REVIEW 4 major objections 5 minor 1 cited by
Generators of the cohomology ring, after Newstead
T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read For a one-nodal curve, the rational cohomology ring of the rank-2, fixed-determinant moduli space is minimally generated by explicit limit classes in degrees 2, 3, 4, and 6.
desk verdict Real new result with a sound degeneration strategy, but the proof leans on unproved companion preprints and a few sketched weight-filtration steps; referee must check those. 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 argument hinges on an auxiliary degeneration of moduli spaces whose central fiber is a simple normal crossings divisor, obtained by a standard degeneration construction. The special fiber splits into two smooth components meeting along a $\mathbb{P}^1 \times \mathbb{P}^1$-bundle over the moduli space of the normalized curve, with one component itself a $\mathbb{P}^3$-bundle over the same base. A specialization morphism from the cohomology of the central fiber to the limit mixed Hodge structure on the smooth fiber is governed by an exact sequence that describes both the kernel, coming from push-forwards along the inclusions of the intersection, and the cokernel, coming from monodromy-weight graded pieces. Cup products of the smooth-curve generators are lifted through this sequence, and the projective-bundle cohomology decomposition shows that the lifted classes generate the whole ring. A proper contraction from the central fiber to the fixed-determinant moduli space then identifies the monodromy-invariant subring with the cohomology of $U_{X_0}(2,L_0)$.
What would settle it
Compute the third cohomology of $U_{X_0}(2,L_0)$ for a low-genus one-nodal curve by an independent method, for instance from the long exact sequence relating this space to the moduli space on the normalized curve; the theorem predicts that $H^3$ is $(2g-1)$-dimensional and spanned by the $\eta^{(3)}_i$. A different dimension, or a cohomology class in any degree not expressible as a polynomial in $\eta^{(2)}_1$, $\eta^{(4)}_1$, $\eta^{(6)}_1$, and the $\eta^{(3)}_i$, would refute the claimed generation.
Extended reading notes
Core claim
The central claim is that for $X_0$ an irreducible nodal curve with exactly one node and $L_0$ an invertible sheaf of odd degree, the rational cohomology ring of the moduli space $U_{X_0}(2,L_0)$ of rank-2 semistable torsion-free sheaves with determinant $L_0$ has a minimal generating set consisting of $\eta^{(2)}_1 \in H^2$, $\eta^{(4)}_1 \in H^4$, $\eta^{(6)}_1 \in H^6$, and $\eta^{(3)}_i \in H^3$ for $1 \le i \le 2g-1$. The proof places $X_0$ as the special fiber of a smooth family of curves and uses an auxiliary degeneration of moduli spaces whose central fiber is a simple normal crossings divisor. The smooth-curve generators of the nearby fibers are transported to the special fiber through specialization maps, and the monodromy-invariant part of the cohomology is shown to be generated by exactly these classes. A proper map from the degeneration's central fiber to $U_{X_0}(2,L_0)$ then carries these classes down to the singular moduli space. The theorem is therefore an existence statement with content: the generators are the natural degeneration limits of the generators in the smooth case.
Load-bearing premise
Everything rests on the imported fact that, as the curve degenerates to a nodal curve, the auxiliary moduli space degenerates into two smooth halves meeting along a simple intermediate space, and on an exact sequence describing how cohomology classes from the smooth fibers lift to this union; if either fact fails for the fixed-determinant family, the generator proof collapses.
Editorial extensions
If this is right
- The rational cohomology ring of the one-nodal fixed-determinant moduli space is generated by $2g+2$ classes: one in degree 2, one in degree 4, one in degree 6, and $2g-1$ in degree 3.
- One of the $2g$ odd degree-3 generators of the smooth case becomes non-monodromy-invariant under degeneration, and its cohomological role is carried by a degree-6 class that is the product of the two special odd classes.
- All generators lie in degrees at most 6, and the only odd-degree generators lie in degree 3, so every cohomology class of the nodal moduli space is a polynomial in these low-degree classes.
- The proof identifies the cohomology of the nodal moduli space with the monodromy-invariant part of the smooth-fiber cohomology under the specialization morphism, so the nodal ring is determined by the monodromy weight filtration.
Reading between the lines
- The same degeneration picture suggests a testable extension to curves with several nodes: each additional node should delete another odd degree-3 generator, with products of exceptional odd classes supplying higher even-degree replacements, so the count of generators would grow with the number of nodes.
- Because the theorem realizes the nodal cohomology ring as a monodromy-invariant subring, computing the weight filtration on the smooth moduli space gives an independent route to the Betti and Hodge numbers of the singular moduli space without resolving it, a calculation the paper does not carry out.
