REVIEW 3 major objections 5 minor 8 references
Poisson structure on the moduli spaces of sheaves of pure dimension one on a surface
T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read The symplectic leaves of the Poisson moduli space of pure dimension-one sheaves are exactly the fibers of the map sending each sheaf to the intersection of its Fitting support with the degeneracy divisor.
desk verdict A probably correct and clean leaf computation for Tyurin–Bottacin Poisson moduli, but the theorem as stated needs a connectedness argument before it is fully proved. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central mechanism is the factorization of the Poisson map into a surjection onto a symplectic subspace $T_0$. For a Poisson structure $\sigma:T^*\to T$, $T_0=\sigma(T^*)$ carries the induced symplectic form, and the paper identifies this $T_0$ with the tangent space to the fiber of $\phi_0$. That identification is made through the local-to-global spectral sequence for Ext: the bottom row of the paper's diagram (4.1), $H^1(\mathcal{H}om(E,E))\to\operatorname{Ext}^1(E,E)\to H^0(\mathcal{E}xt^1(E,E))$, is matched with the tangent sequence of the map $f_0$ sending $E$ to its Fitting support in the Hilbert scheme, while the top-right term $H^0(N_{C/S}(-D))$ is matched with the tangent space to the fibers of the intersection map $g$. The Fitting support $F(E)$ is the subscheme cut out by the zeroth Fitting ideal, and for $E=i_*L$ with smooth $C$ it is simply $C$.
What would settle it
Exhibit a fiber of $g:Q_0\to\operatorname{Sym}^{d'}(D)$ with more than one connected component; then the corresponding fiber of $\phi_0$ is disconnected and cannot be a single symplectic leaf under the standard definition, contradicting Theorem 5.1 as stated.
Extended reading notes
Core claim
For sheaves $E=i_*L$ whose Fitting support $F(E)$ is a smooth curve $C$ meeting $D$ in finitely many points, define $\phi_0(E)=C\cap D$ as a point of the symmetric product $\operatorname{Sym}^{d'}(D)$, where $d'=C\cdot D$. The paper's theorem states that the fibers of $\phi_0$ are the symplectic leaves of the Poisson structure on the open subset $M'_0$. The proof shows that at each such $E$, the tangent space to the fiber equals the symplectic subspace $T_0$ of $\operatorname{Ext}^1(E,E)$ obtained by factoring the Poisson map $\sigma:\operatorname{Ext}^1(E,E\otimes K_S)\to\operatorname{Ext}^1(E,E)$. In particular, the leaf direction is characterized infinitesimally by preserving both the support curve and its intersection with the degeneracy divisor.
Load-bearing premise
The argument assumes that the two maps defining the fiber—the map from sheaves to their support curves and the map from support curves to their intersections with $D$—are smooth enough for tangent-space sequences to describe the fibers, and that each resulting fiber is connected so it is a single maximal symplectic leaf rather than several.
Editorial extensions
If this is right
- For every point $E=i_*L$ with smooth support $C$ and $C\cap D$ finite, the symplectic leaf through $E$ lies inside the fiber of $\phi_0$, consisting of sheaves whose support meets $D$ in the same zero-dimensional subscheme.
- Infinitesimally, a deformation of $E$ lies in the symplectic leaf exactly when it preserves the subscheme $C\cap D$, which is the geometric content of the equality of $T E\phi_0$ with the symplectic subspace $T_0$.
- When $C\cdot D=0$, the intersection datum is empty and the corresponding open moduli space is symplectic: the Poisson map is an isomorphism, so the whole open component is a single symplectic leaf.
- In the spectral-cover description of Higgs bundles, the map $E\mapsto F(E)$ is the analogue of the Hitchin map, so the theorem describes the symplectic foliation in that setting by the relative position of the spectral curve and the divisor $D$.
Reading between the lines
- If the tangent identification is uniform over $M'_0$, the Poisson structure on that open locus should be regular, with all symplectic leaves of the same dimension rather than a foliation whose leaf dimension jumps.
- The statement may need refinement when a fiber of $g:C\mapsto C\cap D$ is disconnected: under the usual definition of a symplectic leaf as a maximal connected submanifold, each connected component would be a leaf, and the whole fiber would be a union of leaves rather than one leaf.
- A similar argument could plausibly extend to sheaves whose Fitting support is singular or has nilpotent thickenings, but the tangent identification would have to be redone because the Hilbert scheme of such supports need not be smooth.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies the moduli space M_H(S,P) of H-stable pure dimension-one sheaves on a smooth complex projective Poisson surface (S,s), where s is a nonzero section of -K_S with degeneracy divisor D. The Poisson structure on the moduli space, due to Tyurin and Bottacin, is taken as external input. The main result (Theorem 5.1) states that, on the open subset M'_0 parametrizing sheaves E = i_*L with smooth support C meeting D in finitely many points, the fibers of the natural morphism phi0: M'_0 -> Sym^{d'}(D), E -> C∩D, are the symplectic leaves of the Poisson structure. The proof identifies, at each point, the tangent space to the fiber of phi0 with the symplectic subspace T0 of Ext^1(E,E) determined by the factorization of the Poisson map sigma, using the local-to-global spectral sequence and the identification of the tangent space of the Hilbert scheme with H^0(N_{C/S}). A vanishing statement H^0(C,K_S|_C)=0 is used to prove smoothness of the Hilbert scheme at the relevant points.
