Integral-integral affine geometry, geometric quantization, and Riemann-Roch

Mark Hamilton; Takahiko Yoshida; Yael Karshon

REVIEW 3 major objections 2 minor 25 references

Integral-integral affine geometry, geometric quantization, and Riemann-Roch

T0 review · 3 major / 2 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read 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…

desk verdict 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. read the letter →

arxiv 2411.10348 v1 pith:MPBXOXUP submitted 2024-11-15 math.SG

Mark Hamilton , Yael Karshon , Takahiko Yoshida This is my paper · ORCID

classification math.SG MSC 53D5053C15

keywords LagrangianfibrationBohr–Sommerfeldgeometricquantizationintegral-integralaffinestructureRiemann–Rochnumberindependenceofpolarizationlocaltorusactionintegralpoints

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

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.

What carries the argument

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.

What would settle it

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.

Watch

Extended reading notes

Core claim

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 $dy^n$.

Load-bearing premise

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.

Editorial extensions

If this is right

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.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

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.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the 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.

Now 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.

The 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.

The 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.

Verdict: 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.

Referee Report

3 major / 2 minor

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.

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 (3)

[Section 2, Theorem 2.2, with Section 4 definitions] 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 7, Theorem 7.6] 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 8, final displayed chain in the proof of Theorem 3.1] 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.

minor comments (2)

[Section 5, Lemma 5.1] 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 8, Lemma 8.1] 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.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the Upstairs and Downstairs theorems are proved independently, and the self-citations are contextual rather than load-bearing.

full rationale

The paper's derivation chain is not circular. Lemma 5.1 proves RR(M)=vol(M) by showing the Todd class is trivial because (TM,J) is isomorphic to the pullback of a flat bundle; Lemma 4.4 proves vol(M)=vol(B) by direct computation of Liouville measure in Arnol'd-Liouville charts; Lemma 4.12 establishes |BS|=|B_Z| by computing holonomy e^{2πix_j} in enhanced Arnol'd-Liouville charts. The remaining step, the Downstairs Theorem vol(B)=|B_Z|, is proved in Section 8 by intersecting the affine-lattice section with the zero section of M^∨ and using Proposition 8.2, whose proof via Theorem 7.6 depends on an averaging/Mayer-Vietoris argument and the external references [O] and [R], not on the target result. No parameter is fitted, no quantity is defined in terms of its predicted value, and the known prior result [FFY] is cited only for attribution and context. Self-citations (e.g., [FFY], [Y], [KSW], [HM]) are not used as premises of the proof. The disconnected-fibre counterexample and any fine-sheaf gap in Theorem 7.6 are correctness risks, not circularity.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

Pure mathematics paper: no free parameters fitted to data and no invented physical entities. The central claim rests on standard background theorems in symplectic geometry and topology (Arnol'd-Liouville, classification of flat bundles by holonomy, invariant cohomology for compact group actions, the Spinc index interpretation of the Riemann-Roch number) and on the domain hypothesis that the Lagrangian fibration has connected fibers. The genuinely new definitions (integral-integral affine structure, enhanced Arnol'd-Liouville charts, local torus actions, the affine lattice section) are proved to exist and to have the stated properties within the paper.

assumptions (5)

standard math Arnol'd-Liouville theorem: every proper Lagrangian fibration with connected fibers admits action-angle coordinates locally (Theorem 4.2, citing [Arn2, §49-50], [MM], [D2]).
Basis for the integral affine atlas on the base B (Lemma 4.4) and for the enhanced local model of a prequantized fibration (Theorem 4.8), Section 4.
standard math A flat Hermitian line bundle over a torus is determined up to isomorphism by its holonomy; flat bundles over a contractible base are trivial ([K, Remark 1.12.1]).
Used in Theorem 4.8 (Enhanced Arnol'd-Liouville) to trivialize the prequantization bundle over each contractible chart.
standard math For a compact connected Lie group action on a manifold, invariant forms compute de Rham cohomology ([O, §9]).
Base case of the induction proving Theorem 7.6 (invariant representatives for local torus actions), which underpins Proposition 8.2.
domain assumption The fibration in Theorem 2.2 is assumed to have connected fibers (a torus fibration), although the theorem statement only says 'Lagrangian fibration'.
Lemmas 4.4, 4.12, and 5.1 are stated for Lagrangian torus fibrations; the proof of Theorem 2.2 in Section 5 silently upgrades the hypothesis.
standard math The Riemann-Roch number of (M,ω) is the index of the Spinc Dirac operator, computed by ∫_M exp(ω) Todd(TM,J) (Section 2, citing [D1]).
Gives definition (2.1) of RR and its interpretation as a quantization dimension; not proved within the paper.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Integral-integral affine geometry, geometric quantization, and Riemann-Roch." pith.science (2026). https://pith.science/paper/MPBXOXUP

