{"id":"951a374c-546d-4bfb-b7dc-d95c5d9700bd","arxiv_id":"2411.10348","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A compact prequantized symplectic manifold with a Lagrangian torus fibration has Riemann-Roch number equal to its number of Bohr-Sommerfeld fibers, proved via a new theorem: volume equals number of integer points on compact integral-integral affine manifolds.","lead":"This paper proves that for a prequantized compact symplectic manifold sliced by Lagrangian tori, the Riemann-Roch number from index theory equals the number of Bohr-Sommerfeld fibers, a count of special quantized cycles. The proof rests on a new standalone result: on a compact integral-integral affine manifold, the total volume equals the number of integer lattice points.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 2.2 as stated fails for disconnected-fiber Lagrangian fibrations: on T^2, π(x,y)=2x mod1 gives RR=1 but |BS|=0.","rationale":"The reader's verdict is CONDITIONAL, and my read does not move that verdict, but it sharpens the condition. The reader identified the unstated connected-fibres hypothesis in Theorem 2.2 as a weakness, yet placed the weakest assumption on Proposition 8.2 and Theorem 7.6. My concern is more direct: under the paper's own terminology, Theorem 2.2 is false without connected fibres, not merely unproved. The T^2 example satisfies the stated hypotheses if a disconnected union of Lagrangian circles is allowed as a Lagrangian submanifold, and it gives |BS| = 0 while RR = 1. This shows that the proof's reliance on Arnol'd-Liouville charts and torus fibrations is not a harmless omission: the stated theorem overclaims. The fix is straightforward: restate Theorem 2.2 for connected-fibre Lagrangian torus fibrations, which is exactly what the abstract and proof use. The sign error in the proof of Theorem 3.1 and the unproved fine-sheaf/Mayer-Vietoris assumption in Theorem 7.6 are genuine issues that should be repaired, but they do not by themselves produce a counterexample. If the intended convention is that a 'Lagrangian fibration' implicitly has connected fibres, then the theorem statement should say so explicitly; as printed, the disconnected example is a counterexample. The Downstairs Theorem may well be true, and the paper's main idea remains plausible, so I do not recommend rejection; I recommend keeping the conditional verdict with the explicit condition that the Upstairs Theorem be restricted to connected-fibre torus fibrations and that the proof gaps in Sections 7 and 8 be fixed.","tokens_in":19872,"tokens_out":27843,"duration_ms":295507,"concrete_test":"Compute the counterexample explicitly: M = R^2/Z^2, ω = dx∧dy, π(x,y) = 2x mod 1, L trivial with connection d − 2πi x dy. For a fibre over b, the components are x = c and x = c + 1/2 with 2c ≡ b mod 1; their holonomies are e^{2πi c} and −e^{2πi c}, so both equal 1 is impossible, and any flat twist preserves the sign difference, giving |BS| = 0. Meanwhile RR = ∫_{T^2} ω = 1. To verify the corrected statement, run the same computation for the connected fibration π(x,y) = y mod 1, where the Bohr–Sommerfeld condition is e^{2πi y} = 1 and |BS| = 1 = RR; this confirms that the theorem holds only after adding the connected-fibre or torus-fibration hypothesis.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Under the paper's own definition in Section 4, a regular Lagrangian fibration does not require connected fibres. Let M = R^2/Z^2 with ω = dx∧dy, and let π: M → S^1 be π(x,y) = 2x mod 1. This is a proper surjective submersion, hence a fibre bundle; each fibre consists of two circles, x = c and x = c + 1/2, both Lagrangian. Take L → M to be the trivial line bundle with connection d − 2πi x dy, whose curvature is ω. The holonomy around the vertical loop at x is e^{2πi x}; on the two components of a fibre the holonomies are e^{2πi c} and −e^{2πi c}. Tensoring with any flat line bundle multiplies both holonomies by the same character, so they can never both be 1. Hence no fibre is Bohr–Sommerfeld and |BS| = 0. But RR(T^2, ω) = ∫_{T^2} ω = 1. Thus Theorem 2.2 is false as written. The proof in Section 5 uses Lemmas 4.4, 4.12, and 5.1, all of which require a Lagrangian torus fibration with connected fibres; the theorem must be restricted to connected-fibre torus fibrations, or the definition must explicitly rule out disconnected fibres. The sign error in the final display of Theorem 3.1 and the fine-sheaf issue in Theorem 7.6 are real but secondary; the disconnected-fibre example is a genuine falsification of the stated Upstairs Theorem.