Constructibility of momentum maps and linear variation for singular symplectic reduced spaces

Maarten Mol

REVIEW 3 major objections 5 minor 1 cited by

Constructibility of momentum maps and linear variation for singular symplectic reduced spaces

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper proves that the image of a Hamiltonian momentum map carries a natural integral affine stratification over which the map is an equivariantly locally trivial fibration, and that cohomology classes of reduced symplectic forms vary l

desk verdict A serious, technically careful preprint that proves constructibility of proper momentum maps and extends Duistermaat–Heckman to singular values; the main theorems are new, and the one real proof gap I found is minor and repairable. read the letter →

arxiv 2508.21284 v1 pith:C6TLK6ZP submitted 2025-08-29 math.SG math.DG

Maarten Mol This is my paper · ORCID

classification math.SGmath.DG MSC 53D2053D1758A35

keywords momentummapsingularsymplecticreductionintegralaffinestratificationlinearvariationequivariantlocaltrivialityconstructiblesheavesHamiltoniangroupactionsproperquasi-symplecticgroupoids

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 singular symplectic reduction has the same qualitative regularity as the regular case, once the momentum-map image is cut into the right pieces. For a proper Hamiltonian action of a compact group, the image admits a canonical integral affine stratification—flat, lattice-structured pieces—such that over each piece the reduced spaces form a locally trivial family. Inside any piece, the cohomology class of the reduced symplectic form varies linearly with the momentum value, and the symplectic volume is polynomial, generalizing the classical linear variation theorem beyond regular values. The strata are not chosen ad hoc: their tangent spaces are determined by the isotropy data of the points mapping to a given value. The same statement is proved for actions of proper quasi-symplectic groupoids.

What carries the argument

The central object is the canonical Hamiltonian stratification, whose strata are connected components of sets where the isotropy pairs are conjugate; its images form the piecewise-affine cover of the momentum image. The workhorse is an abstract proposition that takes any locally finite piecewise-affine cover of a set in a vector space and produces a unique affine stratification by intersecting tangent directions of the covering pieces; Whitney (b)-regularity comes along when the set is locally closed. The load-bearing identity is the tangent formula: the tangent space to the stratum through a value x is the intersection over all points p mapping to x of (g_p^0)^{G_x}, the directions left fix

What would settle it

Check Proposition 1.13 on a compact Hamiltonian action whose momentum image has a non-convex or locally polyhedral part: if some canonical Hamiltonian stratum has momentum image that is not open in its affine hull, or if the momentum map restricted to that stratum drops rank, then the cover is not piecewise-affine and the main theorem cannot hold. A more local falsifier is the tangent formula itself: exhibit two points in the same connected affine subset of the momentum image for which the intersection ∩_{p∈J^{-1}(x)}(g_p^0)^{G_x} has different dimension, since a single affine stratum with the

Watch

Extended reading notes

Core claim

The paper's central claim is that the image of a proper Hamiltonian momentum map carries a natural integral affine stratification—strata are flat pieces with lattices—uniquely determined by the tangent formula T_xσ_+ = ∩_{p∈J^{-1}(x)}(g_p^0)^{G_x}, and over each stratum the momentum map is a G-equivariantly locally trivial fibration. Consequently, comparing reduced spaces within a stratum by Gauss–Manin parallel transport, the cohomology class of the reduced symplectic form varies linearly with the momentum value; this extends the classical regular-value linear variation theorem to singular values. The proof obtains the strata by applying an abstract lemma that turns a piecewise-affine cover

Load-bearing premise

The construction depends on the claim that every stratum of the canonical Hamiltonian stratification has momentum image open in an affine subspace, with the momentum map filling that affine piece with derivative of full rank everywhere; if that fails at any stratum, the piecewise-affine cover and hence the stratification do not exist.

