{"id":"77207b53-2727-4a7c-927f-d700d8d32d47","arxiv_id":"2607.15886","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Normalized Ehrhart h*-polynomials form 'h*-distributions' whose mean/variance are fixed by Ehrhart polynomial coefficients and whose dilation limit is the Eulerian distribution.","lead":"A lattice polytope's Ehrhart h*-polynomial becomes a probability distribution when normalized by volume. The paper derives the mean and variance of these distributions from Ehrhart coefficients and shows that dilating a polytope drives the distribution to the Eulerian (descent) distribution.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4.4's variance lower bound miscomputes E[X]^2, making condition (7) vacuous; the asymptotic-normality theorem needs a corrected derivation.","rationale":"The reader's weakest assumption correctly identifies the miscomputed variance lower bound in Theorem 4.4 as the most serious issue. This is a genuine error: replacing E[X]^2 with 1/4 when E[X] ≈ d/2 introduces a Θ(d^2) error, invalidating the displayed sufficient condition (7). The corrected bound is straightforward and does salvage Corollary 4.5, so the paper should be CONDITIONAL on repairing Theorem 4.4, exactly as the reader concluded. I also note the off-by-one root-count problem in Theorem 3.21's proof (d+1 roots for a degree-≤d polynomial), but this appears to be a repairable indexing error rather than a false conclusion; the Eulerian limit is independently supported by the mean/variance asymptotics and by known results. The manuscript's lack of code/seed for Figure 4 is a reproducibility weakness but not central to the mathematical claim. Overall, the reader's verdict of CONDITIONAL with high confidence is appropriate; no verdict adjustment is needed.","tokens_in":14107,"tokens_out":22714,"duration_ms":212032,"concrete_test":"Independently re-derive the variance lower bound in Theorem 4.4: with E[X] ∈ [d/2, (d+1)/2], minimize (d+1)E − E^2 on this interval and substitute into Theorem 3.5's variance formula. If the resulting bound is Var ≥ 2c_{d−2}/(d(d−1)c_d) + (d−2)/12, then condition (7) as written is vacuous — its (1/12)(3d^2+d−5) term makes the limit automatically infinite — and the theorem's statement must be revised to use the corrected bound.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's asymptotic-normality contribution, Theorem 4.4, is internally flawed. From Theorem 3.5, Var[X_P] = A + (d+1)E[X] − (d+1)(3d+2)/12 − E[X]^2, with A = 2(d−2)!c_{d−2}/Vol(P). Under B(P) ≤ Vol(P), one has E[X] ≥ d/2, but the proof then replaces E[X]^2 by 1/4. Since E[X] is Θ(d), this discards a Θ(d^2) term and produces a lower bound that is too large. Consequently, the sufficient condition (7) contains the spurious term (1/12)(3d^2+d−5), which diverges regardless of the Ehrhart coefficients; the condition is vacuous. The correct lower bound, obtained by minimizing (d+1)E − E^2 over E ∈ [d/2, (d+1)/2], is Var ≥ 2c_{d−2}/(d(d−1)c_d) + (d−2)/12. This corrected bound is what actually supports Corollary 4.5, assuming c_{d−2} ≥ 0. As written, Theorem 4.4 would assert asymptotic normality for every real-rooted B≤Vol sequence, which is not justified and likely false without control on c_{d−2}. This is a load-bearing flaw in a stated contribution, not merely a cosmetic typo.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper defines the h*-distribution of a lattice polytope by normalizing the coefficient vector of its h*-polynomial, and studies this distribution: closed-form mean and variance in terms of Ehrhart coefficients and boundary volume (Thm 3.5), convergence of the h*-distributions of dilates to the Eulerian distribution (Thm 3.21), tail inequalities for real-rooted h*-polynomials with applications to reflexive polytopes, and a sufficient condition for asymptotic normality of sequences of real-rooted h*-distributions (Thm 4.4, Cor 4.5). It also gives explicit formulas for zonotopes and connects the results to Pitman-Stanley polytopes and permutation statistics.","tokens_in":14380,"tokens_out":22958,"duration_ms":174565,"significance":"The paper introduces a useful distributional perspective on h*-vectors, with parameter-free formulas for mean and variance (Thm 3.5) that are derived cleanly from the binomial-basis expansion of the Ehrhart polynomial. The