{"id":"2f9bdea0-f674-404c-ba17-88b7df19a67e","arxiv_id":"2411.11208","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Generic pipe dreams encode the equivariant cohomology classes of lower-upper varieties, unifying classic and bumpless pipe dream formulas and producing new Chern-Schwartz-MacPherson class formulas.","lead":"This math paper introduces a new combinatorial object, generic pipe dreams, whose weighted sums give formulas for the cohomology classes of lower-upper varieties. The same objects unify two classical formulas for Schubert polynomials and yield new Chern class formulas for matrix orbits and double Bruhat cells.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 4 rests on an unproved divided-difference identity for G_w; a computational check or an explicit specialization of [GZJ22, prop. 6] is needed before the central identification is secure.","rationale":"The reader's weakest assumption identifies the unproved divided-difference recurrence for G_w in Theorem 4 as the key gap, and I agree. That assumption is genuinely load-bearing: without it, there is no connection between the combinatorial object G_w and the geometric class of E_w. The reader also flags the 'with a bit more work' step in Theorem 6; I regard that as secondary because Theorem 6 explicitly depends on Theorem 4 and cannot serve as an independent proof. The recovery of the classic and bumpless pipe dream formulae in the leading terms and the independent degree formula from [GZJ22] provide real evidence of plausibility, but they do not prove the full identity. The proposed test is inexpensive and decisive: it checks the divided-difference identity for all permutations up to n = 4, which is feasible with explicit tiles and polynomial arithmetic. The reason for keeping the verdict unchanged is that the concern does not make the claim seem false; it only confirms that the proof as written is incomplete. The right outcome is still CONDITIONAL, conditional on filling the identified gap.","tokens_in":1161,"tokens_out":1104,"duration_ms":126220,"concrete_test":"Independently verify the divided-difference assertion in Theorem 4 for n = 3 and n = 4: implement the recursions from [KZJ14, §4] and the GPD enumeration with the stated tile weights, then check for every w in S_n that (A+B)^{-n} G_w satisfies the same recurrence as the class of E_w, as a polynomial identity in Z[x,y,A,B]. Separately, read [GZJ22, prop. 6] and verify that the K-theoretic divided-difference operators reduce to the cohomological ones when the K-theoretic variable is set to 1. A single counterexample would refute the proof of Theorem 4; full agreement would fill the missing lemma and support the existing CONDITIONAL verdict.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is Theorem 4: the equivariant class of E_w equals (A+B)^{-n} times the generic pipe dream polynomial G_w. The proof sketch has two premises: (1) the equivariant classes of E_w are uniquely determined by divided-difference recursions from [KZJ14, §4]; and (2) the polynomials G_w satisfy the same recursions, asserted as 'not hard to show' with a pointer to [GZJ22, prop. 6]. Premise (2) is not proved in this paper, and it is load-bearing. Theorem 6 cannot fill the gap because its conclusion explicitly invokes Theorem 4 ('Because we already know equality of cohomology classes according to theorem 4'), and its own 'with a bit more work' flux-propagation argument is only sketched. The potential failure modes are concrete: the K-theoretic statement in [GZJ22, prop. 6] might involve the K-theoretic divided-difference operators, which need not specialize to the exact cohomological operators used in [KZJ14, §4]; or the normalization, base case, or action of A and B might differ by a shift or a power of (A+B). If the recursions do not hold for G_w, then the identification of the class of E_w with a weighted enumeration of generic pipe dreams fails, even though the leading-term limits of Theorem 1 and the degree formula from [GZJ22] provide strong independent evidence. This is not a disagreement with consensus but an unverified lemma in the proof of the paper's main theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a new family of combinatorial objects, 'generic pipe dreams' (GPDs), assembled from four tile