Editorial extensions

If this is right

For each stratum, the momentum map is a locally trivial fibration over that stratum, so the homeomorphism type of every reduced space is constant along the stratum.
The cohomology class of the reduced symplectic form is affine in stratum coordinates under Gauss–Manin parallel transport; this extends the classical regular-value linear variation theorem to singular values.
Symplectic volumes of reduced spaces form a polynomial function of degree at most half the reduced dimension on every stratum.
The derived push-forwards of the constant sheaf along the transverse momentum map are constructible, so paths leaving a stratum induce canonical maps between cohomology groups of reduced spaces; a special case is a local invariant-cycle statement for regular-to-singular transitions.
The statements survive extension to disconnected compact groups and to proper quasi-symplectic groupoids, including group-valued momentum maps as an example.

Reading between the lines

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

Beyond the paper, since the stratification is determined by isotropy data alone, it should be computable in concrete torus examples, yielding an algorithmic chamber decomposition of a momentum polytope; the paper leaves general convex-geometric descriptions as work in progress.
Beyond the paper, the flat Gauss–Manin bundle over each stratum could carry nontrivial monodromy; computing it for a non-contractible stratum would test how much of the linear-variation structure is governed by the fundamental group of the stratum.
Beyond the paper, Morita invariance of the stratification suggests that the piecewise-affine volume polynomials are transverse invariants of Hamiltonian actions, so Morita-equivalent actions should exhibit identical linear-variation data.
Beyond the paper, combining polynomial volumes with localization formulas may yield wall-crossing relations for symplectic volumes across adjacent strata, which the paper does not address.

Share X Bluesky LinkedIn Reddit HN

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

The two main results are genuinely new: an integral affine stratification of the transverse image of a proper momentum map, with the momentum map equivariantly locally trivial over each stratum, and a linear variation theorem for the symplectic class on reduced spaces at singular values. The regular case reduces to Duistermaat–Heckman, but the singular case is new. The stratification is determined by isotropy data and is not fitted; that gives the paper real substance.

The proof strategy is transparent and mostly sound: use the canonical Hamiltonian stratification from the author's published work [24], prove that stratum images form a piecewise-affine cover via the MGS normal form, then build local trivializations by lifted flows. The Sjamaar de Rham model is used carefully, and the smooth simplex setup for the pairing is a useful piece of work in its own right. The paper is also honest about what is deferred: the convex-geometric description of the strata for connected compact groups is relegated to a separate note [25].

The stress-test flags a real gap in Lemma 1.18: the proof defines the lifted vector field only on S∩σ, while the claim requires a Gx-invariant field on the slice S. As written, that step is missing. But it is readily fixed: the constant vector field on σ+ is translation by a vector β, and β lies in the Gx-fixed subspace, so defining the same constant translation on all of S gives a Gx-invariant extension. Then the existing partition-of-unity patching works, and completeness of the flows is already handled by Lemma 1.19. So I read this as a minor proof gap, not a load-bearing flaw. Proposition 1.13 is essential and depends on [24], but the reduction to the MGS model looks sound; I did not find a counterexample.

The groupoid extension in Section 3.2 is more speculative and leans on the preprint [7], but it is clearly marked and the Morita-invariance argument is plausible. The citation pattern is fine: [24] is the author's own published tool, and [25] is explicitly a separate note.

Who should read this: symplectic geometers working on reduction, Duistermaat–Heckman theory, or stratified symplectic spaces. It deserves a serious referee. I would send it to peer review, and I would cite it in my own work.

Referee Report

3 major / 5 minor

Summary. The paper claims a constructibility theorem for proper momentum maps of Hamiltonian actions of compact Lie groups and, more generally, of proper quasi-symplectic groupoids. It constructs a natural integral affine stratification of the image of the transverse momentum map such that, over each stratum, the momentum map is an equivariantly locally trivial fibration. It then extends the Duistermaat–Heckman linear variation theorem to singular reduced spaces, proves polynomiality of symplectic volumes on strata, derives constructibility of the direct images of the constant sheaf, and gives an invariant-cycle-type statement for regular values. The proofs use standard machinery: canonical Hamiltonian stratifications, the Marle–Guillemin–Sternberg normal form, Sjamaar's de Rham model, and flows of lifted vector fields.