- The explicit generator set is a natural input for studying related moduli spaces over nodal curves, mirroring how smooth-curve generator results have been used for other moduli problems.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper claims to generalize Newstead's classical generator theorem for the rational cohomology ring of the moduli space of rank-2 stable bundles with fixed odd determinant from smooth curves to irreducible nodal curves with exactly one node. The strategy is to degenerate a smooth curve to the nodal curve, work with the relative Gieseker moduli space whose central fiber is a simple normal crossings divisor, use limit mixed Hodge structures and the specialization morphism, and then pass from the Gieseker space to the Simpson moduli space via the natural proper morphism. The main results are Theorem 2.3, giving a minimal generating set for the cohomology ring of the Gieseker central fiber, and Theorem 3.2, giving a minimal generating set for the cohomology ring of the Simpson moduli space U_{X0}(2,L0). The paper is short and depends heavily on two companion preprints, [3] and [6], for the structure of the central fiber and for the key exact sequence (2.2).
Significance. If the proof is completed, the result would be the first explicit generator theorem for a singular Simpson moduli space with fixed determinant, and the interpretation of the generators as degenerations of Newstead's smooth-case generators is a valuable structural insight. The overall strategy—Gieseker degeneration, limit mixed Hodge structures, and Mayer-Vietoris/Gysin arguments—is coherent and promising. The paper also benefits from being concise and from stating its main theorems precisely. However, the central load-bearing exact sequence and parts of the weight-filtration analysis are imported from unpublished companion preprints, and the generation/minimality arguments in the proof of Theorem 2.3 are asserted too tersely. The significance is therefore conditional on filling those gaps; with the gaps filled, the paper would make a solid contribution.
major comments (4)
- [§2.2, Eq. (2.2)] The exact sequence (2.2) is the central tool of the paper: it identifies ker(sp^i) with Im(f_i) and coker(sp^i) with Gr^W_{i+1}H^i(G(2,L)_∞,Q). It is imported from [6, Corollary 2.4], a companion preprint that is not yet published as far as the manuscript indicates, and neither its statement nor its proof is reproduced here. Since Theorem 2.3 and Theorem 3.2 both rely directly on this sequence, the authors must either supply a self-contained proof of (2.2) in this paper or give a precise reference to a published version; otherwise the main results remain conditional on an external, not-yet-refereed source.
- [§2.3, proof of Theorem 2.3] The assertion 'This implies Im(f_k) ⊂ V_k' after the exact sequence (2.4) is not justified. The map f_k is a Gysin-type morphism, not a ring homomorphism, so the fact that H^*(G0∩G1,Q) is generated by α0, β0, ξ1, ξ2, and ψ'_i does not by itself imply that the images f_k of arbitrary monomials in those generators lie in the subring V generated by the listed ξ^(i)_j. A projection-formula argument or an explicit inductive computation is needed to show that the f_k-images land in V_k for every k.
- [§2.3, proof of Theorem 2.3, and §3, proof of Theorem 3.2] The proof that sp^k(V_k)=Im(sp^k) is too terse. The text says that W_kH^k(G(2,L)_∞,Q) is generated by monomials α^i_∞ β^j_∞ ψ_{i_1}...ψ_{i_t} for which either t<2g or {g,2g} lies in the index set, and then concludes sp^k(V_k)=Im(sp^k). This requires a case-by-case check that each such monomial has a preimage in V_k, including the monomial ψ_g ψ_{2g} and its products with α∞ and β∞; the paper does not provide that check. Similarly, the claim that the generating set is minimal is dismissed with 'The minimality ... is straightforward' without proof. Since minimality is part of the stated theorems, this needs an explicit argument, for example a dimension count in graded pieces or an independence check.
- [§3, proof of Theorem 3.2] The final identification H^i(U_{X0}(2,L0),Q) ≅ Im(θ^*) ≅ Im(sp^i) ≅ W_iH^i(G(2,L)_∞,Q) depends on the same unproved generation statement for the image of sp^i. In particular, the equality Im(sp^i)=W_iH^i(G(2,L)_∞,Q) requires knowing that every weight-pure class in W_iH^i is in the image of the specialization map, which is exactly the content of the missing monomial check in the proof of Theorem 2.3. This is not an independent objection, but it shows that the gap in the proof of Theorem 2.3 directly affects both main theorems.
minor comments (5)
- [Title header] The running title contains a typo: 'GENERA TORS' should be 'GENERATORS'; there are also several OCR-like spacing errors in the body, e.g., 'de gree' in the Abstract and 'th e' in the Introduction. A thorough proofread is needed.