@misc{pith2026241110348,
  author       = {Pith},
  title        = {Pith review of: Integral-integral affine geometry, geometric quantization, and Riemann-Roch},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MPBXOXUP}},
  note         = {Machine review of arXiv:2411.10348}
}

read the original abstract

We give a simple proof that, for a pre-quantized compact symplectic manifold with a Lagrangian torus fibration, its Riemann-Roch number coincides with its number of Bohr-Sommerfeld fibres. This can be viewed as an instance of the "independence of polarization" phenomenon of geometric quantization. The base space for such a fibration acquires a so-called integral-integral affine structure. The proof uses the following simple fact, whose proof is trickier than we expected: on a compact integral-integral affine manifold, the total volume is equal to the number of integer points.

Figures

Figures reproduced from arXiv: 2411.10348 by the authors.

**Figure 1.** The integral lattice and the affine lattice 6.3. Remark. In this section, given any integral-integral affine manifold B, we constructed its affine lattice Λaff ⊂ T B, and we viewed it as a section of the Lie torus bundle T B/Λ. In §4, we started with a prequantized Lagrangian torus fibration L → M → B, and we obtained from it an integralintegral structure on B. When B is obtained in this way, we can identify the Li… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 24 canonical work pages

[1]

J rgen E.\ Andersen, Geometric quantization of symplectic manifolds with respect to reducible non-negative polarizations, Commun.\ Math.\ Phys.\ 183, 401--421 (1997)

work page 1997
[2]

V.I.\ Arnol'd, Mathematical methods of classical mechanics, Second Edition, Springer, New York, 1989

work page 1989
[3]

Diff Geom

Cheeger and Gromov, Collapsing Riemannian manifolds while keeping their curvature bounded I, J. Diff Geom. 23 (1986), 309--346

work page 1986
[4]

J.\ J.\ Duistermaat, The heat kernel Lefschetz fixed point formula for the spin-c Dirac operator, Birkh\"auser, Boston, 2011

work page 2011
[5]

33 (1980), no.\ 6, 687--706

J.\ J.\ Duistermaat, On Global Action-Angle Coordinates, Comm.\ Pure Appl.\ Math. 33 (1980), no.\ 6, 687--706

work page 1980
[6]

Rui Loja Fernandes and Maarten Mol, K\"ahler metrics and toric Lagrangian fibrations, arXiv:2401.02910 [math.DG]

work page Pith review arXiv
[7]

Hajime Fujita, Mikio Furuta, and Takahiko Yoshida, Torus fibrations and localization of index I -- polarization and acyclic fibrations, J.\ Math.\ Sci.\ Univ.\ Tokyo 17 (2010), 1--26

work page 2010
[8]

Viktor Ginzburg, Victor Guillemin, and Yael Karshon Moment Maps, Cobordisms, and Hamiltonian Group Actions, Mathematical Surveys and Monographs, 98 , American Mathematical Society, Providence, RI, 2002

work page 2002

Show all 25 references

[9]

William M.\ Goldman, Geometric structures on manifolds, Graduate Studies in Mathematics 227, Amer.\ Math.\ Soc., 2022

2022
[10]

IHES 56 (1983), 213--307

Gromov, Volume and bounded cohomology, Publ. IHES 56 (1983), 213--307

1983
[11]

Integrable systems and algebraic geometry (Kobe/Kyoto, 1997), 156-193, World Sci

Mark Gross, Special Lagrangian Fibrations I: Topology. Integrable systems and algebraic geometry (Kobe/Kyoto, 1997), 156-193, World Sci. Publ., River Edge, NJ, 1998