cluster-point theorem (Thm 3.21) is true and the use of Beck-Stapledon root convergence is appropriate. The real-rooted tail inequalities and the applications to zonotopes and Pitman-Stanley polytopes are interesting. However, the asymptotic-normality theorem (Thm 4.4) as stated is not correctly proved, and its sufficient condition is vacuous; this is a central stated contribution and requires revision.","major_comments":[{"comment":"The proof of Theorem 4.4 lower-bounds Var[X_{P_j}] by replacing E[X]^2 with 1/4 after using E[X] ≥ d_j/2. This is invalid: E[X] is Θ(d_j), so E[X]^2 is Θ(d_j^2), and discarding it introduces the spurious (3d_j^2+d_j−5)/12 term in (7). That term diverges for any sequence with d_j→∞, so condition (7) is vacuous and the theorem as stated gives no control from the Ehrhart coefficients. The correct reduction minimizes (d+1)E − E^2 over E∈[d/2,(d+1)/2], yielding Var ≥ 2c_{d−2}/(d(d−1)c_d) + (d−2)/12. Theorem 4.4 and the proof of Corollary 4.5 should be restated with this corrected bound; the corollary remains supported when c_{d_j−2}≥0, but the current derivation is invalid.","section":"§4.2, Theorem 4.4 and Eq. (7)"},{"comment":"The proof asserts that h*(tP;z) has d+1 roots β_{t,1},…,β_{t,d+1}, but h*(tP;z) has degree at most d and hence at most d roots; the Eulerian polynomial as defined also has degree d (with one root at 0). The product formula and the limit of the normalized product are therefore not justified as written. The convergence statement itself is true, but the argument needs a rigorous derivation, e.g. by working with the reciprocal polynomial or by treating missing roots as tending to infinity before applying Vieta's formulas.","section":"§3.3, proof of Theorem 3.21"}],"minor_comments":[{"comment":"The proof says that finiteness of lattice simplices of bounded volume implies finiteness of h*-distributions for all lattice polytopes. The implication is not immediate; however, the conclusion follows from the simpler observation that each h_i^* ≤ Vol(P), so for fixed d and V there are at most (V+1)^{d+1} possible h*-vectors. Please replace the justification.","section":"Proposition 3.16"},{"comment":"The indexing ρ_{d+1}=0 is confusing, since a degree-d polynomial has d roots. Clarify that the Eulerian polynomial has a root at 0 and that the root set has d elements, or explain the extended-root convention.","section":"Theorem 3.21 notation"},{"comment":"The isolated point corresponding to (1,7,1) is described as a black 'Y', but the grayscale plot makes it hard to locate. Consider adding a legend or an arrow for this point.","section":"Example 3.3 / Figure 3"},{"comment":"Typos: 'EHRHARTh ∗-DISTRIBUTIONS' and 'HLA V ACEK' should be fixed; also the L= lcm line in Example 3.17 has a formatting break.","section":"Title and author block"}],"recommendation":"major_revision","confidential_remarks":"The main issue is the faulty proof of Theorem 4.4, which makes the stated sufficient condition (7) vacuous. This is a load-bearing gap in a stated contribution, but it is fixable within the manuscript's scope by correcting the variance lower bound; the rest of the paper is largely sound. The proof of Theorem 3.21 also needs a rigorous rewrite. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper makes a good case that normalizing h*-polynomials into probability distributions is a genuinely useful move. The mean/variance formulas and the Eulerian limit for dilates are correct and worth having. But the asymptotic-normality section has a real algebra error in the proof of Theorem 4.4, and Theorem 3.21's root-counting is sloppy. Both are fixable, but the paper as posted needs revision.\n\nWhat's actually new: the distributional viewpoint itself, Theorem 3.5 connecting mean/variance to c_{d-1}, c_{d-2}, and the moment characterization in Theorem 3.14. Theorem 3.21 is a natural corollary of Beck–Stapledon but the distributional formulation is clean and the application to wreath-product Eulerian statistics (Theorem 3.23) is cute. Section 4.1's inequalities via Pitman's tail bounds are a reasonable application.