{"id":"21611d2d-a0e7-4011-a508-340e17f24f27","arxiv_id":"2506.11910","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For tamely ramified reductive groups, the moduli stack of Breuil-Kisin modules with descent data and G-structure is shown to be smoothly equivalent to a twisted Schubert variety.","lead":"This paper builds a new geometric space that packages p-adic symmetries attached to reductive groups, and shows it is equivalent to a known geometric building block. The equivalence gives mathematicians a concrete tool to study p-adic Galois representations for arbitrary reductive groups, not just matrix groups.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem A's smooth equivalence is proved only under Assumption ConFib; when ConFib fails, as it does for PGL3 at an alcove barycenter, the identification of G_x with the Pappas–Zhu group scheme is not available, so the advertised unconditional claim is not established.","rationale":"I agree with the reader's identification of ConFib as the weakest load-bearing assumption. The paper is mathematically careful: the main theorems are proved step by step under explicit standing assumptions, and Remark 1.15 acknowledges that both Theorem A and Theorem B depend on G_x having connected fibers. The concern I raise is not that the proof is internally inconsistent, but that the advertised central claim—smooth equivalence with the Pappas–Zhu twisted Schubert variety—is only proven in the connected-fiber case, and Example 3.10 gives concrete tamely ramified examples where that case fails. This is a real scope limitation, not a manufactured objection. It does not, however, invalidate the paper's results as conditional statements. Since the body already flags the assumption and the reader accepted the paper with that caveat, I would keep the verdict unchanged rather than move to rejection or unverified status. If the formal statements or abstract are meant to be unconditional, they should be amended to include Assumption ConFib, but that is a presentation correction, not a mathematical refutation of the stated theorems.","tokens_in":73620,"tokens_out":15486,"duration_ms":341178,"concrete_test":"Work out the PGL3 counterexample of Example 3.10(2): take Ĝ = PGL3 with I acting trivially, x the barycenter of the base alcove, and a small dominant cocharacter μ. Compute the c-twisted φ-conjugation quotient Y^{≤μ} modulo p, or at least the groupoid map β_x: h ↦ h^{-1}Xφ_c(h)X^{-1} used in Lemma 5.11, and compare the resulting stack with the left-translation quotient Gr^{≤μ,∧ϖ} over F. If the special fibers have different π0 or the stabilizer group schemes of corresponding base points differ, Theorem A genuinely requires ConFib; if the two quotient stacks remain equivalent, then ConFib is unnecessary for Theorem A and the statement should be made unconditional.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central equivalence identifies Y^{≤μ} with the p-adic completion of the Pappas–Zhu twisted Schubert variety. This identification passes through Proposition 3.11, which identifies G_x with the Pappas–Zhu group scheme only when G_x has connected fibers. That condition is Assumption ConFib, imposed from Section 3.1.5 onward and explicitly used in both main theorems (Remark 1.15). It is not automatic: Example 3.10(2) gives Ĝ = PGL3, Γ acting trivially, and x the barycenter of an alcove, where π1(PGL3) = Z/3 is torsion and ˘G_x has disconnected fibers. Proposition 3.7 supplies only a sufficient condition (π1(G*)I torsion-free). Thus the central claim as stated in the abstract and in the introductory Theorem A is broader than what the proof establishes: without ConFib, G_x need not be the Pappas–Zhu model, and no argument in Sections 4–6 covers the disconnected-fiber case. The paper itself flags this limitation, so the issue is not a hidden inconsistency; it is that the main theorems are conditional on a nontrivial hypothesis that the advertised statements do not prominently carry.