Significance. If the main theorems hold, the paper gives a substantial and natural extension of Duistermaat–Heckman theory to singular values, with a canonical stratification determined by isotropy data. The result is well-motivated and the overall strategy is convincing. The detailed treatment of Sjamaar's de Rham model and the extension to proper quasi-symplectic groupoids are valuable. However, the current version has a genuine proof gap in the local triviality argument, which is load-bearing for Theorem 1 and hence for the linear variation theorem and its corollaries.

major comments (3)

[§1.3.1, Lemma 1.18] The proof of Lemma 1.18 contains a step that is not justified. After fixing a slice S, the proof says 'define X^σ := X|_{S∩σ}' and then treats X^σ as a G_x-invariant vector field on S. But S∩σ is in general a proper affine subspace of S, so this restriction is not a vector field on S, nor on any neighbourhood of J^{-1}(S∩σ) in J^{-1}(S). To satisfy the lemma one must extend the constant field from S∩σ to a G_x-invariant field on S, patch the resulting local lifts over σ by a G-invariant partition of unity, and then control completeness of the patched lifts on the cube C_+ used in Theorem 1.15. None of these steps is written down. Since the flows constructed from these lifts are exactly what defines Φ and φ in Theorem 1.15, the equivariant local triviality asserted in Theorem 1 is not fully proved. This is a proof gap that appears repairable, not a detected counterexample.
[§1.3.1, proof of Theorem 1.15] Even if Lemma 1.18 is repaired, the proof of Theorem 1.15 invokes Lemma 1.19 to conclude that the flows of the lifted fields bX_i are defined on J^{-1}(σ) for the required time interval. Lemma 1.19 requires the base flow to be f-related via a proper map. For the pair (bX^σ_i, X_i), the relevant map is q_+ restricted to σ, whose properness is not established. For the pair (bX_i, bX^σ_i), the map J: J^{-1}(σ) → σ is proper only if σ is regarded as a locally closed subset of g^*; this is likely true, but the argument is not given. The proof also asserts the smooth extension of Φ^{-1} on a set that is 'open around J^{-1}(W)' using the tube lemma; this requires a compactness statement for J^{-1}(W) that is not explicitly proved. These are all local-triviality details that must be supplied before Theorem 1.15 can be regarded as complete.
[§3.2.3, proof of Theorem 3.10] The groupoid extension of Theorem 1 is obtained by Morita equivalence and by transporting the local trivializations of Theorem 3.1. This transport relies on the existence of G-equivariant smooth extensions of Φ_{g^*} and Φ_{g^*}^{-1} as in Theorem 1.15. Since the Lie-group case is not fully proved at the level of Lemma 1.18, the gap propagates to the groupoid statement. Once the Lie-group local triviality is repaired, the Morita transport argument appears sound, but in the present form Theorem 3.10 inherits the same unproved completeness and extension steps.

minor comments (5)

[§2.1.2] There is a typographical error: 'Theorem 2.10]' should be 'Theorem 2.10'.
[§1.3.1, Lemma 1.18 proof] The notation p is overloaded: p denotes a point in S and also the H-equivariant splitting p : h^* → g^* in the MGS model. This makes the proof harder to follow. Using a different letter for the splitting would improve readability.
[Introduction, after Example 1] The sentence 'A proof of this will be not be given here, but in a separate note [25]' is an explicit pointer to an in-progress note. This is fine as a statement of scope, but the paper should make clear that the convex-geometric chamber decomposition is not used in the proofs of the main theorems.
[§1.1.1, Example 1.5] The example is useful but the displayed definition of X and the covering A is dense; a small clarifying sentence on why local finiteness of S(A) fails there would help the reader.
[§3.2.4] In the proof of the extension of Corollary 2(a), the phrase 'the homeomorphism h_P restricts to a homeomorphism between the respective images' and the notation for the open V are not always consistent with the diagram in Theorem 3.10. A minor rewriting would improve clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the stratification and linear-variation proofs are constructed from prior published tools, not defined into existence.

