[F]$ moves vertices of the lifted graph in the third coordinate, giving a way to engineer $G_2$ multi-moment graphs with prescribed three-dimensional shapes.","The expectation expressed in Remark 5.3, that versal deformations of toric singularities produce graphs whose components lie in distinct parallel affine planes, is a testable consequence of the same lifting picture."],"forward_implications":["For any compact edge of the base multi-moment graph, the lifted $G_2$ edge's direction and length are computed directly from $[\\omega]\\cap[E]$ and $[F]\\cap[E]$, with no differential equations to solve.","A loop in the base graph lifts to a closed loop in the $G_2$ graph exactly when $[\\omega]\\cup[F]=0$ on the corresponding cycle; in the toric Calabi-Yau setting the paper proves this equivalence as Theorem 4.3.","The cohomological condition $[\\omega]\\cup[F]=0$ used in known circle-bundle constructions of complete non-compact $G_2$-manifolds is not an artefact of the adiabatic-limit method, but is forced by the graph geometry itself.","The additive constants in $\\lambda$ and $\\nu$ correspond to affine changes of the graph, so lifted graphs can be computed relative to an arbitrary basepoint, as the paper's examples do.","In the examples (the $M_{m,n}$ spaces, the star-triangulated hexagon, and the second resolution of the conifold quotient), the formulas produce the full compact part of the $G_2$ graph, including non-planar configurations when $[\\omega]\\cup[F]\\neq 0$."],"supporting_citations":[{"why":"Introduces multi-moment maps for closed invariant forms, the map whose lifting behaviour is the subject of the paper.","marker":"[22, 23]"},{"why":"Introduces the $G_2$ multi-moment graph, proves edges are straight lines with rational tangents, and supplies the zero-tension condition used in Proposition 3.1.","marker":"[24]"},{"why":"Gives the decomposition of a $G_2$-structure on a circle bundle into $\\theta$, $\\omega$, $\\psi_\\pm$, and $h$ used throughout the lifting computation.","marker":"[4]"},{"why":"States the necessary and sufficient condition for the existence of a $T^2$-invariant connection with curvature $F$, namely equation (3.2).","marker":"[27]"},{"why":"Shows a $T^2$-action on the base lifts to a $T^3$-action on a principal circle bundle, giving the effective action used to build the $G_2$ graph.","marker":"[19]"},{"why":"Provides the divisor relations, Picard-group description, and triple-intersection formulas used in the toric Calabi-Yau closing argument.","marker":"[13]"},{"why":"Supplies the circle-bundle construction of complete non-compact $G_2$-manifolds and the condition $[\\omega]\\cup[F]=0$ that Theorem 4.3 interprets as closure of the lifted graph.","marker":"[15]"},{"why":"Provides the cohomogeneity-one $G_2$ examples $M_{m,n}$, the first explicit illustration of the lifting formulas.","marker":"[16]"},{"why":"Supplies the conifold-quotient example and its divisor-intersection coefficients used in the final illustration.","marker":"[1]"}],"fun_headline_variants":["Cohomology dictates G2 multi-toric lifts","Base classes fix lifted G2 edges","Symplectic and curvature classes lift G2 graphs","Multi-toric graphs: lifts decided by base cohomology","G2 graph lifts follow from cohomology"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The lifting only works if the circle bundle's curvature $F$ is Hamiltonian for the torus action: functions $\\lambda_1, \\lambda_2$ must exist with $d\\lambda_i = -F(X_i,\\cdot)$, and if they do not exist there is no lifted three-torus action and no $G_2$ multi-moment graph to compute.","fun_headline_variants_meta":{"raw":{"variants":["Cohomology dictates G2 multi-toric lifts","Base classes fix lifted G2 edges","Symplectic and curvature classes lift G2 graphs","Multi-toric graphs: lifts decided by base cohomology","G2 graph lifts follow from cohomology"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000322,"raw_usage":{"total_tokens":1787,"prompt_tokens":897,"completion_tokens":890,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":513,"completion_tokens_details":{"reasoning_tokens":815}},"tokens_in":513,"tokens_out":890,"duration_ms":8292,"temperature":1.0,"reasoning_tokens":815,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:07:47.447463+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Using the star-triangulated hexagon example with integer coefficients for which $[\\omega]\\cup[F] \\neq 0$, compute $\\nu_3(q_7) - \\nu_3(q_1)$ from (4.5) and (4.6); Theorem 4.3 predicts a non-zero value, so a result of zero—or any edge whose endpoint difference contradicts (3.7)–(3.10)—would falsify the central claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the $G_2$ multi-moment graph, proves edges are straight lines with rational tangents, and supplies the zero-tension condition used in Proposition 3.1."},{"cited_title":"Apostolov and S","cited_arxiv_id":null,"evidence_quote":"Gives the decomposition of a $G_2$-structure on a circle bundle into $\\theta$, $\\omega$, $\\psi_\\pm$, and $h$ used throughout the lifting computation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"States the necessary and sufficient condition for the existence of a $T^2$-invariant connection with curvature $F$, namely equation (3.2)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows a $T^2$-action on the base lifts to a $T^3$-action on a principal circle bundle, giving the effective action used to build the $G_2$ graph."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the divisor relations, Picard-group description, and triple-intersection formulas used in the toric Calabi-Yau closing argument."},{"cited_title":"Foscolo, M","cited_arxiv_id":null,"evidence_quote":"Supplies the circle-bundle construction of complete non-compact $G_2$-manifolds and the condition $[\\omega]\\cup[F]=0$ that Theorem 4.3 interprets as closure of the lifted graph."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the cohomogeneity-one $G_2$ examples $M_{m,n}$, the first explicit illustration of the lifting formulas."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the conifold-quotient example and its divisor-intersection coefficients used in the final illustration."}],"review_version":1}