Pith. sign in

REVIEW 4 major objections 4 minor 5 references

On the Virtual Euler Characteristic of the Moduli Space of Stable Pairs on Surfaces

T0 review · 4 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read For a non-singular toric surface, the virtual Euler characteristic of the stable pair moduli space reduces to a finite sum over partitions, proven for the projective plane in degree 1 and conjectured in general.

desk verdict A genuinely new virtual tangent space formula for stable pairs on toric surfaces, but the headline d=1 result is off by a q-shift and the key regrouping lemma is underproved. read the letter →

arxiv 2505.23531 v1 pith:WGVKOTNB submitted 2025-05-29 math.AG

Ana Pavlaković This is my paper
classification math.AG MSC 14N3514M2514C05
keywords stablepairstoricsurfacesvirtualEulercharacteristiclocalisationnestedHilbertschemesYoungdiagramspartitionfunctionsrationalityconjecture
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

The paper claims that on any non-singular toric surface $S$, the generating function of virtual Euler characteristics of the moduli space of stable pairs (a curve together with points on it) can be written as a finite sum over combinatorial data: a partition $\lambda_\alpha$ inside a $(d_1,d_2)$-ray Young diagram at every corner of the toric polygon. The central step is an explicit formula for the virtual tangent space at torus-fixed points as a sum of Laurent polynomials, the vertex terms $F_\alpha + G_\alpha$ and edge terms $E_{\alpha\beta}$, whose equivariant Euler classes are products of factors $\frac{1+x}{x}$ evaluated at linear forms in the equivariant parameters. If the main theorem is right, these invariants become machine-computable, which the paper demonstrates with a program and tables for $\mathbb{P}^2$ up to degree 4. That evidence supports the paper's conjectures that the global series is a rational function of $q$ with the symmetry $q \leftrightarrow q^{-1}$ and poles only at $0$ and $1$, and for $\mathbb{P}^2$ in degree 1 the closed form $\frac{3}{(1-q)^2}$ is proven. A notable feature is that the local partition functions are provably non-rational, for example $(1-q)^{1/s_1 - 1}$, so the rationality of the global series must come from exact cancellations among formal Laurent series.

What carries the argument

The load-bearing object is the regrouped virtual tangent space of Section 4.6. The raw vertex term $\widetilde{V}_\alpha$ and edge terms $\widetilde{E}_{\alpha\beta}$ are formal Laurent series with poles at $t = 1$; the paper splits each edge contribution into positive and negative parts $P_{\alpha\beta} + N_{\alpha\beta}$ and uses the normal-bundle transition function $(x_1, x_2) \mapsto (x_1^{-1}, x_2 x_1^{-m_{\alpha\beta}})$ (Remark 3.2) to show that the polar part of one chart cancels against the regular part of the adjacent chart. The result is the identity $T^{\mathrm{vir}} = \sum_\alpha (F_\alpha + G_\alpha) + \sum_{\alpha\beta} E_{\alpha\beta}$ with $V_\alpha = G_\alpha + \sum_{k,l} t_{\alpha\beta_1}^k t_{\alpha\beta_2}^l$ and $E_{\alpha\beta i} := t_{\alpha\beta i}^{-1} \frac{F_{\alpha\beta i}(t_{\alpha\beta j})}{1 - t_{\alpha\beta i}^{-1}} - \frac{F_{\alpha\beta i}(t_{\alpha\beta j} t_{\alpha\beta i}^{-m_{\alpha\beta i}})}{1 - t_{\alpha\beta i}^{-1}}$ both Laurent polynomials (Lemma 4.3), where $G_\alpha$ is built from the monomial generating function $Q_\lambda(t_1,t_2)$ of the Young diagram at vertex $\alpha$. Around this sit the standard tools the computation leans on: virtual localisation (Theorem 3.1), the description of torus-fixed points as tuples $(\vec\lambda, \vec d)$ of partitions contained in $(d_1, d_2)$-rays via monomial ideals (Section 3.3), and the vanishing of the fixed part of $T^{\mathrm{vir}}$ (Theorem 4.6), which is what makes the equivariant Euler class well-defined.