full rationale

I walked the derivation chain of Theorems 1 and 2. The stratification SHam(Delta_+) is not fitted or defined as the conclusion: it is produced by Proposition 1.1 from the piecewise-affine cover AHam = {J_+(Sigma) | Sigma in SHam(S)}, and Proposition 1.13 proves the needed affine-open/affine-hull/submersion facts from the MGS normal form together with [24, Prop. 2.56, Lemma 2.57]. Those cited results are prior published work by the same author, but they are parameter-free, do not contain the present theorem, and are used as a normal-form tool rather than as an assertion of the target statement. The uniqueness of the stratification is not imported from a self-citation; it follows from Lemma 1.6 by the tangent-space argument. The local triviality theorem (Theorem 1.15) is proved by explicitly constructing lifts (Lemma 1.18), proving flow completeness (Lemmas 1.19-1.21), and writing down the maps Phi and phi; no equation in this proof equals its input by construction. Theorem 2 is not definitional: Prop. 2.12 proves smoothness of the cohomology-class section, Lemma 2.13 gives an independent criterion for linear variation, and the proof computes the covariant derivative as the class of the Lie derivative of the symplectic form on the reduced space and constructs an explicit primitive using the momentum-map identity. The symplectic-volume polynomiality (Cor. 3) is then a consequence, not a renaming. The only self-citations are [24] (a published stratification theory used as a tool) and [25] (an in-progress note explicitly not used for any proof in this paper). A possible gap in the written details of Lemma 1.18--the extension/patching of the local slice lifts--is a proof-completeness concern, not a circularity: nothing in that lemma reduces the theorem to its assumptions. Therefore no circular step meets the evidence standard, and the appropriate score is 0.

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

No free parameters or invented entities. The paper's contribution is structural: it derives a stratification and linear variation from the input data (isotropy groups and momentum map geometry) rather than from fitted constants. The main imported axioms are prior results in symplectic geometry, including the author's own published stratification work [24].

assumptions (5)

domain assumption The canonical Hamiltonian stratification SHam(S) from Mol 2024 [24] exists and has the stated smooth structure, with J restricting to constant-rank submersions from strata to coadjoint-orbit-type strata.
This is the foundational input for the affine cover AHam in Section 1.2.4. It is published work by the same author, used as a black box; if it were flawed, Theorem 1 would not follow.
standard math The Marle-Guillemin-Sternberg normal form theorem for Hamiltonian spaces at arbitrary points.
Used in the proofs of Proposition 1.13, Lemma 1.18, and Lemma 2.11 to reduce local computations to a standard local model.
standard math Sjamaar's de Rham theorem and Stokes theorem for singular symplectic reduced spaces [29], including at arbitrary coadjoint orbits.
Basis for representing reduced symplectic cohomology classes and for the polynomial volume argument in Corollary 3.
domain assumption The linearization theorem for proper quasi-symplectic groupoids and the existence of integral affine structures on their orbit-type strata from Crainic-Fernandes-Martinez Torres [7, arXiv:2504.06447].
Required for the groupoid extension in Section 3.2. This source is a recent preprint, not yet peer-reviewed, which adds fragility to that portion of the paper.
standard math Standard sheaf-theoretic tools: proper base change, Vietoris-Begle, and G-invariant partitions of unity on manifolds.
Used in Corollary 2, Proposition 1, and local triviality constructions.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Constructibility of momentum maps and linear variation for singular symplectic reduced spaces." pith.science (2026). https://pith.science/paper/C6TLK6ZP

@misc{pith2026250821284,
  author       = {Pith},
  title        = {Pith review of: Constructibility of momentum maps and linear variation for singular symplectic reduced spaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C6TLK6ZP}},
  note         = {Machine review of arXiv:2508.21284}
}

read the original abstract

In this paper we show that the transverse image of the momentum map of a Hamiltonian Lie group action admits a natural integral affine stratification with the property that over each stratum the momentum map is an equivariantly locally trivial fibration, provided the group is compact and the momentum map is proper. Using this we extend the linear variation theorem of Duistermaat and Heckman to singular values of the momentum map by showing that the cohomology classes of the symplectic forms on the reduced spaces at values within a stratum vary linearly. We also point out an instance of an invariant cycle theorem for momentum maps. Finally, we extend all of the above to Hamiltonian actions of proper quasi-symplectic groupoids.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Comparison of K\"{a}hler quotients of torus actions
math.AG 2026-07 conditional novelty 6.0 of 10

Kähler quotients of compact Hamiltonian torus manifolds undergo explicit bimeromorphic transformations across walls, enabling comparison of Kähler classes and computation of Riemann–Roch numbers of singular quotients.