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves two theorems: the \"Upstairs Theorem\" (Theorem 2.2), which asserts that for a compact prequantized symplectic manifold with a regular Lagrangian fibration, the Riemann-Roch number equals the number of Bohr-Sommerfeld fibres, and the \"Downstairs Theorem\" (Theorem 3.1), which asserts that for a compact integral-integral affine manifold, the total volume equals the number of integral points. The Upstairs Theorem is reduced to the Downstairs Theorem through a chain of lemmas: volume equals base volume (Lemma 4.4), Bohr-Sommerfeld points equal integral points (Lemma 4.12), and Riemann-Roch equals volume (Lemma 5.1). The Downstairs Theorem is proved using the dual torus bundle, local torus actions, and a Poincare-duality argument. The paper also contains a section on dual torus fibrations and a discussion of the literature.","tokens_in":19921,"tokens_out":29045,"duration_ms":276488,"significance":"If the missing connectedness hypothesis is added, the paper gives a short, mostly self-contained proof of a result previously obtained by more complicated arguments, and the Downstairs Theorem is a clean and interesting statement in integral-integral affine geometry. The reduction chain RR = vol = |BS| is elegant, and the proof of the Downstairs Theorem via local torus actions and invariant differential forms is a genuine contribution. The paper is also honest about prior attributions and about the open relation to the Markus conjecture. However, as stated, the main theorem is false without a connected-fibres hypothesis, so the central statement needs revision.","major_comments":[{"comment":"The theorem is false as stated because the definition of a regular Lagrangian fibration in Section 4 does not require connected fibres, and the proof uses connectedness essentially. Counterexample: let M = R^2/Z^2 with ω = dx∧dy, B = S^1, and π(x,y) = 2x mod 1. This is a proper surjective submersion, hence a locally trivializable fibre bundle; each fibre is the union of two Lagrangian circles. Let L be the standard prequantum line bundle whose local connection form is d − 2πi x dy. The holonomy around the fibre component at x = c is e^{2πic}, and around the component at x = c + 1/2 it is −e^{2πic}. After tensoring with any flat line bundle the two holonomies are multiplied by the same character, so they cannot both be 1; hence BS = ∅. But RR(T^2, ω) = ∫_{T^2} dx∧dy = 1. Thus Theorem 2.2 fails. The proof in Section 5 relies on Lemmas 4.4, 4.12, and 5.1, all of which require a Lagrangian torus fibration with connected fibres. The theorem and Corollaries 2.3–2.5 should be restated with connected fibres (or with the hypothesis that π is a Lagrangian torus fibration).","section":"Section 2, Theorem 2.2, with Section 4 definitions"},{"comment":"The proof is incomplete at the Mayer-Vietoris step. Exactness of the Mayer-Vietoris sequence for the invariant de Rham complex requires invariant partitions of unity, but the paper does not prove their existence for its local torus actions. This is not merely a cosmetic gap: Proposition 8.2 and hence the Downstairs Theorem depend on Theorem 7.6. The gap is repairable: one can average a smooth partition of unity on each element of a T-atlas; because on overlaps the actions differ by an automorphism of T, invariance is preserved and the averaged functions form an invariant partition of unity. The authors should include this argument explicitly.","section":"Section 7, Theorem 7.6"},{"comment":"The notation [D]∪[Z0] is nonstandard and, if read as the cup product of the Poincare duals, the first equality has the wrong sign for odd n. The subsequent equalities compute |BZ| correctly, but they are consistent only if [D]∪[Z0] is understood as the intersection number [Z0]·[D], which the preceding sentence defines to be (−1)^n|D∩Z0|. Please replace this notation by [Z0]·[D] (or explicitly state the convention) so that the sign in the proof is unambiguous.","section":"Section 8, final displayed chain in the proof of Theorem 3.1"}],"minor_comments":[{"comment":"In the displayed short exact sequence, the third term should be π^*TB, not π^*T^*B; the quotient of TM by the vertical bundle V is π^*TB. The subsequent isomorphism V ≅ π^*T^*B is correct, so this appears to be a typo.","section":"Section 5, Lemma 5.1"},{"comment":"The statement 'because the x and x′ coordinates are oriented' presupposes that the integral affine charts have been chosen to be compatible with a global orientation. After passing to the orientation double cover this is possible, but it would help to state this explicitly before the determinant argument.","section":"Section 8, Lemma 8.1"}],"recommendation":"major_revision","confidential_remarks":"The elementary counterexample in Major Comment 1 shows that the abstract and Theorem 2.2 are too broad; the authors need to add the connected-fibres hypothesis throughout. The proof machinery appears sound for connected-fibre Lagrangian torus fibrations, and the remaining gaps (Theorem 7.6 partitions of unity, sign notation) are repairable. With these amendments the paper would be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read it. First, the Upstairs Theorem (RR = |BS|) is explicitly not new; the authors say it follows from Fujita-Furuta-Yoshida and attribute it to Andersen. Second, the new meat is the Downstairs Theorem: on a compact integral-integral affine manifold, vol(B) = |BZ|. The proof, which realizes integer points as transverse intersections of two sections of the dual torus bundle and computes the Poincare dual via invariant cohomology, is genuinely clever. The reduction chain is short and each link is proved; I checked the circle, torus, and Klein bottle examples and the lemmas hold as stated.\n\nNow the soft spots, in proportion. The real problem is that Theorem 2.2 is false as written. The paper defines a Lagrangian fibration without requiring connected fibers, but the proof needs connected fibers (through Arnol'd-Liouville and its enhanced version). The stress-test counterexample is right: M = T^2 with omega = dx^dy, pi(x,y) = 2x mod 1 over S^1, and the trivial line bundle with connection d - 2pi i x dy. Each fiber has two components; their holonomies differ by a sign, so no fiber is Bohr-Sommerfeld (|BS| = 0), but RR = 1. The theorem survives if you add 'connected fibers' to the hypotheses or redefine Lagrangian fibration to include that. This is a fixable flaw, but the statement needs correction.\n\nThe other issues are minor. The final displayed chain in the proof of Theorem 3.1 has a sign error for odd-dimensional B; on S^1 it would give |BZ| = -1. The intended argument is fine once the sign is moved, but the published chain is wrong. Theorem 7.6, the invariant-representative theorem for local torus actions, is proved by a compressed Mayer-Vietoris induction that silently relies on fine-sheaf properties; the proof should be expanded, though I do not doubt the claim itself.\n\nThe Downstairs Theorem appears true, and the paper's honesty about attribution is a real asset. The authors state clearly what is theirs and what is not. The reduction of a geometric quantization statement to a purely affine-geometry counting statement is the kind of thing that makes the paper worth reading.