- [§2.3, proof of Theorem 2.3] The line defining ξ^(2)_1 says '1 ∈ H^0(G1∩G2,Q)', but the intersection of the two components of G_{X0}(2,L0) is G0∩G1. This appears to be a typo.
- [§3, proof of Theorem 3.2] The phrase 'excessive couple' should be 'excisive couple'; the standard terminology in [26, Example B.5] is 'excisive'. Please correct the terminology.
- [§2.2] The sentence beginning 'Unfortunately, the resulting specialization map is not a morphism of mixed Hodge structures' is confusing because the next sentence explains that after identifying the cohomology of the nearby fiber with the limit cohomology, the modified map is a morphism of mixed Hodge structures. The 'Unfortunately' is misleading and the two-step construction would benefit from being stated more cleanly.
- [§2.3, proof of Theorem 2.3] In the proof, the statement that f2 is injective because G0∩G1 is smooth and rationally connected is plausible but not explained. Since H^0(G0∩G1)=Q and f2 maps this to H^2(G_{X0}(2,L0),Q), injectivity means the Gysin image of the fundamental class is nonzero; a one-line justification would be helpful.
Circularity Check
Main theorem is carried by the authors' own exact sequence (2.2) and Gieseker central-fiber structure; not definitionally circular, but self-citation is load-bearing.
-
self citation load bearing
[Section 2.2, equation (2.2)]
"By [6, Corollary 2.4], we then have the following exact sequence of mixed Hodge structures: H^{i−2}(G0∩G1,Q)(−1) — f_i → H^i(G_X0(2,L0),Q) — sp^i → H^i(G(2,L)_∞,Q) — g_i → Gr^W_{i+1}H^i(G(2,L)_∞,Q) → 0, (2.2)"
This sequence supplies the key identifications ker(sp^i)=Im(f_i) and coker(sp^i)=Gr^W_{i+1}H^i(G∞). Every generator in Theorem 2.3 is then defined through f_i or sp^i: ξ(3)_i=sp^{-1}_3(ψ∞_i), ξ(4)_2=f_4(ξ1⊕ξ2), ξ(5)_i=f_5(ψ'_i), ξ(6)_1=f_6(β0). Theorem 3.2's conclusion H^i(U)≅Im(sp^i) and the final statement that W_iH^i(G∞) is generated by α∞,β∞,ψ∞_i (1≤i≤2g−1) and ψ∞_gψ∞_2g are then read off from this same imported sequence plus [6, Theorem 4.2]. The cited [6] is a companion preprint by two of the present authors (Dan–Kaur, arXiv:1908.02279), so the central generator claim is not derived inside this paper; it is routed through an unverified-in-this-paper result from the same group.
-
self citation load bearing
[Section 2.1, Notation 2.1]
"there exists a regular, flat, projective family π2:G(2,L)→Δ called the relative Gieseker moduli spaces ... such that ... the central fiber ... is a reduced simple normal crossings divisor of G(2,L) (see [3, Notation A.5, Theorem A.7 and Remark A.8])."
The SNC structure of the Gieseker central fiber is the stated reason for using Steenbrink and Schmid's limit mixed Hodge structure; without it the specialization maps sp^i do not have the required properties. This structure is imported from [3], an earlier preprint by the same three authors. The proof of Theorem 2.3 then builds the whole generating set on top of this imported structural theorem. Since [3] is not machine-checked and is not shown to be independent of the present claim, the derivation's foundation is a self-citation chain rather than an external benchmark.
full rationale
There is no definitional circularity of the 'fit renamed as prediction' kind: the paper does not tune a parameter to H^*(U_X0) and then call it a generator; Newstead's theorem [25] is an independent classical input; and the η-classes are constructed through the Gieseker model and θ^*, not by assuming the conclusion. The derivation is a genuine reduction of the Simpson moduli-space statement to a statement about the Gieseker model. However, the paper is not self-contained at the load-bearing points: the exact sequence (2.2), the SNC description of the central fiber, and parts of the weight-filtration analysis (the Gr^W_4H^3(G∞)≅Q claim, the use of [6, Theorem 4.2] and [6, §7.3]) are quoted from the authors' own companion preprints [3] and [6]. If those preprints are accepted as valid external theorems, the argument works as a derivation from Newstead; if they are not, Theorem 3.2 is unsupported. The minimality assertion in Theorem 2.3 is also only declared 'straightforward' rather than proved, which is a proof gap but not circularity. Overall: heavy self-reliance on companion results, but no equation-level equivalence between input and output, so a moderate score of 4 is appropriate.
