{"id":"cf4089d6-c91d-4bff-83a1-662f4b816e43","arxiv_id":"2411.10276","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For minuscule homogeneous spaces, Chevalley polytopes coincide with order polytopes of minuscule posets, producing Newton-Okounkov bodies and Khovanskii bases; the general case is conjectural.","lead":"The authors construct Chevalley polytopes from homogeneous spaces and prove that for minuscule spaces these polytopes are order polytopes of minuscule posets, yielding Newton-Okounkov bodies and Khovanskii bases. They also conjecture a Minkowski-sum decomposition that would give a polytopal shadow of the Littlewood-Richardson rule.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 4.1's degrevlex order is incompatible with the stated poset order: for Gr(2,4) it can select a non-identity embedding, so the proof that the Chevalley polytope equals the order polytope does not go through.","rationale":"The reader's verdict identifies the correct region of the proof, and my check confirms the proof of Proposition 4.1 is not sound. I differ slightly on the mechanism: for the diamond example with F = {top}, the identity embedding is in fact the unique minimum under the stated partial order; what fails is the claimed compatibility of degrevlex with that order, because the larger heap element has the larger coordinate, reversing the order. This is not merely cosmetic: it changes the valuations of Plücker coordinates, so the vertex set of the Chevalley polytope differs from the filter indicators. Interestingly, for Gr(2,4) the resulting polytope may still have normalized volume 2, so the central theorem could survive, but the written chain Proposition 4.1 → Corollary 4.3 → Theorem 4.4 → Theorem 4.6 is broken. The appendix examples do not cover this case and no code is supplied. Conditional acceptance with a request to repair the term-order/embedding argument, or to prove the theorem for a specified ordering, is the appropriate outcome.","tokens_in":24975,"tokens_out":32638,"duration_ms":309237,"concrete_test":"Compute, by hand or with a small script, the Chevalley polytope for Gr(2,4) with reduced expression s2s1s3s2 under Definition 3.2: write each Plücker coordinate as a polynomial in a_top, a_left, a_right, a_bottom, record degrevlex-minimal exponents with a_top > a_left > a_right > a_bottom, and compare the resulting convex hull to the order polytope O_{wP}. If ν(p_13) = e_bottom rather than e_top, or if the hull is not O_{wP}, Proposition 4.1 fails as stated; then test whether reversing the degrevlex order (a_bottom > ... > a_top) restores equality with O_{wP} and preserves the volume-equals-degree argument. Also recompute the semigroup generation in Theorem 4.6 under the original order to see whether the Plücker valuations still generate the value semigroup.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In §4, Proposition 4.1 rests on the claim that the degrevlex term order with coordinates a_{bℓ} > ... > a_{b1} extends the partial order τ1 ≤ τ2 iff τ1(b) ≥ τ2(b) on labeled embeddings of a filter F, so that the valuation of p_F is the identity-embedding monomial. This compatibility is false even in the simplest repeated-label case. For X = Gr(2,4) with reduced expression s2s1s3s2, the heap is the diamond wP = {top, left, right, bottom} with labels 2, 1, 3, 2 and order top > left, right > bottom. For the filter F = {top}, there are two label-preserving embeddings: id(top) = top and τ(top) = bottom. Under the paper's partial order, id ≤ τ because top ≥ bottom, so a valid extension would require a_id ≤ a_τ, i.e. a_top ≤ a_bottom. But degrevlex with a_top > a_bottom gives the opposite. Consequently ν(p_F) is the monomial a_bottom, not a_top: concretely p_13 restricts to a_top + a_bottom and its degrevlex-minimal term is a_bottom, the indicator of {bottom}. Since {bottom} is not a filter of wP, the vertices of P_{X,ω_k} are not simply indicator vectors of filters, and the identification P_{X,ω_k} = O_{wP} is not established. Corollary 4.3 (volume = degree) and the integer-decomposition step in Theorem 4.6 both rely on this identification, so the gap is load-bearing. The claim may be repairable by reversing the coordinate order or by a more careful choice of term order, but as written the proof has a false premise.","agreement_with_reader":"partial"},"referee_report":null,"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First the punchline: the Chevalley polytope construction is a nice idea, and the minuscule story would be quite appealing if the proof held up. It doesn't, as written—there is a concrete counterexample to the key claim in Proposition 4.1. A good referee would send this back for repair, but the paper deserves that referee.