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a moduli stack ℤ of Breuil–Kisin modules with descent data and Ġ-structure, called Breuil–Kisin (Γ,Ġ)-torsors, for a tamely ramified reductive group G over Q_p. For a dominant cocharacter μ it defines the bounded substack ℤ^{≤μ} by a group-theoretic height condition and proves (Theorem A, Theorem 6.4) that ℤ^{≤μ} is a p-adic formal algebraic stack admitting a smooth covering diagram connecting it to the p-adic completion of the Pappas–Zhu twisted Schubert variety Gr^{≤μ}_G. Under unramifiedness, simple-connectivity of the derived group, and genericity hypotheses, Theorem B (Theorem 6.6) gives explicit Zariski charts. The proof uses invariant pushforward to construct Bruhat–Tits group schemes G_x over A^1_O, identifies them with Pappas–Zhu models under the connected-fibers assumption ConFib, and proves contraction/straightening estimates in Section 5. The paper explicitly states in Remark 1.15 and Section 2.7 that the main theorems are proven under Assumption ConFib and, for results using the negative loop group, Assumption Dil.","tokens_in":73956,"tokens_out":10203,"duration_ms":128503,"significance":"If the results are correct, this is a substantial generalization of the local model theorem for moduli of Breuil–Kisin modules from GL_n and GSp_4 to general tamely ramified reductive groups, with a group-theoretic height condition that avoids choosing an embedding into GL_n. The paper provides detailed proofs of the formal algebraic stack property, explicit contraction bounds in Section 5, Zariski chart constructions via negative loop groups, and an extension of Pappas–Zhu group schemes to concave functions (Proposition 3.19). The main theorems are honestly conditional on clearly stated assumptions, and the failure of the connected-fibers condition is explicitly exhibited in Example 3.10. The main concern is that the abstract and introductory theorem statements do not prominently carry this standing assumption, so the advertised scope is broader than the proof establishes.","major_comments":[{"comment":"The abstract and the introductory statement of Theorem A present the smooth equivalence as unconditional, but the proof depends on Assumption ConFib, which is imposed from Section 3.1.5 onward. Proposition 3.11 identifies G_x with the Pappas–Zhu group scheme only when G_x has connected fibers, and Example 3.10(2) shows that ConFib fails for Ġ = PGL_3 at the barycenter of an alcove. No argument in Sections 4–6 covers the disconnected-fiber case. Since the paper itself flags this in Remark 1.15, this is not a hidden inconsistency, but the abstract and the main theorem statements should carry ConFib explicitly; otherwise the claimed scope is broader than what is proved.","section":"§1.5, §2.7, §3.1.4, Example 3.10"},{"comment":"Theorem B as stated omits Assumption Dil and the condition F ⊇ F_q (or E ⊇ L) that Proposition 3.16 requires for G_x to be a dilation, even though Theorem B uses the negative loop group of Section 3.3. Please state these hypotheses in the theorem, or explicitly say that they are part of the standing assumptions in force.","section":"§1.5 Theorem B, §2.7, §3.1.4"},{"comment":"The proof of Proposition 3.19 is carried out in detail only under the assumption that I acts trivially on Ġ; the general case is dispatched by saying that it follows by arguing with ~G_{x,f} and passing to neutral connected components. Since the group schemes G_{x,f} underlie the congruence subgroups used in the straightening argument and in Corollary 4.14, the general case should either be written out or the results that rely on it should be restricted to the unramified case.","section":"Proposition 3.19, §3.1.8"}],"minor_comments":[{"comment":"The displayed characterization of “lowest alcove” contains a stray brace: the condition should read “−1 < ⟨a,β⟩ ≤ 0” with no closing brace after the final 0.","section":"Definition 3.12"},{"comment":"The description of G_x(R[1/(v+p)]) uses the notation “A mod v/(v+p)” without explicitly indicating the ideal (namely v/(v+p) · R[1/(v+p)]); spelling this out would improve readability.","section":"Lemma 3.35"},{"comment":"The fixed choices list n, τ, x, and c, but the dependence of x on the choice of w and λ, and the role of the unramified extension L′ in Lemma 2.37(2), could be stated more explicitly so that the reader can track which choices are canonical and which are auxiliary.","section":"Section 2.7.1"},{"comment":"The phrase “it is also possible to be d-deep” appears to conflate the paper’s notion of d-genericity with the notion of depth used in [Le+23]; please align the terminology.","section":"Remark 2.47"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a reformatted thesis and is mathematically substantial. The main reason for major_revision rather than acceptance is that the advertised theorem statements need to carry the standing assumptions (ConFib and Dil, including F ⊇ F_q where needed), and Proposition 3.19 has a deferred general case that is load-bearing for the non-unramified setting. These are fixable without changing the core method. The paper’s explicit limitation statements are a strength, but they should be reflected in the abstract and theorem statements."