What would settle it

Write out both sides of the identity $T^{\mathrm{vir}} = \sum_\alpha (F_\alpha + G_\alpha) + \sum_{\alpha\beta} E_{\alpha\beta}$ for $\mathbb{P}^2$ with $d = 2$ at one torus-fixed point, expanding the Čech contributions to a fixed total degree in $t_1, t_2$. Computing the residue of the difference at $t_1 = 1$ (or $t_2 = 1$): Theorem 1.1 predicts it is zero, and a nonzero residue, meaning any surviving $\delta(t)$ or $(1-t)^{-k}$ term, would disprove the central formula. The same test can be run numerically by comparing the paper's table values for $d = 2$ against the residue-free sum over the finite partition data.

Watch

Extended reading notes

Core claim

The paper's central discovery is a full computation of the virtual tangent space $T^{\mathrm{vir}}$ of the moduli space of stable pairs $P_n(S,\beta)$ at the fixed points of the torus action on a non-singular toric surface. Decomposing the alternating Ext-character $\chi(\mathcal{O},\mathcal{I}_Z(D)) - \chi(\mathcal{I}_Z,\mathcal{I}_Z)$ by Čech cohomology into vertex contributions from affine charts and edge contributions from pairwise intersections, the author splits each formal Laurent series into a polar and a regular part, then shows that after the change of variables $x_2 \mapsto x_2 x_1^{-m_{\alpha\beta}}$ coming from the normal bundle, the polar parts of neighbouring vertices and edges cancel, leaving honest Laurent polynomials $V_\alpha$ and $E_{\alpha\beta}$ (Lemma 4.3). With the vanishing of the fixed part (Theorem 4.6) making $c_e^\bullet(T^{\mathrm{vir}})$ well-defined, Theorem 1.1 states that the equivariant partition function $\widetilde{Z}^S_\beta(q \mid s_1, s_2)$ is the sum over $[\vec d] = \beta$ and partitions $\vec\lambda \subset \vec d$ of $q^{|\lambda|}$ times the products $\prod_{\alpha\beta} c_e^\bullet(E_{\alpha\beta})$ and $\prod_\alpha c_e^\bullet(F_\alpha + G_\alpha)$, each factor being an explicit product of $f(x) = \frac{1+x}{x}$ terms as in equations (22)--(26). A consequence proven in the paper is that for $S = \mathbb{P}^2$ and $\beta = 1$, the non-equivariant series is the Laurent expansion of $\frac{3}{(1-q)^2}$.

Load-bearing premise

Everything rests on the regrouping claim of Section 4.6: that the infinite Laurent series coming from the individual vertex and edge pieces cancel after the normal-bundle change of variables, leaving only finite polynomials, a step asserted in a single sentence in Lemma 4.3 without an explicit cancellation computation.

Editorial extensions

If this is right

  • For any toric surface, the virtual Euler characteristic of the stable pair space can in principle be computed to arbitrary order by evaluating a finite partition sum, with no further geometric input; the paper implements this for $\mathbb{P}^2$ and degrees $d \le 4$.
  • For $S = \mathbb{P}^2$ and $\beta = 1$, the series is proven to be the Laurent expansion of $\frac{3}{(1-q)^2}$, the first closed form for the virtual Euler characteristic of stable pair spaces on a surface.
  • Low-degree data for $d = 2, 3, 4$ are interpolated by rational functions with denominator $(1-q)^{6d}$ and palindromic numerator of degree $d(d+3)$, supporting the rationality and $q \leftrightarrow q^{-1}$ symmetry conjectures.
  • Since the local pieces are non-rational, for example $(1-q)^{1/s_1 - 1}$, any proof of global rationality must exhibit cancellation of non-rational factors, and Theorem 1.1 locates that cancellation in the edge terms $E_{\alpha\beta}$.
  • The paper expects the same machinery, via capped localisation, to prove rationality for all toric surfaces and for the descendant series (Section 1.2).