types on an n x n grid, and defines polynomials G_w by assigning to each tile a weight A+x_i-y_j, B-x_i+y_j, or A+B. The main geometric claim is Theorem 4: the (B_- x B_+ x (C^×)^2)-equivariant cohomology class of the lower-upper variety E_w is (A+B)^{-n}G_w. The authors also prove leading-term statements recovering classic and bumpless pipe dream formulas for double Schubert polynomials (Theorem 1), derive a formula for the degree of the commuting variety as a sum of powers of 2, and apply GPDs to compute Segre-Schwartz-MacPherson classes of Kazhdan-Lusztig varieties, double Bruhat cells, and matrix Schubert varieties (Theorems 10, 13). A conjectural K-theoretic extension is discussed in Section 4.","tokens_in":10031,"tokens_out":9676,"duration_ms":95636,"significance":"If established, Theorem 4 provides a uniform and elegant weighted-enumeration formula for the equivariant classes of all lower-upper varieties, recovering two previously independent pipe dream models as leading-term specializations. The degree formula for the commuting variety is striking and algorithmically effective. The applications to SSM classes are potentially valuable and connect GPDs to Schubert calculus, cluster algebras, and characteristic class theory. The manuscript is honest about what is proved and what is conjectural, and it explicitly labels the two main proof sections as sketches. However, the paper's central result currently rests on an unverified assertion about divided-difference recursions, and the degeneration argument in Theorem 6 has significant gaps. The overall picture is plausible, but the manuscript is not yet at the level of completeness expected for a journal publication.","major_comments":[{"comment":"The proof of Theorem 4 consists of two premises: the equivariant classes of E_w are uniquely determined by divided-difference recursions from [KZJ14, §4], and the GPD polynomials G_w satisfy the same recursions. The second premise is asserted as 'not hard to show' with a pointer to [GZJ22, prop. 6], but no proof is given here. This is load-bearing: if G_w does not satisfy exactly the same recursions, with the same normalization and the same action of A and B, the identification of the class of E_w with (A+B)^{-n}G_w does not follow. The cited proposition is K-theoretic, and the specialization of its divided-difference operators and base cases to the cohomological setting of [KZJ14, §4] is not automatic. I am not claiming the assertion is false, only that it is unproved in this manuscript. The authors should supply a complete proof or a precise proposition-by-proposition specialization of [GZJ22, prop. 6].","section":"§2, Theorem 4"},{"comment":"The degeneration argument is incomplete in exactly the places that matter. The step 'An easy calculation of its initial term leads to the equation X_ij(Φ_S - Φ_E) = 0' is not shown, and the subsequent flux-propagation step ('starting from the bottom/left and adding one square at a time, we conclude that fluxes Φ_e can only take the values 0, t_1, ..., t_n') is only asserted. The conclusion that each component of the degeneration of E_w lies in a single complete intersection F_δ is the crux of the geometric interpretation, but the proof does not justify why no other flux configurations survive. Moreover, the final comparison explicitly invokes Theorem 4, so Theorem 6 cannot serve as an independent verification of the class formula. Please provide the missing initial-term computation and a rigorous induction for the flux propagation, including a discussion of embedded components.","section":"§2, Theorem 6"},{"comment":"In the proof of Theorem 13, after erasing the pipes coming from the South, the text states that 'the weight of a blank is the sum of the weight of a bump and of that of a cross,' citing Eq. (4). With the weights displayed in Eq. (4), this identity would read 1 = (x-y) + (x-y+1), which is not an identity. If the intended modified weights carry signs, as hinted by the preceding sentence ('It differs from the original weight (1) by signs'), that sign convention must be written explicitly and the identity verified. As written, the CSM formula for B_- w B_+ is not justified, and this result is one of the applications announced in the abstract.","section":"§3.2, Theorem 13 