\n\nThe construction: take a homogeneous space X=G/P, a projective embedding from a highest weight representation, and a reduced expression for w^P. Use the associated torus coordinates to define a degrevlex valuation, then take the convex hull of valuations of the degree-1 coordinate functions. This is natural and new. The paper's Conjecture 5.4—that the Chevalley polytope decomposes as a Minkowski sum of fundamental-weight Chevalley polytopes—is an attractive shadow of the Littlewood-Richardson rule, and the examples support it.\n\nThe problem: Section 4 proves that for minuscule X, P is the order polytope of the minuscule poset by claiming the valuation of p_F is the indicator of F itself. The proof rests on the assertion that the degrevlex order extends the partial order on labeled embeddings given by τ1 ≤ τ2 iff τ1(b) ≥ τ2(b) pointwise. This is false when labels repeat. The stress-test example is Gr(2,4) with s=s2s1s3s2: the heap is a diamond with two incomparable elements both labeled 2. For the filter {top}, p_13 restricts to a_top+a_bottom, and degrevlex with a_top > a_bottom selects a_bottom. That monomial is not the identity embedding and its support is not a filter, so the order-polytope identification fails at that step. This is not a minor typo: Corollary 4.3, Theorem 4.4, and the IDP argument in Theorem 4.6 all depend on that identification. The theorem may still be true—reversing the coordinate order or choosing a more careful term order might fix it—but the proof as written doesn't go through.\n\nWhat the paper does well besides that: the comparison with string polytopes is informative, with concrete examples where Chevalley polytopes are integral, full-dimensional, and satisfy IDP while string polytopes don't. The writing is honest about what is conjectural, and the appendix examples are useful evidence, though they lack the code that would make them reproducible.\n\nMy take: send it to peer review. The construction and conjectures deserve attention, and the minuscule theorem is likely salvageable. I wouldn't cite the theorem in its current form, but I'd keep the paper on the short list to revisit once the embedding-minimality argument is patched.","headline":"A genuinely new polytope construction with a repairable but load-bearing gap in the minuscule theorem's proof.","tokens_in":25907,"tokens_out":3218,"would_cite":false,"duration_ms":30337,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14M15","14M25","52B20","05E10"],"pacs":[],"model":"deepseek-v4-flash","headline":"For minuscule homogeneous spaces, the Chevalley polytope is a Newton-Okounkov body whose Plücker coordinates form a Khovanskii basis.","keywords":["Chevalley polytopes","Newton-Okounkov bodies","Khovanskii bases","minuscule homogeneous spaces","order polytopes","Plücker coordinates","toric degenerations","string polytopes"],"falsifier":"For the Grassmannian $\\mathrm{Gr}(2,4)$ with reduced expression $s_2s_1s_3s_2$, compute the restriction of the Plücker coordinate $p_{13}$ to the torus: if it equals $a_1+a_4$ and the degrevlex-minimal term is $a_1$, then the valuation of $p_{13}$ is not the identity-embedding monomial, directly contradicting the stated premise of Proposition 4.1; the theorem could then hold only after a coordinate permutation.","tokens_in":24732,"feed_emoji":"📐","tokens_out":10885,"duration_ms":89449,"temperature":0.7,"pith_summary":"The paper introduces Chevalley polytopes, built from a valuation on the coordinate ring of a homogeneous space $X=G/P$ with a chosen projective embedding and reduced expression. The valuation is defined by restricting functions to a torus and taking the