Significance. If the result is correct, it gives a clean geometric description of the symplectic foliation of the Tyurin–Bottacin Poisson structure on moduli of pure dimension-one sheaves: leaves are controlled by the intersection of the Fitting subscheme F(E) with the anticanonical divisor D. The paper is concise, relies transparently on the external Poisson-structure results of Tyurin and Bottacin, and contains a concrete tangent-space computation that is the core of the argument. The main conceptual contribution—relating the symplectic subspace to deformations fixing the intersection C∩D—is plausible and useful. The missing connectivity and smoothness arguments are gaps in the proof of the stated theorem, but they appear fixable without changing the main computation.
major comments (3)
- [Section 5, Theorem 5.1] The proof establishes that, for each point E, the tangent space to the fiber of phi0 equals the symplectic subspace T0. This shows the fiber is an integral submanifold of the symplectic distribution, but a symplectic leaf is required to be a maximal connected integral submanifold. The connectedness of the fibers of phi0 is never proved. In particular, the fibers of g: Q0 -> Sym^{d'}(D) could in principle be disconnected, and if so a fiber of phi0 would split into several symplectic leaves, making the statement of Theorem 5.1 false as written. The manuscript should add a proof that g^{-1}(Z) and hence phi0^{-1}(Z) is connected, for instance by realizing g^{-1}(Z) as a connected linear system (a projective space or a connected open subset of one) and using that the fibers of f0 are Picard varieties, which are connected.
- [Section 5, before (5.1)] The assertion that f0 is a smooth morphism is not proved. To use the short exact sequence (5.1) for tangent spaces, one needs f0 to be smooth, or at least to know that the differential of f0 is surjective and that the relative tangent sequence is exact. The identification of the differential df0|E with the map i_{E,E} in the local-to-global sequence is a plausible route to pointwise surjectivity, but the argument should be written out explicitly. Similarly, the identification of the tangent space to the fiber of g with H^0(N_{C/S}(-D)) needs justification: one must explain why the infinitesimal deformations of C fixing the subscheme Z = C∩D are exactly these sections, and that g is a well-behaved morphism on the relevant open subset.
- [Section 5, smoothness of the Hilbert scheme] The statement that H^0(C,K_S|_C)=0 for every effective curve C with C·D≠0 is false when C is reducible and has an irreducible component disjoint from D: on that component, K_S is trivial, so the restriction has nonzero global sections. The vanishing is correct for the smooth curves appearing in Q0, since for a smooth connected C with C·D>0 the line bundle O_C(-D) has negative degree. The overbroad wording should be corrected to apply only to the smooth curves used in the proof, although this error does not invalidate Theorem 5.1 itself.
minor comments (5)
- [Abstract and Introduction] The abstract states that the symplectic leaves of the full moduli space M_H(S,P) are the fibers of the natural map to the symmetric power of D, but the theorem is proved only for the open subset M'_0 where F(E) is smooth and intersects D in a zero-dimensional subscheme. The abstract should be qualified to match the theorem, since the map is not defined on the whole moduli space.
- [Notation throughout] The letter C is used both for a divisor class in NS(S) (Section 2 and later in M = M_H(S,C,P)) and for a specific curve F(E) (Section 5). Also d is used for the degree of the line bundle L while d' = C·D. This is confusing; please introduce distinct notation, for example [C] for the class and C for the curve.
- [Lemma 3.1] In the proof of Lemma 3.1, the claim that K_S|_C is trivial when C·D=0 should be justified explicitly: since s vanishes exactly on D and C∩D is empty (because the intersection is zero-dimensional and has degree zero), the restriction of s to C is nowhere vanishing, so K_S|_C is trivial. As written, 'it follows' skips the key point.
- [Proposition 4.2, diagram (4.1)] The diagram (4.1) is difficult to read because several rows and columns are aligned ambiguously; for example the inclusion on the right column is not clearly labeled. Please redraw the diagram with explicit labels for all arrows, especially the vertical map sigma_2 and the inclusion H^0(Ext^1(E,E⊗K_S)) -> H^0(Ext^1(E,E)).
- [Section 2, definition of M'_0] The definition of the open subset M'_0 says 'parametrizing sheaves E with F(E) smooth'; this should say that F(E) is a smooth curve (and that F(E)∩D is zero-dimensional, as already stated). It may also be worth noting that smoothness of the Fitting subscheme is an open condition, so this is indeed a Zariski open subset.