\n\nVerdict: send it to peer review, but with a referee who will catch the connected-fiber issue and the sign typo. The revision should be quick. I would bring it to a reading group, and I would cite the Downstairs Theorem if I work on integral affine structures.","headline":"A genuinely clever reduction of RR=|BS| to a new affine-geometry counting theorem, but Theorem 2.2 is overstated: without a connected-fibers hypothesis it has a concrete counterexample.","tokens_in":20759,"tokens_out":2399,"would_cite":true,"duration_ms":25326,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53D50","53C15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that for a compact symplectic manifold with a Lagrangian torus fibration and a prequantum line bundle, the Riemann–Roch number equals the number of Bohr–Sommerfeld fibres, via a volume-equals-lattice-points theorem for…","keywords":["Lagrangian fibration","Bohr–Sommerfeld","geometric quantization","integral-integral affine structure","Riemann–Roch number","independence of polarization","local torus action","integral points"],"falsifier":"Compute $\\mathrm{vol}(B)$ and $|B_{\\mathbb{Z}}|$ for the integral-integral Klein bottle and the Kodaira–Thurston-like quotients listed in the paper: if they ever disagree, Theorem 3.1—and with it the Upstairs Theorem—is false.","tokens_in":19390,"feed_emoji":"📐","tokens_out":12498,"duration_ms":102556,"temperature":0.7,"pith_summary":"This paper proves that for a compact symplectic manifold equipped with a Lagrangian torus fibration and a prequantum line bundle, the Riemann–Roch number—the index-theoretic count of quantized states—equals the number of Bohr–Sommerfeld fibres. The proof reduces this to a theorem in affine geometry: on a compact integral-integral affine manifold, the total volume equals the number of integer lattice points. The authors' aim is a short, direct proof of a result previously known only through more elaborate theorems, and the route illustrates the independence-of-polarization phenomenon in geometric quantization. The equality matters because it connects an analytic count (a Dirac–Dolbeault index) with a geometric count (Bohr–Sommerfeld fibres), and the elementary-looking volume/lattice-point theorem turns out to be the delicate part.","feed_headline":"Riemann–Roch number equals Bohr–Sommerfeld fibre count","feed_subtitle":"Proof reduces the quantum index to a simple affine fact: total volume equals the number of integer points.","key_machinery":"The load-bearing construction is the integral-integral affine structure on the base: an affine atlas whose transition maps lie in $\\mathrm{GL}(n,\\mathbb{Z})\\ltimes\\mathbb{Z}^n$ rather than $\\mathrm{GL}(n,\\mathbb{Z})\\ltimes\\mathbb{R}^n$. Enhanced Arnol'd–Liouville charts, which trivialize the prequantized fibration together with its connection, produce such a structure, and the set of integral points $B_{\\mathbb{Z}}$ coincides with the Bohr–Sommerfeld set (Lemma 4.12). The Downstairs proof then works on the dual torus bundle $M^{\\vee}=TB/\\Lambda$: Proposition 8.2 identifies the closed $n$-form $dy^n$ as the Poincaré dual of the zero section, and the section $s(x)=(x,[-x])$ intersects the zero section exactly over $B_{\\mathbb{Z}}$. Theorem 7.6, stating that every de Rham cohomology class on a compact manifold with a local torus action has an invariant representative, is the technical device that makes the Poincaré-duality computation valid.","core_discovery":"The central claim is Theorem 2.2: for a compact symplectic manifold $(M,\\omega)$ with a regular Lagrangian fibration $\\pi\\colon M\\to B$ and a prequantization line bundle $L\\to M$, the Riemann–Roch number $\\mathrm{RR}(M,\\omega)=\\int_M \\exp(\\omega)\\,\\mathrm{Todd}(TM,J)$ equals $|\\mathrm{BS}|$, the number of Bohr–Sommerfeld fibres. The proof reduces this to Theorem 3.1: every compact integral-integral affine manifold $B$ satisfies $\\mathrm{vol}(B)=|B_{\\mathbb{Z}}|$. Lemma 4.12 supplies the bridge, showing that the prequantized fibration induces an integral-integral affine structure on $B$ whose integral points are exactly the Bohr–Sommerfeld points; Lemmas 4.4 and 5.1 then identify the Riemann–Roch number with $\\mathrm{vol}(M)$ and $\\mathrm{vol}(M)$ with $\\mathrm{vol}(B)$. The Downstairs Theorem is proved by passing to the dual torus bundle $M^{\\vee}=TB/\\Lambda$, where a natural section meets the zero section precisely over $B_{\\mathbb{Z}}$, with the intersection number computed via a Poincaré duality statement for the form $dy^n$.","pith_inferences":["The volume/lattice-point equality suggests a polytopal route: triangulate an integral-integral affine base into simple integral convex polytopes and apply exact Euler–Maclaurin formulas, a direction