exponent vector of the degrevlex-minimal monomial, and the Chevalley polytope is the convex hull of valuations of degree-one coordinate functions. The main theorem states that when $X$ is minuscule in its minimal embedding, this polytope is the order polytope of the minuscule poset $w^P$, hence a Newton-Okounkov body, and that the Plücker coordinates form a Khovanskii basis, yielding a toric degeneration to the associated projectively normal toric variety. This matters because Newton-Okounkov bodies and Khovanskii bases are usually hard to construct, while here they come directly from poset combinatorics.","feed_headline":"Chevalley polytopes are Newton-Okounkov bodies for minuscule spaces","feed_subtitle":"They are explicit order polytopes; Plücker coordinates give a Khovanskii basis and toric degeneration.","key_machinery":"The central object is the valuation $\\nu_{X,\\varpi,s}$, which sends a function on $X$ to the exponent vector of the degrevlex-minimal monomial in its restriction to the torus $X^\\circ = U_-^\\circ/P$ determined by a reduced expression $s$ for $w^P$; the Chevalley polytope is the convex hull of these valuations over the homogeneous degree-one part of $\\mathbb{C}[X]$. In the minuscule case, full commutativity makes the heap of any reduced expression equal to the minuscule poset $w^P$, and the valuation of each Plücker coordinate is the indicator vector of a filter. Proposition 4.1 therefore identifies the Chevalley polytope with Stanley's order polytope $O_{w^P}$. The proof then combines three facts: the normalized volume of $O_{w^P}$ counts linear extensions of $w^P$, which equal the degree of $X$; order polytopes satisfy the integer decomposition property; and the general valuation machinery turns one-dimensional leaves plus a Khovanskii basis into a Newton-Okounkov body and a toric degeneration.","core_discovery":"The paper's central claim is Theorem 1.1: for a minuscule homogeneous space $X=G/P_k$ in its minimal embedding, the Chevalley polytope $P_{X,\\varpi_k}$ is a Newton-Okounkov body for $X$ with respect to the valuation $\\nu_{X,\\varpi_k}$; the Plücker coordinates on $X$ form a Khovanskii basis for $\\mathbb{C}[X]$; and $X$ admits a toric degeneration to the projectively normal toric variety associated to $P_{X,\\varpi_k}$. The key identification is $P_{X,\\varpi_k}=O_{w^P}$, the order polytope of the minuscule poset, whose vertices are precisely the indicator vectors of filters of $w^P$. Because order polytopes have no interior lattice points and satisfy the integer decomposition property, the valuations of Plücker coordinates generate the valuation semigroup, which is exactly the condition that they form a Khovanskii basis.","pith_inferences":["If the proof's labeling premise is repaired, the most likely salvage is that every minuscule Chevalley polytope coincides with an order polytope up to a permutation of coordinates depending on the reduced expression; the theorem would survive in that weaker form.","The same valuation scheme suggests a practical recipe for non-minuscule spaces: look for reduced expressions whose heaps make the weighted embeddings of Chevalley-basis elements behave like filters; the appendix examples show some expressions work and others do not.","A computational test of the Minkowski decomposition conjecture in small rank would strengthen the analogy with the Littlewood-Richardson rule: for each representation appearing in the tensor product, there should be a subpolytope of the Chevalley polytope whose lattice points count the multiplicity.","The construction may generalize beyond $G/P$ to any projective variety carrying a torus chart and a distinguished set of algebra generators whose valuations can be described combinatorially."],"forward_implications":["Every minuscule homogeneous space in its minimal embedding acquires an explicit 0/1-polytope Newton-Okounkov body, the order polytope of its minuscule poset, so the body can be written down from the poset alone.","The Plücker coordinates generate the valuation semigroup, so they form a Khovanskii basis; in particular, the homogeneous coordinate ring of $X$ degenerates to the toric ring of the Chevalley polytope, and