proof"}],"minor_comments":[{"comment":"The manuscript relies heavily on custom tile diagrams that are not legible in the version I received; since the definition of GPD and the weights in Eqs. (1), (4), and the definition of F_δ in Theorem 6 all depend on these diagrams, please ensure they are typeset clearly in the final version.","section":"§1, Eq. (1) and throughout"},{"comment":"The GPDs for w = 2431 are displayed but not labelled; it would help the reader if the 'first two' and 'last' GPDs were identified explicitly, for instance by numbering the pictures.","section":"§1, Example 2"},{"comment":"There is a typo: 'a set of GDPs' should read 'a set of GPDs'.","section":"§3.2"},{"comment":"In formula (3), the index set I is used without an explicit definition; please state that I ⊆ {1, ..., |Q|} is the set of positions at which the subword R has letters, and clarify the notation ∏_{i∈I} Q_i as a product in the Weyl group.","section":"§3.1, Theorem 10"},{"comment":"The phrase 'these GPDs have no blanks' is slightly confusing in light of Eq. (4), which assigns weight 1 to the 'otherwise' case; please make explicit whether blanks occur in this setting and, if not, why the 'otherwise' weight is needed.","section":"§3.2, Corollary 11"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know about this one. It gives a single weighted sum over generic pipe dreams (GPDs) that recovers the classic pipe dream formula at the highest B-power and the bumpless pipe dream formula at the highest A-power. That is a real unification, and the paper is upfront about which terms survive the two limits. The reason it matters is Theorem 4: the (B_- × B_+ × (C^×)^2)-equivariant class of the lower-upper variety E_w is (A+B)^{-n} times the GPD polynomial. That theorem is only sketched, and the sketch has a load-bearing gap.\n\nWhat is new and good: the GPD model itself; Theorem 1's exchange argument is clear and self-contained; the commuting-variety degree formula as a sum of powers of 2 is a nice byproduct, and they compute it up to n=16 and cite an independent proof from [GZJ22]; the CSM section produces closed formulas for B_-wB_+ and double Bruhat cells, with an explicit small example. The heavy self-citation to [KZJ14] and [GZJ22] is legitimate—those are the sources of the uniqueness theorem and the K-theoretic recursion, not padding.\n\nThe soft spot is Theorem 4's proof. The authors say the classes are uniquely determined by divided-difference recursions from their 2014 paper and that 'it is not hard to show' the GPD polynomials satisfy the same recursions, pointing to [GZJ22, prop. 6], a K-theoretic statement. The stress-test note is right: a K-theoretic recursion does not automatically specialize to the exact cohomological recursion with the right normalization, base case, and action of A and B. The paper gives none of those details. And Theorem 6 cannot backstop Theorem 4, since its conclusion explicitly assumes Theorem 4. So the central identification is likely true—the two leading-term limits and the independent degree formula give real support—but as written it is a conjecture-with-evidence, not a proof. That is a serious gap for a paper whose main theorem is Theorem 4. A referee should demand a complete proof of the recursion check or a precise pointer to where it appears.\n\nThis is for Schubert calculus people and anyone working on CSM classes of orbits; the reading group would enjoy it. It deserves peer review, but with the expectation of a revision that fills or precisely cites the missing computation. I would cite the GPD construction and the CSM formulas, with a caveat on Theorem 4.","headline":"A genuinely useful unification of pipe dream formulas, likely correct, but the main equivariant-class theorem is presented as a sketch with a load-bearing proof gap.","tokens_in":10538,"tokens_out":3476,"would_cite":true,"duration_ms":33766,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14M15","14N15","14C17"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the equivariant cohomology class of every lower-upper variety $E_w$ is, up to a power of $A+B$, the generic pipe dream polynomial $G_w$, a finite weighted sum over three-tile pipe dream grids.","keywords":["generic pipe dreams","lower-upper varieties","equivariant cohomology","double Schubert