Graded Ehrhart theory and toric geometry

Ian Cavey

REVIEW 1 major objections 3 minor 6 references

Graded Ehrhart theory and toric geometry

T0 review · 1 major / 3 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 The harmonic algebra of a lattice polytope is not always finitely generated: the triangle (0,0),(7,56),(-45,30) is a concrete counterexample, established by identifying the harmonic algebra with a quotient of the blowup section ring.

desk verdict Theorems 1.2 and 1.4 are clean and useful, but Example 4.1's non-finite-generation argument has a divisibility gap that leaves Corollary 1.5 unproved. read the letter →

arxiv 2508.19176 v1 pith:6TJK3XSK submitted 2025-08-26 math.CO math.AG

Ian Cavey This is my paper · ORCID

classification math.COmath.AG MSC 05A1552B2014M25

keywords q-EhrhartseriesharmonicalgebralatticepolytopeEhrharttheoryassociatedgradedtoricvarietyblowupfinitegeneration

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 establishes two structural descriptions of the harmonic algebra of a lattice polytope, a bigraded algebra whose Hilbert series is the q-Ehrhart series. First, the harmonic algebra is the associated graded algebra of the semigroup algebra filtered by order of vanishing at the torus identity e=(1,...,1). Second, it is the quotient of the section ring of the blowup of the toric variety at e by the canonical section of the exceptional divisor. Using an existing example of a non-finitely generated section ring of a blowup of a weighted projective plane, the paper shows that for the lattice triangle with vertices (0,0),(7,56),(-45,30) the harmonic algebra is not finitely generated. This refutes the finite-generation part of the main conjecture in q-Ehrhart theory, while the weaker rationality conjecture remains open.

What carries the argument

The central mechanism is the descending filtration F_{m,d} on the semigroup algebra A_P, where F_{m,d} consists of Laurent polynomials supported on mP that vanish to order at least d at e=(1,...,1). The harmonic algebra is shown to be the associated graded algebra gr A_P = ⊕ F_{m,d}/F_{m,d+1}. Geometrically, the same filtered pieces are identified with sections H^0(Bl_e X_P, O(mH - dE)) of the blowup section ring R_P, and the quotient by the canonical section s of the exceptional divisor collapses the filtration to the graded pieces, giving H_P ≅ R_P/(s). Non-finite generation transfers from the known non-finite generation of R_P for the blowup of the weighted projective plane P(15,26,7) at

What would settle it

Take the triangle with vertices (0,0),(7,56),(-45,30) and compute the bigraded pieces (H_P)_{m,d} for m,d up to, say, 200. If a finite set of low-bidegree generators accounts for every piece, the algebra is finitely generated and the paper's main conclusion fails; likewise, exhibiting one Laurent polynomial supported on mP with order of vanishing at e at least (104/105)m that is not divisible by y-1 would contradict the stable-base-locus fact on which the counterexample depends.

Watch

Extended reading notes

Core claim

The paper proves Theorem 1.2: for any lattice polytope P, the harmonic algebra H_P is isomorphic, as a bigraded algebra, to gr A_P, the associated graded algebra of the semigroup algebra A_P filtered by order of vanishing at e=(1,...,1). It then proves Theorem 1.4: H_P is isomorphic to R_P/(s), where R_P is the bigraded section ring of the blowup of the toric variety X_P at the point e, and s is the canonical section of the exceptional divisor E. The applied conclusion is Corollary 1.5: there exist lattice triangles, for example the triangle with vertices (0,0),(7,56),(-45,30), for which H_P is not finitely generated, disproving the finite-generation conjecture for harmonic algebras.

Load-bearing premise

The non-finite-generation conclusion stands or falls with the imported stable-base-locus facts about the blowup of the projective plane P(15,26,7): if the divisor H-(104/105)E does not actually put the curve y-1=0 in its stable base locus, or the three order-of-vanishing properties fail for some m, then the counterexample triangle may not have a non-finitely generated harmonic algebra.

Editorial extensions

If this is right

The harmonic algebra of a lattice polytope is not guaranteed to be finitely generated; the triangle (0,0),(7,56),(-45,30) is a concrete counterexample to the finite-generation conjecture.
Theorem 1.2 makes the harmonic algebra computable in principle from vanishing orders of Laurent polynomials supported on dilations of P, giving a combinatorial handle on the q-Ehrhart coefficients.
Theorem 1.4 transfers any non-finite-generation result for section rings of blowups of toric varieties at the identity point into a non-finite-generation result for harmonic algebras.
The rationality of the q-Ehrhart series remains open; the non-finite-generation result does not settle it.