Assumptions & free parameters
assumptions (5)
- standard math There exists a regular, flat family pi1: X -> Delta with smooth general fiber and central fiber X0, for any irreducible one-nodal curve X0.
- domain assumption The relative Gieseker moduli space G(2,L) exists and its central fiber G_X0(2,L0) is a reduced simple normal crossings divisor with G1 a P3-bundle and G0 cap G1 a P1 x P1-bundle over M_{\tilde X0}(2,\tilde L0).
- domain assumption For the Gieseker degeneration, the specialization map fits into the exact sequence (2.2), with Gysin maps factoring through the graded pieces of the mixed Hodge structure on H^i(G_X0(2,L0), Q).
- domain assumption The cohomology of the generic fiber H^*(G(2,L)_infty, Q) is generated by monomials alpha_infty^i beta_infty^j psi_infty^{i1} ... psi_infty^{it} with the index restrictions from [21, Remark 5.2].
- standard math Leray-Hirsch decompositions hold for the P1 x P1 and P3 bundles over M_{\tilde X0}(2,\tilde L0).
Cite this review
Pith. "Pith review of Generators of the cohomology ring, after Newstead." pith.science (2026). https://pith.science/paper/ASMQEWXC
@misc{pith2026190801330,
author = {Pith},
title = {Pith review of: Generators of the cohomology ring, after Newstead},
year = {2026},
howpublished = {\url{https://pith.science/paper/ASMQEWXC}},
note = {Machine review of arXiv:1908.01330}
}
read the original abstract
Newstead gave the generators of the cohomology ring of the moduli space of rank 2 semi-stable, torsion-free sheaves with fixed odd degree determinant over a smooth, projective curve. In this article, we generalize this result to the case when the underlying curve is irreducible, nodal. We show that these generators (of the cohomology ring in the nodal curve case) arise naturally as degeneration of Newstead's generators in the smooth curve case.
Forward citations
Cited by 1 Pith paper
-
Generalization of a conjecture of Mumford
For an irreducible nodal curve of genus at least two, the cohomology ring of the moduli space of rank-two semistable sheaves with fixed odd determinant obeys the Mumford relations restricted to monodromy-invariant cla...
Reference graph
Works this paper leans on
-
[3]
S. Basu, A. Dan, and Kaur I. Degeneration of intermediate Jacobians and the Torelli theorem. arXiv preprint arXiv:1809.08604, 2018
work page Pith review arXiv 2018
-
[6]
Generalization of a conjecture of Mumford
A. Dan and I. Kaur. Generalization of a conjecture of mumf ord. arXiv preprint arXiv:1908.02279 , 2019
work page Pith review arXiv 1908
-
[1]
M. F. Atiyah and R. Bott. The Yang-Mills equations over Ri emann surfaces. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and En gineering Sciences, 308(1505):523–615, 1983
work page 1983
-
[2]
M. Baker. Specialization of linear systems from curves t o graphs. Algebra & Number Theory , 2(6):613–653, 2008
work page 2008
-
[4]
U. N. Bhosle. Vector bundles with a fixed determinant on an irreducible nodal curve. In Proceedings of the Indian Academy of Sciences-Mathematical Sciences , volume 115, pages 445–451. Springer, 2005
work page 2005
-
[5]
I. Biswas and N Raghavendra. Canonical generators of the cohomology of moduli of parabolic bundles on curves. Mathematische Annalen , 306(1):1–14, 1996
work page 1996
-
[7]
M. A. de Cataldo, T. Hausel, and L Migliorini. Topology of Hitchin systems and Hodge theory of character varieties: the case A1. Annals of Mathematics , 175:1329–1407, 2012. GENERATORS OF THE COHOMOLOGY RING 7
work page 2012
-
[8]
S. del Ba˜ no. On the Chow motive of some moduli spaces. Journal fur die Reine und Angewandte Mathematik , pages 105–132, 2001
work page 2001
Show all 34 references
-
[9]
del Ba˜ no
S. del Ba˜ no. On the motive of moduli spaces of rank two vector bundles over a curve. Compositio Mathematica, 131(1):1–30, 2002
2002
-
[10]
U. V. Desale and S. Ramanan. Poincar´ e polynomials of th e variety of stable bundles. Mathematische Annalen, 216(3):233–244, 1975
1975
-
[11]