the paper notes but does not pursue.","The proof's reliance on connected fibres suggests that Theorem 2.2, as stated for arbitrary regular Lagrangian fibrations, is really a torus-fibration statement; a disconnected-fibre example would clarify the true scope.","Extending the argument to integrable systems with focus-focus or elliptic singularities is a natural next step, and the paper explicitly leaves that extension open.","The paper does not treat the half-form correction; a natural consequence of its method would be to test whether the equality persists, possibly with a shifted Bohr–Sommerfeld count, in that setting."],"forward_implications":["Different Lagrangian torus fibrations on the same prequantized compact symplectic manifold have the same number of Bohr–Sommerfeld points.","The Bohr–Sommerfeld count is independent of the choice of prequantum line bundle, and it is unchanged by adding a closed integral two-form pulled back from the base.","For a compact Lagrangian torus fibration, the Riemann–Roch number equals the symplectic volume, because the complexified tangent bundle admits a flat connection.","The equality $\\mathrm{RR}(M,\\omega)=|\\mathrm{BS}|$ gives a concrete instance of independence of polarization: the Dirac–Dolbeault quantization dimension equals the Bohr–Sommerfeld quantization dimension.","The Downstairs theorem, $\\mathrm{vol}(B)=|B_{\\mathbb{Z}}|$, holds as a statement about integral-integral affine manifolds in their own right, independent of any symplectic origin."],"supporting_citations":[{"why":"It supplies the Arnol'd–Liouville theorem used to establish action-angle charts for proper Lagrangian fibrations with connected fibres.","marker":"[Arn2]"},{"why":"It provides the global action-angle coordinate statement that, together with [Arn2], underlies the chart construction in Section 4.","marker":"[D2]"},{"why":"It gives Lemma 2.1, which fixes the explicit form of the transition maps between Arnol'd–Liouville models used in Lemma 4.3.","marker":"[Se]"},{"why":"It supplies the prequantization conventions and the classification of flat bundles by holonomy used in the proof of the enhanced Arnol'd–Liouville theorem.","marker":"[K]"},{"why":"It provides the global torus-action result that serves as the base case in the induction proving Theorem 7.6.","marker":"[O]"},{"why":"It justifies differentiation under the integral sign in Lemma 7.5, which shows that averaging commutes with the exterior derivative.","marker":"[R]"},{"why":"It documents the known characterization of Bohr–Sommerfeld points as integer action coordinates, which contextualizes Lemma 4.12.","marker":"[GS]"}],"fun_headline_variants":["Riemann-Roch number equals Bohr-Sommerfeld count","Counting lattice points proves Riemann-Roch equality","Volume equals lattice points: key to Riemann-Roch","Simple affine fact proves Riemann-Roch number"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on the claim that every cohomology class on a compact manifold with a local torus action has an invariant representative, together with the tacit assumption that the Lagrangian fibration has connected fibres so that Arnol'd–Liouville applies.","fun_headline_variants_meta":{"raw":{"variants":["Riemann-Roch number equals Bohr-Sommerfeld count","Counting lattice points proves Riemann-Roch equality","Volume equals lattice points: key to Riemann-Roch","Simple affine fact proves Riemann-Roch number"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000706,"raw_usage":{"total_tokens":3164,"prompt_tokens":906,"completion_tokens":2258,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":522,"completion_tokens_details":{"reasoning_tokens":2197}},"tokens_in":522,"tokens_out":2258,"duration_ms":18757,"temperature":1.0,"reasoning_tokens":2197,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T19:48:44.563915+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $\\mathrm{vol}(B)$ and $|B_{\\mathbb{Z}}|$ for the integral-integral Klein bottle and the Kodaira–Thurston-like quotients listed in the paper: if they ever disagree, Theorem 3.1—and with it the Upstairs Theorem—is false.","supporting_citations":[],"review_version":1}