polynomials","commuting variety","Schwartz-MacPherson classes","Kazhdan-Lusztig varieties"],"falsifier":"Take $n=2$, $w$ the transposition, expand $(A+B)^{-2}G_w$ as a polynomial in $A,B,x_1,x_2,y_1,y_2$, and compare one coefficient, say that of $x_1^2$, with the same coefficient obtained by applying the divided-difference recursion the paper cites; if they differ, Theorem 4 is false. A geometric check would compute the initial ideal of the defining equations of $E_w$ for a small $n$ and look for any component not among the $F_\\delta$.","tokens_in":9439,"feed_emoji":"🧩","tokens_out":15529,"duration_ms":123643,"temperature":0.7,"pith_summary":"This paper aims to show that one combinatorial gadget, the generic pipe dream (GPD), controls the topology of the whole family of lower-upper varieties $E_w$. The central claim is Theorem 4: under the $(B_- \\times B_+ \\times (\\mathbb{C}^\\times)^2)$-action, the equivariant cohomology class of $E_w$ equals $(A+B)^{-n}G_w$, where $G_w$ is the sum, over all GPDs for $w$, of products of tile weights $A+x_i-y_j$, $B-x_i+y_j$, and $A+B$. If this is right, the same enumeration contains the classic and bumpless pipe dream formulas for double Schubert polynomials, the degree of the $n$th commuting variety as a sum of powers of 2, and Chern-Schwartz-MacPherson classes of Kazhdan-Lusztig varieties, matrix Schubert varieties, and double Bruhat cells. The reason to care is that an infinite geometric invariant becomes a finite, explicit combinatorial count.","feed_headline":"Pipe-dream sum equals class of every lower-upper variety","feed_subtitle":"The same weighted tile count recovers double Schubert polynomials and commuting-variety degrees.","key_machinery":"The central object is the generic pipe dream (GPD): an $n\\times n$ square tiled by three tiles — a crossing, a bump, and a blank — with labeled pipes entering from the West and North according to the permutation $w$. Each tile contributes a linear equivariant weight, $A+x_i-y_j$ for a crossing, $B-x_i+y_j$ for a bump, and $A+B$ for a blank, and $G_w$ is the sum of products of these weights over all GPDs. The argument is carried by two mechanisms: a combinatorial one, in which $G_w$ is shown to obey the same divided-difference recursions that uniquely determine the equivariant class in the cited Brauer loop scheme work; and a geometric one, in which a degeneration of $E_w$ with flux variables on the grid forces every limiting component to lie in one of the complete intersections $F_\\delta$, whose classes are the individual GPD terms.","core_discovery":"The paper's discovery is a closed formula for the equivariant cohomology class of each lower-upper variety $E_w = \\{(X,Y) : XY \\text{ lower triangular},\\ YX \\text{ upper triangular},\\ \\operatorname{diag}(XY) = w \\cdot \\operatorname{diag}(YX) \\text{ nonrepeating}\\}$ inside $(\\mathrm{Mat}_{n\\times n})^2$. With the scaling action of $(\\mathbb{C}^\\times)^2$ added to the $B_- \\times B_+$ action, the class is $(A+B)^{-n}$ times the generic pipe dream polynomial $G_w$, which is the sum over $n\\times n$ grids of crossing, bump, and blank tiles (with boundary labels determined by $w$) of the product of the three corresponding weights. The proof has two halves: one shows combinatorially that $G_w$ satisfies the same divided-difference recursions that uniquely determine the class; the other exhibits an equivariant degeneration of $E_w$ into complete intersections $F_\\delta$, one per GPD, whose classes are exactly the individual terms of the formula. As the paper stresses, the leading terms of $G_w$ recover the classical and bumpless pipe dream formulas for double Schubert polynomials, and the same GPDs, with modified weights, compute CSM classes.","pith_inferences":["Inference: If Theorem 4 is correct, the lower-upper variety is a geometric realization of pipe-dream connectivity: its equivariant class depends only on the weighted set of pipe configurations, so any deformation of the variety that preserves that set has the same class.","Inference: The flux-degeneration argument that produces the complete intersections $F_\\delta$ could plausibly be applied to other schemes defined by matrix product conditions, yielding pipe-dream-type enumerations for their classes; the paper does not