Circularity Check
No circularity: the symplectic-leaf theorem is an honest derivation from the externally supplied Tyurin–Bottacin Poisson structure.
full rationale
The paper does not exhibit any of the circularity patterns. The Poisson structure sigma is taken as an external input from Tyurin and Bottacin, and the main theorem derives the symplectic-leaf description from it rather than assuming it. Proposition 4.2 computes the symplectic subspace T0 using the local-to-global spectral sequence for Ext and the section s, with no fitted parameters and no equation whose conclusion is substituted for an assumption. Theorem 5.1 then identifies the tangent space to the fiber of phi0 with T0 through canonical identifications: the differential of f0 is identified with i_{E,E} via H^1(C,O_C)=H^1(S,End(E,E)), and the right-hand vertical inclusion is identified with the tangent space to the fiber of g via H^0(N_{C/S}(-D))=H^0(S,Ext^1(E,E⊗K_S)). This is a genuine computation, not a renaming or a self-citation chain. There are no self-citations that are load-bearing; the cited prior results of Tyurin, Bottacin, Mukai, and Markman are independent external inputs. The known gap concerning connectedness of the fibers of phi0, which would be needed to conclude that each entire fiber is one maximal symplectic leaf rather than a union of leaves, is a correctness or completeness issue, not a circularity issue. Likewise, the overbroad claim in Section 5 that H^0(C,K_S|_C)=0 for every effective curve with C·D≠0 is false for reducible curves with a component disjoint from D, but the proof only needs the smooth-curve case, and in any event this is an error in an auxiliary smoothness argument, not a circular reduction. The derivation chain is self-contained relative to the stated external Poisson-structure input.
Assumptions & free parameters
assumptions (5)
- domain assumption Tyurin-Bottacin construction yields a holomorphic Poisson structure σ on the moduli space (external input).
- domain assumption H-stable pure dimension one sheaves are simple with vanishing trace-free Ext^2, so the moduli space is smooth.
- standard math The local-to-global spectral sequence for Ext and Serre duality give the exact sequences in (4.2).
- standard math For smooth C with C·D≠0, H^0(C,K_S|_C)=0.
- domain assumption The quotient H^0(N_C/S)/H^0(N_C/S(-D)) is the tangent space to the fiber of g: Q0 -> Sym^{d'}(D).
Cite this review
Pith. "Pith review of Poisson structure on the moduli spaces of sheaves of pure dimension one on a surface." pith.science (2026). https://pith.science/paper/CXG2GM4W
@misc{pith2026190808260,
author = {Pith},
title = {Pith review of: Poisson structure on the moduli spaces of sheaves of pure dimension one on a surface},
year = {2026},
howpublished = {\url{https://pith.science/paper/CXG2GM4W}},
note = {Machine review of arXiv:1908.08260}
}
read the original abstract
Let S be a smooth complex projective surface equipped with a Poisson structure s and also a polarization H. The moduli space M_H(S,P) of stable sheaves on S having a fixed Hilbert polynomial P of degree one has a natural Poisson structure given by s, studied by Tyurin and Bottacin. We prove that the symplectic leaves of M_H(S,P) are the fibers of the natural map from it to the symmetric power of the effective divisor on S given by the singular locus of s.
Reference graph
Works this paper leans on
-
[1]
Altman and S
A. Altman and S. Kleiman, Compactifying the Picard scheme, Adv. in Math. 35 (1980), 50--112
1980
-
[2]
Bottacin, Symplectic geometry on moduli spaces of stable pairs, Ann
F. Bottacin, Symplectic geometry on moduli spaces of stable pairs, Ann. Sci. École Norm. Sup. 28 (1995), 391--433
work page 1995
-
[3]
Bottacin, Poisson structures on moduli spaces of sheaves over Poisson surfaces, Invent
F. Bottacin, Poisson structures on moduli spaces of sheaves over Poisson surfaces, Invent. Math. 121 (1995), 421--436
work page 1995
- [4]
-
[5]
Markman, Spectral curves and integrable systems
E. Markman, Spectral curves and integrable systems. Compositio Math. 93 (1994), 255--90
work page 1994
-
[6]
Mukai, Symplectic structure of the moduli space of sheaves on an abelian or K3 surface, Invent
S. Mukai, Symplectic structure of the moduli space of sheaves on an abelian or K3 surface, Invent. Math. 77 (1984), 101--116
work page 1984
-
[7]
C. T. Simpson, Moduli of representations of the fundamental group of a smooth projective variety I, Inst. Hautes \'Etudes Sci. Publ. Math. 79 (1994), 47--129
1994
-
[8]
A. N. Tyurin, Symplectic structures on the varieties of moduli of vector bundles on algebraic surfaces with p_g>0 , Math. USSR Izvestiya 33 (1989), 139--177
work page 1989
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.