the degeneration is projectively normal.","The Chevalley polytope has no lattice points except its vertices, and its dilations decompose into sums of vertices, so the associated toric variety is projectively normal.","For the same spaces, string polytopes can fail to be integral or to satisfy the integer decomposition property, while minuscule Chevalley polytopes always have these properties.","The paper conjectures that the same construction works for arbitrary homogeneous spaces and embeddings, and that Chevalley polytopes decompose under Minkowski sums according to the expansion of the weight into fundamental weights, a polytopal analogue of the Littlewood-Richardson rule."],"supporting_citations":[{"why":"Supplies the full-commutativity theorem that makes the heap of any reduced expression equal to the minuscule poset.","marker":"[Ste96]"},{"why":"Establishes that the weight poset of a minuscule representation is a distributive lattice, identifying the minuscule poset whose filters label Plücker coordinates.","marker":"[Pro84]"},{"why":"Provides the order polytope facts used to identify the Chevalley polytope with $O_{w^P}$: vertices are filters, normalized volume counts linear extensions, and dilations have the integer decomposition property.","marker":"[Sta86]"},{"why":"Provides the definitions and general theorems connecting valuations with one-dimensional leaves and finite Khovanskii bases to Newton-Okounkov bodies and toric degenerations.","marker":"[KM19]"},{"why":"Gives the refined condition under which the toric degeneration is to the toric variety of the Newton-Okounkov body itself rather than its normalization.","marker":"[RW19]"},{"why":"Used in Lemma 4.2 to compute the degree of a minuscule $X$ as the number of maximal chains via the Chevalley multiplication formula.","marker":"[CMP08]"},{"why":"Supplies the lattice-theoretic fact that maximal chains of a distributive lattice are counted by linear extensions of its poset of join-irreducibles, used in the degree computation.","marker":"[Sta72]"},{"why":"Identifies the Schubert variety poset of $G/P$ with minimal coset representatives under Bruhat order, used in the degree computation.","marker":"[Hil82]"}],"fun_headline_variants":["Minuscule Chevalley polytopes are Newton-Okounkov bodies","Chevalley polytopes = Newton-Okounkov bodies for minuscule spaces","Minuscule Chevalley polytopes yield Newton-Okounkov bodies","Chevalley polytopes: Newton-Okounkov bodies for minuscule spaces","Minuscule spaces get toric degenerations from Chevalley polytopes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that the Chevalley polytope equals the order polytope relies on the identity embedding of a filter into the minuscule poset being the unique minimal embedding under the chosen term order; this can fail when two elements of the poset carry the same label, because the degrevlex-minimal term may then come from a non-identity embedding.","fun_headline_variants_meta":{"raw":{"variants":["Minuscule Chevalley polytopes are Newton-Okounkov bodies","Chevalley polytopes = Newton-Okounkov bodies for minuscule spaces","Minuscule Chevalley polytopes yield Newton-Okounkov bodies","Chevalley polytopes: Newton-Okounkov bodies for minuscule spaces","Minuscule spaces get toric degenerations from Chevalley polytopes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.002386,"raw_usage":{"total_tokens":9207,"prompt_tokens":992,"completion_tokens":8215,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":8107}},"tokens_in":608,"tokens_out":8215,"duration_ms":55888,"temperature":1.0,"reasoning_tokens":8107,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T19:51:38.255235+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For the Grassmannian $\\mathrm{Gr}(2,4)$ with reduced expression $s_2s_1s_3s_2$, compute the restriction of the Plücker coordinate $p_{13}$ to the torus: if it equals $a_1+a_4$ and the degrevlex-minimal term is $a_1$, then the valuation of $p_{13}$ is not the identity-embedding monomial, directly contradicting the stated premise of Proposition 4.1; the theorem could then hold only after a coordinate permutation.","supporting_citations":[],"review_version":1}