state this.","Inference: Because the pipe-dream weights are linear in the equivariant parameters, the theorem suggests that all the Schubert and CSM formulas discussed here are specializations of one master polynomial; checking the conjectural K-theoretic pipe-dream formula for small $n$ would test that uniformity."],"forward_implications":["For every permutation $w$, the equivariant class of the lower-upper variety becomes an explicit finite sum over generic pipe dreams, so it can be computed by enumeration rather than by solving geometric recursions.","The leading terms of the pipe-dream sum reproduce the classical double Schubert polynomial formula and the bumpless pipe dream formula, making the generic pipe dream the common parent of both calculi.","The degree of the $n$th commuting variety equals a sum of powers of 2 indexed by generic pipe dreams for the identity permutation, giving an effective way to compute these degrees for large $n$.","The same modified pipe-dream weights compute the Chern-Schwartz-MacPherson classes of open Kazhdan-Lusztig varieties, matrix Schubert varieties, and double Bruhat cells, so all of these classes are governed by one combinatorial model.","The K-theoretic analogue replaces tile weights by K-theoretic ones and conjecturally provides K-classes of lower-upper varieties, including the commuting variety."],"supporting_citations":[{"why":"Defines the lower-upper scheme and its components E_w, the geometric objects whose classes are computed.","marker":"[Knu05]"},{"why":"Supplies the divided-difference recursions that uniquely determine the equivariant classes, which the GPD polynomials must match.","marker":"[KZJ14]"},{"why":"Establishes that matrix Schubert variety classes are double Schubert polynomials, the classical formula recovered as a limit.","marker":"[KM05]"},{"why":"Introduces bumpless pipe dreams, whose formula is recovered as another limit.","marker":"[LLS21]"},{"why":"Gives the K-theoretic analogue and an independent proof of the commuting-variety degree formula.","marker":"[GZJ22]"},{"why":"Supplies the transversality product rule for Segre-Schwartz-MacPherson classes used in Section 3.","marker":"[Sch17]"},{"why":"Provides the toric degeneration of Bott-Samelson varieties used to compute SSM class restrictions.","marker":"[PP16]"},{"why":"Gives the restriction formula for stable bases that the paper re-derives and generalizes.","marker":"[Su15]"}],"fun_headline_variants":["All lower-upper classes from a single pipe-dream sum","Pipe-dream formula recovers Schubert and commuting degrees","One pipe-dream sum determines every lower-upper class","Generic pipe dreams yield all lower-upper classes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the paper's assertion, given as a sketch rather than a proof, that the generic pipe dream polynomials satisfy exactly the same divided-difference recursions that uniquely determine the equivariant classes, together with the unproved geometric step that the degeneration of $E_w$ has no components outside the complete intersections $F_\\delta$.","fun_headline_variants_meta":{"raw":{"variants":["All lower-upper classes from a single pipe-dream sum","Pipe-dream formula recovers Schubert and commuting degrees","One pipe-dream sum determines every lower-upper class","Generic pipe dreams yield all lower-upper classes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000826,"raw_usage":{"total_tokens":3616,"prompt_tokens":954,"completion_tokens":2662,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":570,"completion_tokens_details":{"reasoning_tokens":2596}},"tokens_in":570,"tokens_out":2662,"duration_ms":30579,"temperature":1.0,"reasoning_tokens":2596,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T18:47:37.535220+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $n=2$, $w$ the transposition, expand $(A+B)^{-2}G_w$ as a polynomial in $A,B,x_1,x_2,y_1,y_2$, and compare one coefficient, say that of $x_1^2$, with the same coefficient obtained by applying the divided-difference recursion the paper cites; if they differ, Theorem 4 is false. A geometric check would compute the initial ideal of the defining equations of $E_w$ for a small $n$ and look for any component not among the $F_\\delta$.","supporting_citations":[],"review_version":1}