REVIEW 1 major objections 6 minor
The mysterious bivariate polynomials that proved real-rootedness for M̄0,n are stratified virtual Poincaré polynomials.
Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →
T0 review · grok-4.5
2026-07-30 14:40 UTC pith:EHGJVKPW
load-bearing objection Clean geometric meaning for the BK26 bivariate polynomials via new inductive formulas for stratified motivic invariants of M̄0,n and Y[n]. the 1 major comments →
Stratified motivic invariants and bivariate deformations of Poincar\'e polynomials
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
The bivariate deformations Fn(y,t) and eFn(y,t) of the even-degree Poincaré polynomials of M̄0,n+1 and P1[n] that appeared in earlier work are precisely the stratified virtual Poincaré polynomials of those spaces (up to the elementary factor y in the first case). Both families of stratified invariants are uniquely determined by explicit first-order inductive formulae coming from solid echelon towers.
What carries the argument
A solid echelon tower: a sequence of stratified varieties X[n] linked by morphisms πn that send the open codimension-j stratum into the union of the codimension-j and codimension-(j-1) strata of X[n], with the two restricted maps motivically locally trivial and of constant fibre invariant. The constant-fibre condition converts the stratification into a linear recurrence for the generating function FX[n](y).
Load-bearing premise
The restricted forgetful maps on open dual-graph strata must be motivically locally trivial with fibres of constant motivic invariant; without that constancy the inductive formulae fail.
What would settle it
Compute the stratified virtual Poincaré polynomial of M̄0,6 or P1[4] both from the inductive formula and by direct enumeration of dual-graph strata; any mismatch of coefficients falsifies the claim that the earlier bivariate polynomials coincide with the stratified invariants.
If this is right
- The coefficients of the earlier bivariate polynomials are now identified as virtual Poincaré polynomials of explicit strata, giving them independent geometric content.
- Specialising the inductive formulae at y=1 recovers all previously known generating functions for the ordinary Poincaré polynomials of M̄0,n and Y[n] by elementary calculus.
- The same echelon-tower formalism applies verbatim to any other sequence of moduli spaces whose forgetful maps satisfy the solid-fibre condition.
- The paper notes that the ideas may extend to Chow rings of loopless matroids and their Poincaré polynomials.
Where Pith is reading between the lines
- If the solid-echelon condition can be verified for other compactified configuration spaces or for Hassett spaces, the same inductive machine would immediately produce stratified invariants and candidate real-rooted polynomials.
- The failure of solidity for Mg,n (g>0) and for stable-map spaces suggests that a weaker “virtual-fibre” version of the tower might still control generating functions in those settings.
- Once the coefficients are known to be stratum-wise virtual Poincaré polynomials, positivity or unimodality questions about those strata become equivalent to coefficient-wise questions about Fn.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper introduces, for a stratified quasi-projective variety X and a motivic invariant H, the stratified invariant F_X(y) = Σ y^j H_{X°_j}, which interpolates between H_{X°} (y=0) and H_X (y=1). It axiomatizes families of stratified moduli spaces connected by forgetful morphisms as "echelon towers" (Definition 2.3), with a "solidity" condition (motivic local triviality of the two restricted maps π_{n,j,0}, π_{n,j,1}) that yields the inductive identity (7). The two main theorems compute the fiber invariants explicitly for the moduli spaces M̄_{0,n} (Proposition 3.1, Eq. (12)) and the Fulton–MacPherson compactifications Y[n] (Eqs. (18)–(20)), producing differential recursions (13) and (21). As an application, the bivariate polynomials F_n(y,t) and F̃_n(y,t) introduced ad hoc in the companion paper [BK26] to prove real-rootedness of the even-degree Poincaré polynomials of M̄_{0,n+1} and P^1[n] are identified with stratified virtual Poincaré polynomials (Corollaries 3.3 and 4.4). A characteristic-method solution of the generating-function PDE (24) yields a new elementary proof of Manin's formulas for H_{Y[n]} (Theorem 4.5).