Reading between the lines

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

An elementary proof of non-finite generation for the triangle example may be reachable: the filtration description reduces the question to whether elements of bidegree (km,kd) for d/m in a certain interval can be generated by finitely many low-degree elements, a purely polynomial divisibility statement.
Rationality of the q-Ehrhart series could still hold even when H_P is not finitely generated; the geometric interpretation suggests searching for rationality through the structure of the section ring R_P rather than through finite generation of H_P.
The same geometric dictionary predicts that finite generation of H_P is governed by the effective cone of the blowup: polytopes whose blowup section rings contain non-finitary stable-base-locus behavior should give further non-finitely generated examples, and lattice triangles may be classified by such cone geometry.
The relationship between harmonic spaces and lowest-degree parts of spans of (1+x)^a suggests that q-Ehrhart invariants could be studied purely from point configurations, without passing through toric geometry, potentially leading to algorithms for arbitrary lattice polytopes.

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 structural results are the real news here. Theorem 1.2, identifying HP with the associated graded of AP under the order-of-vanishing filtration at e, is clean and non-obvious; the proof via Proposition 3.1 is short and correct. Theorem 1.4, the quotient R_P/(s) description, follows directly from the filtration and is also correct. Anyone working in q-Ehrhart theory or harmonic algebras will want these.

The trouble is the advertised resolution of Reiner–Rhoades Conjecture 1.1(i). It hangs entirely on Example 4.1, and the final paragraph of that example doesn't hold together. The argument needs to show that a finitely generated subalgebra of HP can't contain all elements of certain bidegrees. It uses properties 1–3, which are statements about Laurent polynomials f supported on mP. Property 3 gives f not divisible by y−1 with order kd at e. But to get an element of HP not divisible by y (the shifted variable), you need the initial form of f, i.e., the lowest-degree homogeneous part after shifting, to be not divisible by y. That does not follow from f not being divisible by y−1. Example: (y−1)+(x−1)^2 is not divisible by y−1, yet its degree-1 initial form is y. So the elements produced by property 3 could still be y-divisible in HP, and the contradiction with the subalgebra's α doesn't occur. This is not a minor exposition slip; it's the load-bearing step for Corollary 1.5. The geometric facts from Gonzalez–Karu are about sections and the curve C, not about their jets at e, and the passage to the associated graded loses that distinction.

The paper is honest that it relies on [4] and that an elementary proof is unknown. But the problem is internal: the stated facts don't imply the claimed conclusion.

Net: the structure is good, the final result is unproved as written. A referee should ask for a repair of Example 4.1, or a different proof of non-finite generation. The paper is worth taking seriously and sending to review, but not accepting in its current form.

Referee Report

1 major / 3 minor

Summary. The paper studies the harmonic algebra H_P of a lattice polytope P, a bigraded algebra whose bigraded Hilbert series is the q-Ehrhart series of Reiner and Rhoades. The main positive results are Theorem 1.2, which identifies H_P with the associated graded algebra gr A_P of the semigroup algebra A_P with respect to the filtration by order of vanishing at e=(1,...,1), and Theorem 1.4, which identifies H_P with the quotient R_P/(s) of the blowup section ring by the canonical section of the exceptional divisor. The paper then uses an example of Gonzalez and Karu, presented in Example 4.1, to claim that there exist lattice triangles P for which H_P is not finitely generated (Corollary 1.5), thereby disproving Conjecture 1.1(i) of Reiner and Rhoades.

Significance. If the non-finite-generation claim is correct, the paper resolves a central conjecture in graded Ehrhart theory and gives a new bridge between q-Ehrhart theory and birational geometry. The algebraic description in Theorem 1.2 is clean and self-contained, and Theorem 1.4 gives a natural geometric interpretation of H_P. These positive results are likely to be useful regardless of the fate of the counterexample. However, the counterexample proof in Example 4.1 has a serious gap, and since Corollary 1.5 is the main advertised resolution of the conjecture, the paper is not yet suitable for publication in its current form.

major comments (1)

[Example 4.1, final paragraph] The step from properties 1-3 to non-finite-generation of H_P conflates divisibility in A_P with divisibility in H_P. Property 3 gives a Laurent polynomial f supported on kmP that is not divisible by y-1 and has order kd at e. But the element of H_P represented by f is the lowest-degree homogeneous part of f(1+u,1+v) in the shifted variables u=x-1, v=y-1; divisibility of f by y-1 in the Laurent polynomial ring does not imply, and is not implied by, divisibility of this initial form by v. For example, f=(y-1)+(x-1)^2 = v+u^2 is not divisible by y-1 and has order 1 at e, yet its image in H_P is v, which is divisible by the class of y-1. Thus property 3 does not yield elements of H_P whose classes are outside the subalgebra generated by elements of smaller d/m. Since the final paragraph relies exactly on this implication, the proof of Corollary 1.5 is incomplete.