\n\nThe soft spots: In Theorem 4.4's proof, the authors lower-bound Var[X] by replacing E[X]^2 with 1/4. Since E[X] ≥ d/2, that is off by Θ(d^2) and condition (7) becomes vacuous. The corrected lower bound is Var ≥ 2c_{d−2}/(d(d−1)c_d) + (d−2)/12, which still yields Corollary 4.5 when c_{d−2} ≥ 0, but the theorem as stated needs that correction. Also, the proof of Theorem 3.21 lists d+1 roots for a degree-≤d polynomial; the indexing is inconsistent and should be cleaned up. Minor: the figures are not reproducible because the SageMath code and sampling seed are not included.\n\nOverall: the core of the paper is solid, the errors are repairable, and the distributional framework is likely to be useful to people working on Ehrhart positivity and unimodality.\n\nRecommendation: send it to a serious referee. It is not desk-reject material. If I were editing, I'd ask the authors to fix Theorem 4.4 and polish Theorem 3.21. A referee should verify the corrected bound and that Corollary 4.5 holds.","headline":"Useful distributional take on Ehrhart h*-polynomials with a correct core; the asymptotic-normality theorem has a fixable but real proof error.","tokens_in":14979,"tokens_out":5273,"would_cite":true,"duration_ms":42266,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52B20","05A15","60F05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every lattice polytope yields an h*-distribution, whose dilates converge to the descent-statistic distribution.","keywords":["Ehrhart h*-polynomial","lattice polytope","finite probability distribution","Eulerian distribution","real-rooted polynomial","asymptotic normality","zonotopes","reflexive polytopes"],"falsifier":"For P=[0,1]^d, the h*-distributions of tP are known explicitly as wreath-product Eulerian distributions; checking whether these normalized coefficients converge to the descent distribution as t→∞ at fixed d would directly test Theorem 3.21. For the asymptotic-normality criterion, evaluating the variance formula from Theorem 3.5 on a sequence of reflexive zonotopes with interior lattice points and checking whether the variance diverges would test Corollary 4.5.","tokens_in":13945,"feed_emoji":"📊","tokens_out":7777,"duration_ms":62496,"temperature":0.7,"pith_summary":"This paper defines the h*-distribution of a lattice polytope by normalizing the coefficients of its Ehrhart h*-polynomial to sum to one, turning an integer-encoding polynomial into a probability law. It proves closed formulas for the mean and variance of this distribution in terms of normalized boundary volume and the Ehrhart coefficient c_{d-2}, and it shows that moments of the distribution and Ehrhart coefficients determine each other through a triangular system. The central structural result is that the h*-distributions of dilates tP converge, as t tends to infinity, to the distribution of the descent statistic on permutations, regardless of the original polytope. The paper then specializes to real-rooted h*-polynomials, transferring tail bounds to obtain linear inequalities for h*-vectors (including a 2^d volume lower bound for reflexive polytopes) and giving sufficient conditions for sequences of h*-distributions to be asymptotically normal, with applications to zonotopes and related families.","feed_headline":"Every polytope's dilates converge to the Eulerian distribution","feed_subtitle":"Treating Ehrhart coefficients as probabilities yields universal limits and new inequalities for real-rooted polytopes.","key_machinery":"The central object is hdis(P), the normalized coefficient vector of the h*-polynomial. Its distributional moments are connected to Ehrhart polynomial coefficients through the binomial-basis identity L_P(t) = Σ_j h_j^* binom(t+d-j,d), which makes the moment/coefficient correspondence triangular and invertible. The limit theorem rests on a root-continuity theorem for h*(tP;z) as t→∞, transferring roots to those of the Eulerian polynomial. The real-rooted results use the equivalence between real-rooted distributions and sums of independent Bernoulli trials, which supplies tail inequalities and a variance-divergence criterion for asymptotic normality.","core_discovery":"Every d-dimensional lattice polytope P yields a finite distribution hdis(P) = (h_0^*/Vol(P), …, h_d^*/Vol(P)). Its mean is (d+1)/2 − B(P)/(2Vol(P)) and its variance is an explicit rational expression in the Ehrhart coefficient c_{d-2}, the mean, and d; conversely, all Ehrhart coefficients are recoverable from the first moments. The paper's main limit theorem states that hdis(tP) tends