A. Dhillon. On the cohomology of moduli of vector bundle s and the Tamagawa number of SL n. Canadian Journal of Mathematics , 58(5):1000–1025, 2006
2006
-
[12]
R. Earl. A note on the cohomology of moduli of rank two sta ble bundles. arXiv preprint alg-geom/9603009 , 1996
1996 arXiv
-
[13]
Earl and F
R. Earl and F. Kirwan. The Hodge numbers of the moduli spa ces of vector bundles over a Riemann surface. The Quarterly Journal of Mathematics , 51(4):465–483, 2000
2000
-
[14]
Gieseker
D. Gieseker. A degeneration of the moduli space of stabl e bundles. Journal of Differential Geometry , 19(1):173–206, 1984
1984
-
[15]
Gonz´ alez-Mart ´ ınez
C. Gonz´ alez-Mart ´ ınez. The Hodge-Poincar´ e polynomial of the moduli spaces of stable vector bundles over an algebraic curve. manuscripta mathematica , 137(1-2):19–55, 2012
2012
-
[16]
Hausel and F
T. Hausel and F. Rodriguez-Villegas. Mixed Hodge polyn omials of character varieties. Inventiones mathe- maticae, 174(3):555–624, 2008
2008
-
[17]
Hausel and M
T. Hausel and M. Thaddeus. Mirror symmetry, Langlands d uality, and the Hitchin system. Inventiones mathematicae, 153(1):197–229, 2003
2003
-
[18]
Hausel and M
T. Hausel and M. Thaddeus. Generators for the cohomolog y ring of the moduli space of rank 2 Higgs bundles. Proceedings of the London Mathematical society , 88(3):632–658, 2004
2004
-
[19]
Herrera and S Salamon
R. Herrera and S Salamon. Intersection numbers on modul i spaces and symmetries of a Verlinde formula. Communications in mathematical physics , 188(3):521–534, 1997
1997
-
[20]
Huybrechts and M
D. Huybrechts and M. Lehn. The geometry of moduli spaces of sheaves . Springer, 2010
2010
-
[21]
A. D. King and P. E. Newstead. On the cohomology ring of th e moduli space of rank 2 vector bundles on a curve. Topology, 37(2):407–418, 1998
1998
-
[22]
F. Kirwan. The cohomology rings of moduli spaces of bund les over Riemann surfaces. Journal of the American Mathematical Society, 5(4):853–906, 1992
1992
-
[23]
Periods of a moduli space of bun dles on curves
D Mumford and P Newstead. Periods of a moduli space of bun dles on curves. American Journal of Mathe- matics, 90(4):1200–1208, 1968
1968
-
[24]
P. E. Newstead. Topological properties of some spaces o f stable bundles. Topology, 6(2):241–262, 1967
1967
-
[25]
P. E. Newstead. Characteristic classes of stable bundl es of rank 2 over an algebraic curve. Transactions of the American Mathematical Society , 169:337–345, 1972
1972
-
[26]
Peters and J
C. Peters and J. H. M. Steenbrink. Mixed Hodge structures , volume 52. Springer Science & Business Media, 2008
2008
-
[27]
W. Schmid. Variation of Hodge structure: the singulari ties of the period mapping. Inventiones mathematicae, 22(3-4):211–319, 1973
1973
-
[28]
Siebert and G
B. Siebert and G. Tian. Recursive relations for the coho mology ring of moduli spaces of stable bundles. Turkish J. Math , 19(2):131–144, 1995
1995
-
[29]
Steenbrink
J. Steenbrink. Limits of Hodge structures. Inventiones mathematicae , 31:229–257, 1976
1976
-
[30]
J. H. M. Steenbrink. Some remarks about the Hodge conjecture, Hodge theory (Sant Cugat, 1985) . Lecture Notes in Maths. -1246. Springer, Berlin, 1987
1985
-
[31]
X. Sun. Moduli spaces of SL(r)-bundles on singular irre ducible curves. Asia Journal of Mathematics , 7(4):609– 625, 2003
2003
-
[32]
C. Voisin. Hodge Theory and Complex Algebraic Geometry-I . Cambridge studies in advanced mathematics-76. Cambridge University press, 2002
2002
-
[33]
Elementary aspects of the Verlinde formula and of the Harder -Narasimhan-Atiyah-Bott formula
D Zagier. Elementary aspects of the Verlinde formula and of the Harder -Narasimhan-Atiyah-Bott formula . Max-Planck-Institut f¨ ur Mathematik, 1994
1994
-
[34]
D. Zagier. On the cohomology of moduli spaces of rank two vector bundles over curves. In The moduli space of curves , pages 533–563. Springer, 1995. 8 S. BASU, A. DAN, AND I. KAUR Institute of Mathematical Sciences, HBNI, CIT Campus, Thar ama ni, Chennai 600113, India E-mail a...
1995
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.