Reading between the lines

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

  • The cancellation in Lemma 4.3 suggests a general principle the paper leaves implicit: a local equivariant vertex whose exponents are linear forms in $s_1, s_2$ produces a rational global series whenever the residues gathered along the closed chain of edges of the Newton polygon sum to zero, a cycle-compatibility condition checkable combinatorially without deriving closed forms.
  • Testable extension: run the paper's formulas for $\mathbb{P}^2$ at $d = 5, 6$, or for $\mathbb{P}^1 \times \mathbb{P}^1$ at $(1,1)$, and test whether the denominator $(1-q)^{6d}$ and the palindromic numerator of degree $d(d+3)$ persist; a failure would show the pattern is a low-degree artefact, while success would strengthen the conjectures of Section 6.4.
  • For $\mathbb{P}^2$ and $d = 1$ the computed virtual values equal the ordinary Euler characteristics $3(m+1)$, which the paper obtains independently from a fibration over the space of lines; if that agreement extends beyond $d = 1$, it suggests a virtual-to-topological comparison for stable pair spaces that virtual localisation alone does not explain.
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.

Referee Report

4 major / 4 minor

Summary. The paper studies stable pairs on toric surfaces via virtual localization. It describes the torus-fixed loci by pairs of edge multiplicities and vertex partitions, computes the virtual tangent space as a sum of vertex and edge contributions, and claims (Theorem 1.1, restated as Theorem 4.7) that after a regrouping these contributions become Laurent polynomials, reducing the equivariant partition function to a finite sum over partitions weighted by Euler classes. The paper further proves that the virtual tangent space has no fixed part, derives product formulas for the relevant Euler classes, computes the P2, d=1 case explicitly, provides numerical tables for d up to 4, and formulates conjectures on rationality and q ↔ 1/q symmetry.

Significance. If Theorem 1.1 is correct, it gives an effective and explicit combinatorial formula for PT-type invariants of toric surfaces, going beyond the Hilbert-scheme and Quot-scheme rationality results cited in the introduction. A notable strength is that the structural theorem is derived from standard localization and obstruction theory rather than from the conjectures, and the paper ships concrete tables and a program, so the claims are checkable. The d=1 computation and the low-degree tables are encouraging evidence. However, the central regrouping step is only sketched, one displayed definition is ambiguous, and one closed formula is derived only for the P2 value of the normal-bundle slope; these points must be fixed before the main claims are fully supported.