to the Eulerian distribution Eul_d as t→∞, meaning the scaled coefficient vector of h*(tP;z) approaches the polynomial that counts descents in permutations. This is proved via a root-continuity result: the roots of h*(tP;z) approach the roots of the Eulerian polynomial. For real-rooted h*-polynomials, the paper","pith_inferences":["The limit theorem suggests a quantitative refinement: the rate at which hdis(tP) approaches Eul_d likely depends on the distance between the roots of h*(tP;·) and the Eulerian roots; the paper does not address rates, and a bound here could yield finite-dilation error estimates.","The cluster-point analysis only shows all limits with nonzero first coordinate must have that coordinate zero; a natural research program would be to characterize the subset of that facet attainable as limits, using the one-row Hermite normal form families as building blocks.","The paper's sufficient condition for asymptotic normality is stated with a variance lower bound that, as written, is not valid at large dimension because E[X]^2 is Θ(d^2) rather than ≤ 1/4; the corrected bound Var ≥ 2c_{d-2}/(d(d-1)c_d) + (d-2)/12 would still support the corollary for the cited families.","Treating h*-vectors as distributions gives a probabilistic language for Ehrhart positivity: asking when c_{d-2} ≥ 0 becomes asking when the distribution has enough spread, and higher cumulants could be mapped onto higher Ehrhart coefficients."],"forward_implications":["The mean and variance formulas convert Ehrhart-theoretic quantities (boundary volume, c_{d-2}) into probabilistic facts about a lattice point sampled with probability equal to its height in the fundamental parallelepiped.","The convergence hdis(tP) → Eul_d gives a universal asymptotic shape for all sufficiently large dilates of any polytope, independent of the original polytope.","For real-rooted reflexive polytopes, normalized volume must be at least 2^d, with the crosspolytope being the extremal case; this is a new lower bound from probabilistic tail bounds.","Sequences of real-rooted h*-polynomials whose variance grows without bound have asymptotically normal h*-distributions, allowing normal approximations to Ehrhart coefficient sums.","The formulas specialize to exact statistics for lattice zonotopes in terms of gcds of minors, so the h*-distribution can be computed from the generators alone."],"fun_headline_variants":["Ehrhart h*-distributions: mean, variance, Eulerian limit","Lattice polytope dilates approach Eulerian distribution","h*-coefficients as probabilities: limits and inequalities","From polytopes to Eulerian: Ehrhart h*-distributions","Ehrhart polynomials as probability distributions"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The asymptotic-normality results assume real-rooted h*-polynomials, B(P) ≤ Vol(P), and c_{d-2} ≥ 0; moreover, the printed variance lower bound in the proof is invalid because E[X_P]^2 grows like d^2, so the stated sufficient condition (7) does not actually follow from the assumptions.","fun_headline_variants_meta":{"raw":{"variants":["Ehrhart h*-distributions: mean, variance, Eulerian limit","Lattice polytope dilates approach Eulerian distribution","h*-coefficients as probabilities: limits and inequalities","From polytopes to Eulerian: Ehrhart h*-distributions","Ehrhart polynomials as probability distributions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00027,"raw_usage":{"total_tokens":1460,"prompt_tokens":744,"completion_tokens":716,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":488,"completion_tokens_details":{"reasoning_tokens":631}},"tokens_in":488,"tokens_out":716,"duration_ms":6784,"temperature":1.0,"reasoning_tokens":631,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T22:05:08.350965+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For P=[0,1]^d, the h*-distributions of tP are known explicitly as wreath-product Eulerian distributions; checking whether these normalized coefficients converge to the descent distribution as t→∞ at fixed d would directly test Theorem 3.21. For the asymptotic-normality criterion, evaluating the variance formula from Theorem 3.5 on a sequence of reflexive zonotopes with interior lattice points and checking whether the variance diverges would test Corollary 4.5.","supporting_citations":[],"review_version":1}