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a solid PhD-thesis-level paper that genuinely extends the GLn/GSp4 results of [Le+23] and [Lee23] to arbitrary tamely ramified reductive groups, introducing Breuil–Kisin (Γ, Ĝ)-torsors as new objects and proving a local model theorem relating their bounded moduli stack Y^{≤μ} to Pappas–Zhu twisted Schubert varieties. It deserves a serious referee. But the advertised main theorem is slightly broader than what the proofs establish, because Assumption ConFib (connected fibers of the Bruhat–Tits group scheme Gx) is load-bearing.\n\nWhat is actually new: the moduli stack Y and its quotient presentation via the c-twisted Frobenius action, the invariant pushforward formalism in Section 2.6 and Appendix A, the building-theoretic classification of Galois types (Proposition 2.23), and the contraction/straightening machinery in Section 5 with explicit bounds. The proofs are detailed and the assumptions are stated up front. The paper is honest about its limitations: Remark 1.15 and Section 3.1.4 flag ConFib, Example 3.10 shows it can fail (PGL3 at an alcove barycenter), and the main theorems are stated conditionally there.\n\nThe soft spot is packaging, not correctness. Theorem A in the abstract and introduction reads unconditional, but Proposition 3.11's identification of Gx with the Pappas–Zhu group scheme requires connected fibers, and Sections 4–6 run under ConFib throughout. When ConFib fails, the central smooth equivalence is simply not proved. Since the author explicitly flags this, it is not a hidden flaw—but a referee should ask that the main theorems carry the hypothesis in their statements, or that the disconnected-fiber case be addressed.\n\nI found the standing assumptions clear and the contraction bounds plausible; I did not verify every estimate, but nothing looks circular. The results are conditional on ConFib (and Dil/genericity for Theorem B), which limits scope but is honestly disclosed. For people working on Emerton–Gee stacks or categorical p-adic Langlands, this is a useful and citable step.\n\nRecommendation: send it to peer review. The referee should check whether ConFib can be weakened and push for a statement of Theorem A that matches the proofs.","headline":"A careful, honest thesis-level generalization of the GLn local-model story to tamely ramified reductive groups; the main theorem is real work but conditional on an assumption the intro states more loosely than the proofs do.","tokens_in":74480,"tokens_out":2307,"would_cite":true,"duration_ms":31810,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14D23","11S37","20G25"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper constructs the moduli stack of Breuil–Kisin modules with reductive-group descent data and proves it is smoothly equivalent to a Pappas–Zhu twisted Schubert variety.","keywords":["Breuil-Kisin modules","descent data","reductive groups","affine Grassmannian","twisted Schubert variety","p-adic Langlands program","Bruhat-Tits group schemes","formal algebraic stacks"],"falsifier":"Take the explicit failure of connectedness from Example 3.10: $\\hat{G} = \\mathrm{PGL}_3$ with trivial $\\Gamma$-action and $x$ the barycenter of an alcove, so that $G_x$ has disconnected fibers. For a dominant $\\mu$ and a Frobenius-invariant type represented by $x$, compute the map $Z^{\\wedge\\varpi} \\to Y^{\\leq\\mu}$ constructed in Section 6; if any geometric fiber fails to be formally smooth, or the map is not a covering of $p$-adic formal algebraic stacks, then Theorem A fails without Assumption ConFib.","tokens_in":73388,"feed_emoji":"🧮","tokens_out":7778,"duration_ms":90210,"temperature":0.7,"pith_summary":"This paper introduces the moduli stack $Y^{\\leq \\mu}$ of Breuil–Kisin modules with $\\hat{G}$-structure and descent data, called Breuil–Kisin $(\\Gamma,\\hat{G})$-torsors, where $\\Gamma$ is the Galois group of a tamely ramified extension of $\\mathbb{Q}_p$ and $\\mu$ is a dominant cocharacter bounding the Hodge–Tate weights. The paper proves that $Y^{\\leq \\mu}$ is a $p$-adic formal algebraic stack and that it is smoothly equivalent to the $p$-adic completion of the Pappas–Zhu twisted Schubert variety $\\mathrm{Gr}^{\\leq \\mu}_G$ attached to $\\mu$. This equivalence gives a concrete loop-group model for a moduli problem defined through $p$-adic Hodge theory, extending to general reductive groups the relation previously known for $\\mathrm{GL}_n$ and used in the study of crystalline Emerton–Gee stacks and Serre weight conjectures. If the result is right, it makes the geometry of integral crystalline $L$-parameters for reductive groups accessible through affine Grassmannian techniques.","feed_headline":"Breuil-Kisin torsor stack matches twisted Schubert variety","feed_subtitle":"For reductive groups, bounded-height Breuil-Kisin modules with descent data form p-adic formal stacks with explicit loop-group models.","key_machinery":"The central object is the Bruhat–Tits group scheme $G_x$ over $\\mathbb{A}^1_O$, obtained as the invariant pushforward of a $\\Gamma$-twisted form of the dual group $\\hat{G}$ along the ramified cover $u^e = v$; under the connected-fiber assumption it coincides with the Pappas–Zhu group scheme attached to a point $x$ of the enlarged Bruhat–Tits building, where $x$ encodes the Galois type of the descent data. The stack is then realized as the quotient $Y^{\\leq \\mu} = [L^{\\leq \\mu}G_x^{\\wedge \\varpi} /_{c,\\varphi}\\, L^+G_x^{\\wedge \\varpi}]$ by the $c$-twisted Frobenius action $X \\star A = A^{-1} X \\varphi_c(A)$. The proof's engine is straightening: on $\\mathbb{Z}/p^a$-algebras, once the depth inequality $(p-1)\\lfloor f\\rfloor + d - h_\\mu - 2a + 2 > 0$ holds, the operator $A \\mapsto X\\varphi_c(A)X^{-1}$ is a contraction in the $v$-adic metric, so the $c$-twisted $\\varphi$-conjugation orbits coincide with left-translation orbits modulo sufficiently deep congruence subgroups; Banach's fixed point theorem turns one action into the other. The negative loop group $L^{--}G_x$ provides the affine open charts $U(z) = L^{--,\\leq \\mu}G\\cdot z$ in $\\mathrm{Gr}^{\\leq \\mu}$, which are used to split the relevant torsors Zariski locally.","core_discovery":"The paper's central claim is Theorem A (Theorem 6.4): for every dominant cocharacter $\\mu : \\mathbb{G}_m \\to \\hat{T}$, the substack $Y^{\\leq \\mu}$ of Breuil–Kisin $(\\Gamma,\\hat{G})$-torsors of height at most $\\mu$ is a $p$-adic formal algebraic stack over $\\mathrm{Spf}\\,O$, and there is a $p$-adic formal scheme $Z^{\\wedge \\varpi}$ over $\\mathrm{Spf}\\,O$ with smooth covering maps to both $Y^{\\leq \\mu}$ and $\\mathrm{Gr}^{\\leq \\mu,\\wedge\\varpi}_G$. In other words, the two stacks are smoothly equivalent. Under the additional hypotheses that $G$ is unramified, the derived group of $\\hat{G}$ is simply connected, and the point $x$ is lowest-alcove and $d$-generic with $d > h_\\mu$, Theorem B (Theorem 6.6) upgrades this to explicit Zariski-open charts, indexed by the admissible set $\\mathrm{Adm}(\\mu)$, on which the equivalence is fibered by $\\hat{T}$-torsors.","pith_inferences":["The paper leaves the monodromy condition to future work; if the smooth equivalence survives after imposing a monodromy condition, it would give local models for crystalline Emerton–Gee stacks for general reductive groups, not just the Breuil–Kisin side.","The connected-fiber assumption is probably not essential at the level of formal algebraic stacks: one could work with the neutral connected component of $G_x$ and track the finite component group through the quotient, with Example 3.10 suggesting the failure is a finite étale phenomenon rather than a breakdown of smoothness in the interior.","Because the straightening estimates are purely group-theoretic and often optimal, they could be implemented algorithmically to compute explicit $p$-adic charts, making deformation-ring computations for reductive groups more tractable."],"forward_implications":["For every dominant cocharacter $\\mu$, $Y^{\\leq \\mu}$ is a $p$-adic formal algebraic stack, giving finite-type approximations