major comments (4)
  1. [§4.6.2, Lemma 4.3] The proof of Lemma 4.3 is not sufficient for the Laurent-polynomial claim on which Theorem 1.1 depends. It checks the pole at t_{αβ1}=1 for Vα but does not address the pole at t_{αβ2}=1, and for E_{αβ_i} it only notes that a numerator vanishes at t=1 without showing that the full quotient, including the factor (1-t^{-1}_{αβ_i}) and the change of variables from Remark 3.2, is a Laurent polynomial in both variables for arbitrary m_{αβ}. Since the Euler classes c•_e(E_{αβ}) and c•_e(Fα+Gα) in Theorem 1.1 are well-defined only if this Laurent property holds, the proof needs a complete verification, for example by explicitly writing Vα and E_{αβ_i} as finite sums. The P2, d=1 example has m_{αβ}=1 and therefore does not test the cancellation for general m_{αβ}.
  2. [§4.6.2, Eq. (12); Theorem 1.1/4.7] The definition (12) of Vα begins with the expression Fα + Gα − ..., but Fα has not been defined at that point. If Fα denotes the infinite series ~Fα from §4.4, then the equality to Gα + the finite double sum is essentially the content of Lemma 4.2 and should be stated as such; if Fα denotes a new regrouped vertex term, its definition is missing. As written, Theorem 1.1's factor c•_e(Fα+Gα) is ambiguous because the reader cannot tell which Fα is meant. Please define Fα explicitly and reconcile (12) with the statement of Theorem 4.7.
  3. [§6.4.1, Theorem 1.2] The claimed unshifted series is off by a factor of q. The d=1 table in §6.3 gives evir = 3(m+1) for m≥0, and since m = n − 1 for d=1, this means evir(P_n(P^2,1)) = 3n for n≥1; the unshifted generating series is therefore 3q/(1−q)^2, not 3/(1−q)^2. The computation displayed in (36) appears to produce the shifted series ~Z = 3/(1−q)^2, and the q-shift from (7) is missing in the statement of Theorem 1.2. The symmetry discussion and any comparison with the table should be adjusted accordingly.
  4. [§5, Eq. (26)] The closed formula for c•_e(E_{αβ1}) is derived only for the value m_{αβ}=1. The derivation starts from (13) with a general m, but the displayed factorization uses the identity (1−t^{-1})∑_{k=0}^l t^k = t^l − t^{-1}, which gives the quotient for m=1 (up to the sign conventions already built into F). For general m, the factor t^{-1} − t^{lm} leads to a different Laurent polynomial, so (26) is not the general edge Euler class. Since Theorem 1.1 is stated for arbitrary toric surfaces, either prove the general edge formula, or state explicitly that (26) is specific to the P^2 example and that the general computation is to be done from (13) alone.
minor comments (4)
  1. [§6.4.2] The text says the conjectural degree-2 polynomial is obtained by interpolating the first 12 coefficients but then says it agrees with all 25 values in Figure 4; please clarify how many coefficients were used in the interpolation and include the precise interpolation command or data file for reproducibility.
  2. [§6.4.3] The displayed conjectural rational function for d=4 contains an ellipsis inside an otherwise explicit formula; please either display all coefficients or refer the reader to the accompanying code for the full expression.
  3. [§5] The notation f(x) := (1+x)/x is used before the convention that c•_e is multiplicative and that f is evaluated on first Chern classes of the torus characters is spelled out; a short sentence explaining this would improve readability.
  4. [Throughout] There are several small typos, e.g. 'This the Laurent expansion' in §6.2.2 and 'the non-rationality of controlled' in §1.1; these should be corrected in a final pass.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the virtual tangent computation is assembled from localization and explicit character/edge calculations; the noted weaknesses are exposition and correctness gaps, not question-begging.

full rationale

I walked the claimed derivation chain and found no step that reduces to its own input by construction. The paper's main formula is assembled from the standard virtual localization theorem, the PT/nested-Hilbert scheme isomorphism, and explicit T-character computations of χ(O_d|Uα), χ(O_λ|Uα), and the Čech local-to-global decomposition. Equations (11), (12), and (13) are explicit definitions/computations, not fitted parameters. The d=1 evaluation in Section 6.4.1 is a direct cancellation calculation using the local partition functions and edge terms from Section 6.2; it is not imported as a conjecture, even though the q-normalization appears to be off by a factor of q relative to the paper's own table in Section 6.3 (m=n-1 gives Z=3q/(1-q)^2 rather than 3/(1-q)^2). The d=2,3,4 formulas are explicitly labeled conjectural and are interpolated from the computed coefficients, so they are not disguised predictions. All citations are external; there is no load-bearing self-citation chain. The genuinely concerning issues are correctness/exposition gaps rather than circularity: Fα in Theorem 4.7 and equation (12) is not cleanly defined (the displayed equality defines Vα and appears to drop the tilde from F˜α), and Lemma 4.3 gives only a one-sentence verification of the Laurent-polynomial property. These would affect whether the formula is fully specified and proved, but they do not make the derivation circular.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard obstruction theory, the nested Hilbert scheme identification, and the combinatorial localization setup. The most fragile assumption is the regrouping lemma, which is not proved in detail. There are no invented physical or geometric entities; the conjectural interpolated formulas are fitted but are not load-bearing for the main theorem.