Reference graph

Works this paper leans on

32 extracted references · 30 canonical work pages · cited by 1 Pith paper

[24]

M. Mol. Stratification of the transverse momentum map. Compos. Math., 160(3):518–585, 2024

work page 2024
[25]

Mol and D

M. Mol and D. Sepe. On the chamber decomposition of the momentum polytope. In progress

work page
[7]

Crainic, R.L

M. Crainic, R.L. Fernandes, and D. Mart ´ ınez Torres. Poisson manifolds of compact types. arXiv:2504.06447, 2025

arXiv 2025
[1]

Alekseev, A

A. Alekseev, A. Malkin, and E. Meinrenken. Lie group valued moment maps. J. Differential Geom., 48(3):445– 495, 1998

work page 1998
[2]

M. Atiyah. Convexity and commuting hamiltonians. Bulletin of the London mathematical society , 1982

work page 1982
[3]

Behrend, G

K. Behrend, G. Ginot, B. Noohi, and P. Xu. String topology for stacks. Ast´ erisque, (343):xiv+169, 2012

work page 2012
[4]

G.E. Bredon. On the continuous image of a singular chain complex. Pacific J. Math. , 15:1115–1118, 1965

work page 1965
[5]

Bursztyn, M

H. Bursztyn, M. Crainic, A. Weinstein, and C. Zhu. Integration of twisted Dirac brackets. Duke Mathematical Journal, 2004

work page 2004

Show all 32 references

[6]

Crainic, R.L

M. Crainic, R.L. Fernandes, and D. Mart ´ ınez Torres. Regular Poisson manifolds of compact types (PMCT 2). Ast´ erisque, (413):viii + 154, 2019

2019
[8]

Crainic and J.N

M. Crainic and J.N. Mestre. Orbispaces as differentiable stratified spaces. Letters in Mathematical Physics , 2017

2017
[9]

Duistermaat and G.J

J.J. Duistermaat and G.J. Heckman. On the variation in the cohomology of the symplectic form of the reduced phase space. Invent. Math. , 69(2):259–268, 1982

1982
[10]

Duistermaat and J.A.C

J.J. Duistermaat and J.A.C. Kolk. Lie groups. Springer-Verlag, 2000

2000
[11]

Gibson, K

C.G. Gibson, K. Wirthm¨ uller, A.A. de Plessis, and E.J.N. Looijenga.Topological stability of smooth mappings. Springer-Verlag, 1976

1976
[12]

Godement

R. Godement. Topologie alg´ ebrique et th´ eorie des faisceaux, volume No. 13 of Publications de l’Institut de Math´ ematiques de l’Universit´ e de Strasbourg [Publications of the Mathematical Institute of the University of Strasbourg]. Hermann, Paris, 1958. Actualit´ es Scien...

1958
[13]

Goresky and R

M. Goresky and R. MacPherson. Stratified Morse theory , volume 14 of Ergebnisse der Mathematik und ihrer Grenzgebiete (3) [Results in Mathematics and Related Areas (3)] . Springer-Verlag, Berlin, 1988

1988
[14]

Grothendieck

A. Grothendieck. Sur quelques points d’alg` ebre homologique. Tohoku Math. J. (2) , 9:119–221, 1957

1957
[15]

Guillemin, E

V. Guillemin, E. Lerman, and S. Sternberg. Symplectic fibrations and multiplicity diagrams . Cambridge Uni- versity Press, Cambridge, 1996

1996
[16]

Guillemin and S

V. Guillemin and S. Sternberg. Convexity properties of the moment mapping. Inventiones mathematicae, Springer-Verlag, 1982

1982
[17]