1997
[12]

Guillemin and S

V. Guillemin and S. Sternberg, The Gel'fand--Cetlin system and quantization of the complex flag manifolds, J. Funct.\ Anal. 52 (1983), no.\ 1, 106--128

1983
[13]

M.\ D.\ Hamilton, Classical and quantum monodromy via action-angle variables, Journal of Geometry and Physics, 115 (May 2017), 37--44

2017
[14]

M. D. Hamilton and Zoe McIntyre, Quantum monodromy in the Kodaira-Thurston manifold, in progress
[15]

B.\ Kostant, Quantization and unitary representations. I. Prequantization, Lectures in modern analysis and applications, III, Lecture Notes in Mathematics, 170 , pp. 87--208
[16]

Yael Karshon, Shlomo Sternberg, and Jonathan Weitsman, Exact Euler-Maclaurin formulas for simple polytopes, Adv.\ Appl.\ Math.\ 39 (2007), 1--50

2007
[17]

L.\ Markus and K.R.\ Meyer, Generic Hamiltonian systems are neither integrable nor ergodic, Mem.\ A.M.S.\ 144, 1974

1974
[18]

Maarten Mol, On the classification of multiplicity-free Hamiltonian actions by regular proper symplectic groupoids, arXiv:2401:00570 [math.SG]
[19]

Arkadi L.\ Onishchik, Topology of transitive transformation groups, Barth Verlagsgesellschaft, Barth 1994

1994
[20]

W.\ Rudin, Principles of Mathematical Analysis, third edition, McGraw-Hill, 1976

1976
[21]

Daniele Sepe, Topological classification of Lagrangian fibrations, Journal of Geometry and Physics 60, 341--351 (2010)

2010
[22]

Jedrzej \'Sniatycki, On cohomology groups appearing in geometric quantization, in: Lecture Notes in Math.\ 570, Springer-Verlag, Berlin, 1977, pp. 46--66

1977
[23]

Andrei Tyurin, Geometric quantization and mirror symmetry, arXiv:math/9902027v1

arXiv
[24]

N.M.J.\ Woodhouse, Geometric Quantization, 2nd edition, Clarendon Press, Oxford, 1991

1991
[25]

Takahiko Yoshida, Local torus actions modeled on the standard representation, Advances in Mathematics, 227 5 (Aug 2011), 1914--1955

2011

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

J rgen E.\ Andersen, Geometric quantization of symplectic manifolds with respect to reducible non-negative polarizations, Commun.\ Math.\ Phys.\ 183, 401--421 (1997)

work page 1997

[2] [2]

V.I.\ Arnol'd, Mathematical methods of classical mechanics, Second Edition, Springer, New York, 1989

work page 1989

[3] [3]

Diff Geom

Cheeger and Gromov, Collapsing Riemannian manifolds while keeping their curvature bounded I, J. Diff Geom. 23 (1986), 309--346

work page 1986

[4] [4]

J.\ J.\ Duistermaat, The heat kernel Lefschetz fixed point formula for the spin-c Dirac operator, Birkh\"auser, Boston, 2011

work page 2011

[5] [5]

33 (1980), no.\ 6, 687--706

J.\ J.\ Duistermaat, On Global Action-Angle Coordinates, Comm.\ Pure Appl.\ Math. 33 (1980), no.\ 6, 687--706

work page 1980

[6] [6]

Rui Loja Fernandes and Maarten Mol, K\"ahler metrics and toric Lagrangian fibrations, arXiv:2401.02910 [math.DG]

work page Pith review arXiv

[7] [7]

Hajime Fujita, Mikio Furuta, and Takahiko Yoshida, Torus fibrations and localization of index I -- polarization and acyclic fibrations, J.\ Math.\ Sci.\ Univ.\ Tokyo 17 (2010), 1--26

work page 2010

[8] [8]

Viktor Ginzburg, Victor Guillemin, and Yael Karshon Moment Maps, Cobordisms, and Hamiltonian Group Actions, Mathematical Surveys and Monographs, 98 , American Mathematical Society, Providence, RI, 2002

work page 2002