Significance. The paper gives a genuine geometric interpretation of the bivariate deformation that appeared "out of nowhere" in the AI-assisted real-rootedness proof of [BK26]: the coefficients of F_n(y,t²) are exactly the virtual Poincaré polynomials of the dual-graph strata (Eq. (15)). This is a satisfying and non-obvious demystification. Independently of the companion paper, the recursions (13) and (21) are effective, graph-sum-free computational tools; the paper earns credibility by reproducing known results in checkable cases — F_{M̄0,5} at y=1 gives the degree-5 del Pezzo polynomial L²+5L+1 with the correct stratum-by-stratum count (10(L−2) from ten boundary divisors, 15 from codimension-2 strata), the n=2 Fulton–MacPherson example matches the blowup Y[2] = bl_Δ(Y²), and specialization to y=1 recovers Manin's formulas with a fresh characteristic-curve proof. I independently verified the load-bearing computations: the fiber class (j+1)L−j−n+1 in Proposition 3.1 equals H_C − (n+j) with H_C = (j+1)(L+1)−j; Eq. (18) follows from H_{Y′} = H_Y + j(H_{P^m}−1); the characteristic computation (26)–(31) is correct line by line. The results are correct as far as I can check and should be of interest
major comments (1)
- [§2.2, Definition 2.3 and its use in Proposition 3.1 and §4 (Eqs. (17)-(20))] Solidity — the motivic local triviality of π_{n,j,0} and π_{n,j,1} — is the hypothesis that makes (7), and hence the recursions (13) and (21), valid; it is therefore load-bearing for both main theorems. However, its verification is compressed to one sentence in each case ('by (8), we can decompose (11) ... and find that the morphisms in (11) are motivically locally trivial'; similarly after (17)). The fiber motivic invariants are computed explicitly and check out (I verified all of (12), (18), (19) independently), but local triviality itself is not argued. For M̄_{0,n} one expects: over each stratum M_{0,n,γ} the restriction of the universal curve with its special sections removed is Zariski locally trivial because, after normalization, the family is obtained by gluing trivial families of pointed P¹'s, and the attaching sections remain disjoint within the stratum. For (19), one needs tha
minor comments (6)
- [§4, Example after (22)] The formula for F_{Y[3]} contains H_{P^{d−1}} (twice), with d undefined; this should be H_{P^{m−1}}.
- [§4, Eqs. (28), (30), (31)] The exponent 'Lm−1' is typographically ambiguous between L^{m}−1 and L^{m−1}. Solving the characteristic equation myself confirms the intended meaning is L^{m}−1 (e.g., (28) reads y/(y+L−1) = (1/L)(z/u)^{L^m−1}); please add parentheses to remove the ambiguity.
- [§3, Eq. (10) and Proposition 3.1] Notation oscillates between M_{0,n,j} and M̄_{0,n,j} for the same locally closed strata; since M_{0,n,γ} denotes the open locus inside M̄_{0,n}, a uniform convention would help.
- [Remark 3.4] The assertion that M̄_{g,n} for g≥1 forms a non-solid echelon tower would benefit from one line of explanation: the fiber of π_{n,j,0} over C is still C minus special points with constant motivic class, so the failure is genuinely in Zariski local triviality of the universal family over strata (smooth higher-genus curves have moduli). As stated, a reader might mistakenly think the fiber invariant itself fails to be constant.
- [§1 and Corollary 3.3] In Corollary 3.3 the identification F_n(y,t²) = yF_{M̄0,n+1}(y) implicitly uses R = Z[t] with L = t²; stating this substitution explicitly (it is only explained via P^even_n(t²) = P_{M̄0,n}(t)) would make the corollary self-contained.
- [Typos] §4: 'autormorphism group' (twice); §4 after (19): 'the union of of n+j−1 copies'; §2.2: 'motivially locally trivialif it factors'. Also [BK26] should include the arXiv identifier in the bibliography as it currently does; fine, but 'poincaré' is mis-capitalized in the [BK26] title.
Simulated Author's Rebuttal
We thank the referee for the careful and engaged report, for the positive assessment of the results, and especially for independently verifying the load-bearing computations (Proposition 3.1, Eq. (18), and the characteristic computation (26)–(31)). We are glad the referee finds the geometric interpretation of the bivariate deformations of [BK26] satisfying. The referee's sole major comment concerns the verification of solidity, which is indeed load-bearing for both main theorems and is currently compressed to one sentence in each case. We agree this deserves a fuller treatment and will expand the arguments in the revision, as detailed below.
read point-by-point responses
-
Referee: Solidity (motivic local triviality of pi_{n,j,0} and pi_{n,j,1}) is load-bearing for (7) and hence for (13) and (21), but its verification is compressed to one sentence in each case. The fiber invariants are computed explicitly and check out, but local triviality itself is not argued. For M̄_{0,n} one expects an argument via normalization: over each stratum the universal curve is glued from trivial families of pointed P^1's, and the attaching sections remain disjoint within the stratum. For (19) one needs a similar argument.