free parameters (1)
  • Conjectural numerator coefficients for d=2,3,4 rational functions = For d=2: 1872360, -360008, 171325, -21714, 1265, -28 (and constants)
    Interpolated to match the first 12 computed coefficients in Section 6.4.2 and analogous fits for d=3,4; explicitly labeled conjectural, not used in the proof of the main theorem.
assumptions (5)
  • domain assumption The moduli space P_n(S,β) admits a 2-term perfect obstruction theory with virtual tangent space given by χ(O, I_Z(D)) - χ(I_Z, I_Z) (equations (5), (8)).
    Quoted from Lin18 Theorem 2 and GSY20 Proposition 2.5; not proved in this paper. It is the starting point of the whole computation.
  • domain assumption P_n(S,β) is isomorphic to the nested Hilbert scheme S^{[0,m]}_β with m = n + β(β+K_S)/2 (Proposition 2.4).
    Quoted from PT10 Proposition B.8; used throughout to describe fixed points via partitions.
  • domain assumption Torus fixed loci are isolated and in bijection with pairs (λ,d) as described in Section 3.3.
    Stated without proof; needed for the localization sum to be a finite sum over partitions.
  • standard math The local-to-global spectral sequence and Čech cohomology give the vertex/edge decomposition (9), with no missing higher cohomology terms.
    Uses standard spectral sequence; the paper applies it directly.
  • ad hoc to paper The regrouping in Section 4.6 turns the formal Laurent series into Laurent polynomials, specifically Lemma 4.2 and Lemma 4.3.
    The proof of Lemma 4.3 is only sketched; the cancellation of infinite series relies on the normal bundle change of variables (Remark 3.2). This is the most delicate step.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On the Virtual Euler Characteristic of the Moduli Space of Stable Pairs on Surfaces." pith.science (2026). https://pith.science/paper/WGVKOTNB

@misc{pith2026250523531,
  author       = {Pith},
  title        = {Pith review of: On the Virtual Euler Characteristic of the Moduli Space of Stable Pairs on Surfaces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WGVKOTNB}},
  note         = {Machine review of arXiv:2505.23531}
}
read the original abstract

We study the stable pair theory on toric surfaces and determine the virtual tangent space over the fixed point loci. Further, we present a program to compute the virtual Euler characteristic, illustrated by the case of the projective plane. As an application, conjectures regarding rationality and symmetry are supported by verification of a special case.

Figures

Figures reproduced from arXiv: 2505.23531 by the authors.