Haefliger

A. Haefliger. Homotopy and integrability. InManifolds–Amsterdam 1970 (Proc. Nuffic Summer School), volume Vol. 197 of Lecture Notes in Math. , pages 133–163. Springer, Berlin-New York, 1971

1970
[18]

Hilgert, K.-H

J. Hilgert, K.-H. Neeb, and W. Plank. Symplectic convexity theorems and coadjoint orbits. Compositio Math., 94(2):129–180, 1994

1994
[19]

W.-Y. Hsiang. Cohomology theory of topological transformation groups , volume Band 85 of Ergebnisse der Mathematik und ihrer Grenzgebiete [Results in Mathematics and Related Areas] . Springer-Verlag, New York- Heidelberg, 1975

1975
[20]

F. Kirwan. Convexity properties of the moment mapping, iii. Inventiones mathematicae, Springer-Verlag, 1984

1984
[21]

Lerman and R

E. Lerman and R. Sjamaar. Stratified symplectic spaces and reduction. Annals of Mathematics , 1991

1991
[22]

J.N. Mather. Notes on topological stability. Bull. Amer. Math. Soc. (N.S.) , 49(4):475–506, 2012

2012
[23]

M. Mol. On the classification of multiplicity-free Hamiltonian actions by regular proper symplectic groupoids. arXiv:2401.00570, 2024

2024 arXiv
[26]

R.S. Palais. When proper maps are closed. Proc. Amer. Math. Soc. , 24:835–836, 1970. 42 MAARTEN MOL UNIVERSITY OF TORONTO

1970
[27]

Pflaum, H

M.J. Pflaum, H. Posthuma, and X. Tang. Geometry of orbit spaces of proper Lie groupoids. Journal f¨ ur die reine und angewandte Mathematik , 2014

2014
[28]

D. Quillen. The spectrum of an equivariant cohomology ring: I. Annals of Mathematics , 94(3):549–572, 1971

1971
[29]

R. Sjamaar. A de Rham theorem for symplectic quotients. Pacific J. Math. , 220(1):153–166, 2005

2005
[30]

J. Watts. The orbit space and basic forms of a proper Lie groupoid. In Trends in Mathematics, Research Perspectives: Proceedings of the 12th ISAAC Congress, Aveiro, Portugal . Birkh¨ auser, 2019

2019
[31]

P. Xu. Momentum maps and Morita equivalence. J. Differential Geometry , 2004

2004
[32]

N.T. Zung. Proper groupoids and momentum maps: linearization, affinity, and convexity. Annales scientifiques de l’ ´Ecole Normale Sup´ erieure, 2006. University of Toronto, Canada E-mail address : maarten.mol.math@gmail.com

2006

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [24]

M. Mol. Stratification of the transverse momentum map. Compos. Math., 160(3):518–585, 2024

work page 2024

[2] [25]

Mol and D

M. Mol and D. Sepe. On the chamber decomposition of the momentum polytope. In progress

work page

[3] [7]

Crainic, R.L

M. Crainic, R.L. Fernandes, and D. Mart ´ ınez Torres. Poisson manifolds of compact types. arXiv:2504.06447, 2025

arXiv 2025

[4] [1]

Alekseev, A

A. Alekseev, A. Malkin, and E. Meinrenken. Lie group valued moment maps. J. Differential Geom., 48(3):445– 495, 1998

work page 1998

[5] [2]

M. Atiyah. Convexity and commuting hamiltonians. Bulletin of the London mathematical society , 1982

work page 1982

[6] [3]

Behrend, G

K. Behrend, G. Ginot, B. Noohi, and P. Xu. String topology for stacks. Ast´ erisque, (343):xiv+169, 2012

work page 2012

[7] [4]

G.E. Bredon. On the continuous image of a singular chain complex. Pacific J. Math. , 15:1115–1118, 1965

work page 1965

[8] [5]

Bursztyn, M

H. Bursztyn, M. Crainic, A. Weinstein, and C. Zhu. Integration of twisted Dirac brackets. Duke Mathematical Journal, 2004

work page 2004