minor comments (3)

[Example 4.1] The rational triangle is said to have vertices (0,0), (2/15,6/15), and (-6/7,4/7), but the displayed figure and the subsequent integral triangle (7,56),(-45,30) indicate that the second vertex should be (2/15,16/15). Please correct this typo.
[Abstract and Introduction] Typos: 'it's relationship' should be 'its relationship', 'one the main conjectures' should be 'one of the main conjectures', and 'algberas' in Theorem 1.4 should be 'algebras'.
[Example 1.3] The displayed q-Ehrhart series '1+(1+2q+1)t+...' appears to be missing a q^2 term; the coefficient of t should be 1+2q+q^2 according to the filtration dimensions given immediately above.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: main derivations are self-contained; external citations are independent evidence.

full rationale

The paper's central structural theorems are derived from definitions and from Proposition 3.1, which is proved directly via Lemma 3.2; there is no fitted parameter, no target conclusion assumed as input, and no equation whose definition already contains the result it is said to predict. Theorem 1.2 identifies H_P with gr A_P through a genuine proof using the lowest-degree homogeneous parts of formal power series, and Theorem 1.4 follows from the geometric identification of the filtration F_{m,d} with sections vanishing along E; this is a rephrasing with independent content, not a renaming of H_P as its own conclusion. The non-finite-generation example relies on imported stable-base-locus facts from Gonzalez-Karu, which the paper explicitly acknowledges: 'Our argument for non-finite generation depends on the geometric results of Gonzalez and Karu. It would be interesting to give an elementary proof...' That is an external, cited mathematical input, not a self-citation or an ansatz smuggled in by the present author. The paper does not cite itself load-bearingly, and the conjecture of Reiner and Rhoades is being tested, not assumed, via an independent geometric counterexample. Any concern about whether divisibility in A_P transfers correctly to H_P is a mathematical correctness issue, not a circularity pattern under the specified criteria. Therefore no circular step is identified.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

No free parameters or invented entities appear; the paper is pure mathematics. The main external inputs are standard toric geometry, the prior harmonic algebra framework of Reiner-Rhoades, and the stable-base-locus theorem of Gonzalez-Karu.

assumptions (6)

standard math Standard toric-geometric identification A_P ≅ ⊕_{m≥0} H^0(X_P, O(mH))
Used in equation (3) and in the proof of Theorem 1.4 to translate between Laurent polynomials supported on mP and global sections.
domain assumption For a finite set Z, the space W_Z is the span of e^{a·x}, and V_Z is its set of lowest-degree homogeneous parts (Corollary 4.4 and Lemma 4.6 of Reiner-Rhoades [6])
Used in the proof of Proposition 3.1; these are prior results from the paper that introduced harmonic algebras.
domain assumption The harmonic spaces assemble into a bigraded algebra H_P with Hilbert series E_P(t,q) (Proposition 5.4 of Reiner-Rhoades [6])
This is the background structure the paper builds on; it is needed to identify H_P with gr A_P as an algebra, not just as a vector space.
standard math Basic blowup facts: O(E) has a canonical section s, and for k>0 the line bundle O(mH+kE) restricts to O_E(-k), so all its global sections vanish along E to order k
Used in the proof of Theorem 1.4 to justify that the negative-degree pieces of R_P/(s) vanish and that the quotient is H_P.
domain assumption Gonzalez-Karu facts for Bl_e P(15,26,7): the curve C has class H-E on the boundary of the effective cone, H-(104/105)E has C in its stable base locus, and properties 1-3 in Example 4.1 hold
Load-bearing for Corollary 1.5. The paper imports these from [4] and does not prove them.
ad hoc to paper The rational triangle P in Example 4.1 is a convenient representative, and its non-finite-generation properties transfer to the minimal integral triangle asserted in Corollary 1.5
Example 4.1 switches to a rational dilation without giving the full scaling or Veronese-subring argument for the transfer to the integral triangle.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Graded Ehrhart theory and toric geometry." pith.science (2026). https://pith.science/paper/6TJK3XSK

@misc{pith2026250819176,
  author       = {Pith},
  title        = {Pith review of: Graded Ehrhart theory and toric geometry},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6TJK3XSK}},
  note         = {Machine review of arXiv:2508.19176}
}

abstract

We give two new constructions of the harmonic algebra of a lattice polytope $P$, a bigraded algebra whose character is the $q$-Ehrhart series of $P$ defined by Reiner and Rhoades. First, we show that the harmonic algebra is the associated graded algebra of the semigroup algebra of $P$ with respect to a certain natural filtration, clarifying it's relationship with the more classical semigroup algebra. We then give a geometric interpretation of the harmonic algebra as a quotient of the ring of global sections of a certain family of line bundles on the blowup of the toric variety associated to $P$ at a generic point. Using this connection to toric geometry we resolve one the main conjectures of Reiner and Rhoades by showing that the harmonic algebra is not finitely generated in general.