Figure 1
Figure 1. Case (d1, d2) = (1, 0), b = 4 Therefore, Qλ = X b−1 i=0 t i 1 . In this case the finite vertex contribution Gα is Gλ⊂(1,0) = Qαt −1 1 t −1 2 − QαQα(1 − t −1 1 )(1 − t −1 2 ) + Qα(1 − t −1 2 ) = Qαt −1 1 t −1 2 − Qα(t −1 2 )Qα(1 − t −1 1 − 1) = Qαt −1 1 t −1 2 + Qαt −b 1 (1 − t −1 2 ) = X b i=1 t −i 1 , where m = b is the size of the partition λ. This yields Z˜C 2 (0,1)(q | s1, s2) = X b≥0 q b Y b i=1 1 − is1 −is1 = … view at source ↗
Figure 2
Figure 2. Case (d1, d2) = (1, 1), a = 5, b = 4 In this case, Qλ = 1 +X b i=1 t i 1 + Xa j=1 t j 2 . Thus, A + B = X b i=1 t i 1 + Xa j=1 t j 2 + 1 − t −1 2 X b−1 i=0 t i 1 − t −1 1 aX−1 j=0 t j 2 + t −1 2 X b i=1 t −(i+1) 1 + t −1 1 Xa j=1 t −(j+1) 2 . Furthermore, QλQλ = t1 t b 1 − 1 t1 − 1 X b i=1 t −i 1 + (t b 1 − 1)(t a 2 − 1) (t1 − 1)(t2 − 1)t1t −a 2 + X b i=1 t i 1 + (t b 1 − 1)(t a 2 − 1) (t1 − 1)(t2 − 1)t −b 1 t2 + t2… view at source ↗
Figure 3
Figure 3. Case (d1, d2) = (2, 0), a = 4, b = 2 We want to compute Gλ⊂(2,0) = Qλ(1 − t −2 2 ) | {z } =:A + Qλ t1t2 | {z } =:B −(t −1 1 − 1)(t −1 2 − 1)QλQλ | {z } =:C . Observe that A = (1 + t2)(1 − t −2 2 ) aX−1 i=0 t i 1 + (1 − t −2 2 ) X b−1 j=0 t a+j 1 (30) = (1 − t −1 2 )(1 + t −1 2 )(1 + t2) aX−1 i=0 t i 1 + (1 − t −2 2 ) X b−1 j=0 t a+j 1 and B = (t −1 2 + t −2 2 ) Xa i=1 t −i 1 + X b j=1 t −(a+j) 1 t −1 2 (31) = t −1 1… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Virtual Euler characteristic for d = 1, 2 and low m 24 [PITH_FULL_IMAGE:figures/full_fig_p024_4.png]
Figure 5
Figure 5. Figure 5: Virtual Euler characteristic for d = 3, 4 and low m Note that when d = 1 and for the cases where d is general and n ≤ d + 1, we can determine the Euler characteristic via different methods. When d = 1. For S = P 2 , consider the forgetful map S [0,m] d → Hilbd,(Z, D) 7…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

5 extracted references · 5 canonical work pages

  1. [1]

    Fantechi and L

    [FG10] B. Fantechi and L. G¨ ottsche. ‘Riemann-Roch theorems and elliptic genus for virtually smooth schemes’. In: Geom. Topol. 14 (2010), pp. 83–115. [G¨ ot90] L. G¨ ottsche. ‘The Betti numbers of the Hilbert scheme of points on a smooth projective surface’. In: Math. Ann. 286 (1990), pp. 193–207. [GP99] T. Graber and R. Pandharipande. ‘Localization of v...

    work page 2010

  2. [7]

    [Tho99] R.P

    url: http://dx.doi.org/10.1017/CBO9781107279544.007. [Tho99] R.P. Thomas. ‘A holomorphic Casson invariant for Calabi-Yau 3-folds, and bundles on K3 fibrations’. In: Journal of Differential Geometry 53 (1999), pp. 367–438. 29

    work page doi:10.1017/cbo9781107279544.007 1999

  3. [17]

    [Lim21] W Lim

    url: http://dx.doi.org/10.14231/AG-2014-017. [Lim21] W Lim. ‘Virtual chi y-genera of Quot schemes on surfaces’. In: J. Lond. Math. Soc. 104 (2021), pp. 1300–1341. doi: 10.1112/jlms.12460 . url: https://doi.org/10.1112/ jlms.12460. [Lin18] Y. Lin. ‘Moduli spaces of stable pairs’. In: Pacific J. Math. 294 (2018), pp. 123–158. [MNOP] D. Maulik, N. Nekrasov, ...

    work page doi:10.14231/ag-2014-017 2021

  4. [143]

    Pandharipande and A

    [PP13] R. Pandharipande and A. Pixton. ‘Descendent theory for stable pairs on toric 3-folds’. In: J. Math. Soc. Japan 65 (2013), pp. 1337–1372. [PT09a] R. Pandharipande and R.P. Thomas. ‘Curve Counting via Stable Pairs in the Derived Category’. In: Invent. math. 178 (2009), pp. 407–477. [PT09b] R. Pandharipande and R.P. Thomas. ‘The 3-fold vertex via stab...

    work page 2013

  5. [411]

    282–333.doi: 10.1017/cbo9781107279544

    Cambridge University Press, 2014, pp. 282–333.doi: 10.1017/cbo9781107279544

    work page doi:10.1017/cbo9781107279544 2014

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.