Authors: We agree entirely. Solidity is the hypothesis that makes identity (7), and therefore both recursions, valid, so it should be argued rather than asserted. In the revision we will add a dedicated lemma in each section giving the full local triviality argument along exactly the lines the referee sketches. For M̄_{0,n}: over the stratum M_{0,n,γ} the restriction of the universal stable curve, after normalization, decomposes as a disjoint union of constant families of pointed P^1's (one per vertex of γ), since the moduli of a k(v)-pointed P^1 are rigid once the tree structure is fixed; the attaching sections stay disjoint within the stratum because the dual graph is constant, so π_{n,j,0} is Zariski locally trivial with fiber a pointed P^1-tree minus its n+j special sections, and π_{n,j,1} is locally trivial with fiber the finite set of special points (n+j−1 of them). For Y[n]: over each stratum Y[n]°_γ the family of degenerate varieties Y′ is locally trivial because γ fixes the blowup tree and the blowup centers and gluing loci vary in disjoint families; the fiber of π_{n,j,0} is the complement of the marked points and the P^{m−1}-intersections, and the fiber of π_{n,j,1} is the union of n+j−1 copies of P^{m−1} arising as C*-quotients as described before (19). We will also spell out explicitly why the complement description gives a motivic (not merely Zariski) local trivialization in the sense of Definition 2.3. The fiber invariant computations (12), (18), (19), which the referee verified, remain unchanged; the revision adds the missing triviality argument. revision: yes
Circularity Check
No significant circularity: stratified invariants and inductive formulae are derived from fiber geometry; match to [BK26] is a post-hoc identification, not an input.
full rationale
The load-bearing chain is self-contained. Stratified invariants F_X are defined from the stratification (Def. 2.1). Solid echelon towers are defined by motivically locally trivial restrictions of forgetful maps with constant fiber invariants (Def. 2.3). For M̄0,n the fibers are computed explicitly: H of πn,j,0-fiber = (j+1)L−j−n+1 and of πn,j,1-fiber = n+j−1 (Prop. 3.1), yielding the stratum recurrence (12) and, by elementary generating-function algebra, the inductive formula (13) with F_M̄0,3=1. The same pattern for Y[n] uses the blowup-and-glue fiber counts (18)–(19) to obtain (21). These derivations cite only standard dual-graph geometry and motivic local triviality; they do not presuppose the bivariate polynomials of [BK26]. Corollary 3.3 (resp. 4.4) then observes that the independently obtained recurrence for y F_M̄0,n+1 (resp. F_P1[n]) coincides with the defining recurrence of Fn (resp. eFn) in the companion paper. That is a comparison of two objects, one geometric and one previously introduced by formula, not a derivation that takes the target as input. Self-citation of [BK26] is limited to naming the object being identified and to the real-rootedness application; it is not load-bearing for the inductive formulae or the fiber calculations. Manin’s formulae are recovered as a specialization (y=1) of the same PDE, again without circular dependence. No fitted parameters, no uniqueness imported to forbid alternatives, and no renaming of an empirical pattern as a first-principles result.
Axiom & Free-Parameter Ledger
axioms (3)
- standard math A motivic invariant H is a ring homomorphism from the Grothendieck ring of quasi-projective varieties to an integral domain R with H(pt)=1 and L=H(A^1) invertible.
- domain assumption The dual-graph stratification of M̄0,n (resp. Y[n]) has smooth strata of pure codimension equal to the number of edges.
- domain assumption The restricted forgetful maps πn,j,0 and πn,j,1 are motivically locally trivial with constant fiber invariants An,j and Bn,j.
invented entities (2)
-
stratified motivic invariant F_X(y)
no independent evidence
-
echelon tower (solid)
no independent evidence
read the original abstract
For a stratified variety $X$ and a motivic invariant $\mathbb{H}$, we consider the stratified invariant which interpolates the invariant of $X$ and that of its interior $X^{\circ}$. Based on observations in moduli theory, we introduce the notion of an echelon tower of stratified varieties and then prove explicit inductive formulae for the stratified invariants of the moduli spaces $\overline{M}_{0,n}$ of stable curves of genus $0$ and the Fulton-MacPherson varieties $Y[n]$ for any smooth projective variety $Y$. Using these, we show that the mysterious bivariate deformations of the even degree Poincar\'e polynomials of $\overline{M}_{0,n+1}$ and $\mathbb{P}^1[n]$ in \cite{BercziKiem2026} are nothing but stratified virtual Poincar\'e polynomials.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.