to the full moduli stack $Y$ of Breuil–Kisin $(\\Gamma,\\hat{G})$-torsors.","The smooth equivalence $Y^{\\leq \\mu} \\simeq \\mathrm{Gr}^{\\leq \\mu,\\wedge\\varpi}_G$ transfers geometric information from the explicit Pappas–Zhu local model to the Breuil–Kisin side.","Under unramified hypotheses with simply connected derived group, the equivalence admits explicit Zariski-open charts indexed by $\\mathrm{Adm}(\\mu)$, making the local model diagram concrete enough for deformation-ring computations.","The straightening inequalities quantify exactly which congruence subgroups of $L^+G_x$ are needed, with the bound $h_\\mu = \\max_{a\\in\\Phi}\\langle a,\\mu\\rangle$ shown to be sharp.","The theorem directly generalizes the local model diagrams of earlier $\\mathrm{GL}_n$-type and $\\mathrm{GSp}_4$ results to arbitrary tamely ramified reductive groups."],"supporting_citations":[{"why":"Constructs the Bruhat–Tits group schemes over the affine line and the twisted Schubert variety $\\mathrm{Gr}^{\\leq \\mu}_G$ that form the smooth-equivalence target.","marker":"[PZ13]"},{"why":"Supplies the three-step template and the prior $\\mathrm{GL}_n$-level relation between Breuil–Kisin moduli and twisted Schubert varieties that the paper generalizes.","marker":"[Le+23]"},{"why":"Originates the loop-group local model diagram and the Frobenius contraction arguments adapted in Section 5.","marker":"[PR09]"},{"why":"Proves a previous local model diagram for explicitly ramified groups, which Theorem A directly extends.","marker":"[CL18]"},{"why":"Treats the $\\mathrm{GSp}_4$ case with symplectic Breuil–Kisin modules and descent data, the immediate predecessor in the same family.","marker":"[Lee23]"},{"why":"Introduces Breuil–Kisin modules with $\\hat{G}$-structure without descent data at the level of deformation rings.","marker":"[Lev15]"},{"why":"Supplies the theory of Breuil–Kisin modules and their resolution of potentially crystalline deformation rings that motivates the whole construction.","marker":"[Kis09]"},{"why":"Provides the building-theoretic language, parahoric group schemes, and root-group filtrations used to define $G_x$ and the height and genericity conditions.","marker":"[KP23]"},{"why":"Shows that the affine Grassmannian is an fpqc sheaf, justifying the sheaf-theoretic definition of $\\mathrm{Gr}$ used here.","marker":"[Ces24]"}],"fun_headline_variants":["Breuil-Kisin torsor stack smoothly equivalent to twisted Schubert","p-adic formal stack from Breuil-Kisin modules matches Schubert","Breuil-Kisin torsor stack: a p-adic analog of Schubert varieties","Smooth equivalence: Breuil-Kisin torsor stack and twisted Schubert"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The main theorems assume that the group scheme $G_x$ has connected fibers; if this fails, the identification of $G_x$ with the Pappas–Zhu group scheme, and hence the smooth equivalence, can break, with an explicit failure for $\\hat{G} = \\mathrm{PGL}_3$ at the barycenter of an alcove.","fun_headline_variants_meta":{"raw":{"variants":["Breuil-Kisin torsor stack smoothly equivalent to twisted Schubert","p-adic formal stack from Breuil-Kisin modules matches Schubert","Breuil-Kisin torsor stack: a p-adic analog of Schubert varieties","Smooth equivalence: Breuil-Kisin torsor stack and twisted Schubert"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000304,"raw_usage":{"total_tokens":1754,"prompt_tokens":963,"completion_tokens":791,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":579,"completion_tokens_details":{"reasoning_tokens":709}},"tokens_in":579,"tokens_out":791,"duration_ms":8890,"temperature":1.0,"reasoning_tokens":709,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T01:00:26.200565+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the explicit failure of connectedness from Example 3.10: $\\hat{G} = \\mathrm{PGL}_3$ with trivial $\\Gamma$-action and $x$ the barycenter of an alcove, so that $G_x$ has disconnected fibers. For a dominant $\\mu$ and a Frobenius-invariant type represented by $x$, compute the map $Z^{\\wedge\\varpi} \\to Y^{\\leq\\mu}$ constructed in Section 6; if any geometric fiber fails to be formally smooth, or the map is not a covering of $p$-adic formal algebraic stacks, then Theorem A fails without Assumption ConFib.","supporting_citations":[],"review_version":1}