Figures

Figures reproduced from arXiv: 2508.19176 by the authors.

**Figure 1.** The lattice triangle P triangle P contains 4 integer points so we have EP (t) = 1+4t+· · · . Using the definition (2), the degree-one piece of the semigroup algebra is the linear span (AP )1 ≃ ⟨1, xy, x2 y, xy2 ⟩, where we write x = x1 and y = x2, and drop the grading parameter z. Let us write F1,d for the subset of (AP )1 consisting of polynomials that vanish at e with order at least d. The filtration (AP )1 = F1,0… view at source ↗

**Figure 2.** A rational triangle P As an intermediate step towards showing that RP is not finitely generated, Gonz´alez and Karu show that the nef divisor D = H − 104 105E contains C in its stable base locus. Concretely, the geometric facts listed above mean that for any positive integer m: 1. The maximum order of vanishing at e of a Laurent polynomial supported on mP is exactly m, which is attained by (y − 1)m. 2. Any Laurent p… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

6 extracted references · 5 canonical work pages

[4]

Some non-finitely generated Cox rings.Compositio Mathematica, 152(5):984–996, 2016

Jos´ e Luis Gonz´ alez and Kalle Karu. Some non-finitely generated Cox rings.Compositio Mathematica, 152(5):984–996, 2016

work page 2016
[1]

The geography of negative curves

Javier Gonz´ alez Anaya, Jos´ e Luis Gonz´ alez, and Kalle Karu. The geography of negative curves. Michigan Math. J. , 75(1):89–118, 2025

work page 2025
[2]

Symbolic algebras of monomial primes

Steven Dale Cutkosky. Symbolic algebras of monomial primes. J. Reine Angew. Math. , 416:71–89, 1991

work page 1991
[3]

Introduction to Toric Varieties

William Fulton. Introduction to Toric Varieties. (AM-131) . Princeton University Press, 1993

work page 1993
[5]

Non-cohen-macaulay symbolic blow- ups for space monomial curves and counterexamples to cowsik’s question

Shiro Goto, Koji Nishida, and Kei-Ichi Watanabe. Non-cohen-macaulay symbolic blow- ups for space monomial curves and counterexamples to cowsik’s question. Proceedings of the American Mathematical Society , 120(2):383–392, 1994

work page 1994
[6]

Harmonics and graded Ehrhart theory

Victor Reiner and Brendon Rhoades. Harmonics and graded Ehrhart theory. arXiv:2407.06511, 2024. 8

arXiv 2024

Pith tools

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

[1] [4]

Some non-finitely generated Cox rings.Compositio Mathematica, 152(5):984–996, 2016

Jos´ e Luis Gonz´ alez and Kalle Karu. Some non-finitely generated Cox rings.Compositio Mathematica, 152(5):984–996, 2016

work page 2016

[2] [1]

The geography of negative curves

Javier Gonz´ alez Anaya, Jos´ e Luis Gonz´ alez, and Kalle Karu. The geography of negative curves. Michigan Math. J. , 75(1):89–118, 2025

work page 2025

[3] [2]

Symbolic algebras of monomial primes

Steven Dale Cutkosky. Symbolic algebras of monomial primes. J. Reine Angew. Math. , 416:71–89, 1991

work page 1991

[4] [3]

Introduction to Toric Varieties

William Fulton. Introduction to Toric Varieties. (AM-131) . Princeton University Press, 1993

work page 1993

[5] [5]

Non-cohen-macaulay symbolic blow- ups for space monomial curves and counterexamples to cowsik’s question

Shiro Goto, Koji Nishida, and Kei-Ichi Watanabe. Non-cohen-macaulay symbolic blow- ups for space monomial curves and counterexamples to cowsik’s question. Proceedings of the American Mathematical Society , 120(2):383–392, 1994

work page 1994

[6] [6]

Harmonics and graded Ehrhart theory

Victor Reiner and Brendon Rhoades. Harmonics and graded Ehrhart theory. arXiv:2407.06511, 2024. 8

arXiv 2024