{"id":"dfdca0c1-9ecb-4c8a-a44b-27f9059ca7bf","arxiv_id":"2411.16054","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A stratification and smoothening method gives exact stable orbital integral formulas for gl2, gl3, and u2, and new lower bounds for all n.","lead":"This paper introduces a geometric method for computing stable orbital integrals, the p-adic volumes of matrices with a fixed characteristic polynomial, for general linear, unitary, and symplectic Lie algebras. It obtains exact formulas in small ranks and improved lower bounds in all ranks, with conjectures on optimality.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Any-characteristic gl2/gl3 formulas rely on a char-restricted reduction: Corollary 3.3 (via Proposition 2.4) supplies the S=0 unramified endpoint, so the proof does not cover char(F)=2 or 3 as stated.","rationale":"The reader's conditional verdict is appropriate, and their identification of the measure comparison as the fragile assumption is related, but the specific load-bearing issue is sharper: the papers' unconditional any-characteristic closed formulas for gl2 and gl3 depend on Corollary 3.3 at the S=0 unramified endpoint, even though Corollary 3.3 is justified only for char(F)=0 or char(F)>n. This does not invalidate the whole paper, because the formulas are very likely true and the gap is localized and repairable, but it does mean the proof as written is incomplete for small characteristics. The rest of the analysis, including the stratification framework and the explicitly conditional lower bounds, appears consistent; no internal inconsistency beyond this endpoint gap was found. The verdict therefore remains conditional, with the added condition that the S=0 case and the low-characteristic reduction must be addressed directly or the theorem statements must be restricted.","tokens_in":88686,"tokens_out":31672,"duration_ms":269560,"concrete_test":"Take F=κ((t)) with char(κ)=2 and γ∈gl2(o) whose characteristic polynomial reduces to an irreducible quadratic over κ, e.g., x^2+x+1. Compute SOγ directly from Lemma 2.3 as lim_{N→∞} q^{-2N} #{A mod t^N : χ_A(x) ≡ χ_γ(x)}. If the limit equals (q-1)/q, the formula survives and the gap is a missing proof; if not, Theorem 1.3(1) fails in char 2. Equivalently, try to re-derive Corollary 3.3 for the geometric measure without invoking Proposition 2.4; verifying the constant q^{n^2-n^2/l} in all characteristics would close the gap.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper states Theorems 1.3(1) and 1.4(1) as holding for a local field F of any characteristic, but the proof chain does not support this. The reduction to the case where χγ(x) reduces to x^n is made in Section 3.1 using Lemma 3.2 and Proposition 2.4; the latter is explicitly proved only under char(F)=0 or char(F)>n. This reduction is not merely a convenience: it is used at the unramified S(γ)=0 endpoint. In the proof of Theorem 5.4 (gl2, unramified case), after summing k1=0,...,S(γ)-1, the terminal term is q^{-S(γ)}SO_{γ(S(γ))}, and when S(γ)=0 this is exactly the original element whose reduction is irreducible (not x^2). The value (q-1)/q is then obtained by invoking Corollary 3.3, which is derived from Proposition 2.4. Thus, for char(F)=2, n=2, the stated theorem is not proved. The same issue occurs for gl3 in Appendix B, where the l=0 term in the unramified case is evaluated by Corollary 3.3, leaving char(F)=3 uncovered. No direct computation of the S=0 endpoint via smoothness or Lemma 2.3 is supplied. This is an internal proof gap in the central unconditional closed formulas, not merely a restriction on the conditional lower bounds or the dμ comparison.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a new geometric method, based on stratifying the set of matrices with fixed characteristic polynomial and then smoothening each stratum, to compute stable orbital integrals for the Lie algebras gl_n, u_n, and sp_2n over a non-Archimedean local field F. For gl_2 and gl_3 it obtains exact closed formulas (Theorems 1.3 and 1.4), for u_2 a closed formula under a hypothesis on the splitting behavior (Theorem 1.11), and for general n a lower bound (Theorems 1.6 and 1.10) that improves the second leading term of Yun's lower bound. The paper also formulates conjectures about the optimality of these bounds. Part 1 contains detailed proofs, including appendices for gl_3; Part 2 is conditional on explicit factorization, characteristic, and residue-characteristic assumptions, which the authors state carefully.","tokens_in":41,"tokens_out":8086,"duration_ms":191956,"significance":"If the results are correct, the closed formulas for gl_2 and gl_3 are genuine advances, as is the improvement over Yun's lower-bound second leading term in Theorem 1.6. The stratification-and-smoothening method, built on modules over a PID and Weil's formula, is original and likely to be transferable to other settings. The paper is careful in attributing earlier results (Weil, FLN, Yun, Gross, Gordon) and in stating the hypotheses needed in Part 2. The exposition is detailed, with proofs of the gl_3 formulas relegated to appendices, and the conjectures are presented with concrete evidence. The main reservation is that the claim of characteristic-free validity for the gl_2 and gl_3 closed formulas is not supported by the proof chain.","major_comments":[{"comment":"Theorems 1.3(1) and 1.4(1) are stated for local fields of arbitrary characteristic, but the proof chain does not support this. The reduction to the case where the reduction of χγ(x) is x^n is made in Section 3.1 using Corollary 3.3, which is derived from Proposition 2.4 under the assumption char(F)=0 or char(F)>n. This restriction is inherited at the unramified S(γ)=0 endpoint in the proof of Theorem 5.4: the terminal term q^{-S(γ)}SO_{γ(S(γ))} is evaluated using Corollary 3.3, and when S(γ)=0 this is the original element with irreducible reduction, not one with reduction x^2. The same issue occurs for gl_3 in Appendix B, where the l=0 term in the unramified case is evaluated by Corollary 3.3, leaving char(F)=3 uncovered. No direct computation of the S=0 endpoint via Lemma 2.3 or via smoothness is supplied. Therefore the stated any-characteristic conclusions are not established for char(F)=2 (gl_2) and char(F)=3 (gl_3); the theorems should either be restricted to char(F)=0 or char(F)>n, or the missing endpoint computation must be provided.","section":"§3.1, Corollary 3.3; §5.1, proof of Theorem 5.4; Appendix B"}],"minor_comments":[{"comment":"The remark immediately following the Notations section of Part 2 is numbered 'Remark 7.1', but it belongs to a later section (near Section 8); the numbering should be corrected.","section":"Part 2, Notations"},{"comment":"Proposition 5.1 states a result for 'a prime number n', but the proof implicitly uses n ≥ 2; the n=1 case is trivial (no translation is needed) and could be mentioned to avoid an edge-case gap in the exposition.","section":"Proposition 5.1"},{"comment":"In the proof of Theorem 5.4, the notation SO_{γ(S(γ))} is used before the definition of γ(k) is recalled; adding a forward reference to Proposition 3.17 or a one-line reminder would improve readability.","section":"Theorem 5.4 proof"}],"recommendation":"major_revision","confidential_remarks":"The main issue is the unproven characteristic-free claim for the gl_2 and gl_3 closed formulas. I believe the gap is fixable either by a direct computation of the S(γ)=0 endpoint in small characteristic or by adjusting the theorem statements to char(F)=0 or char(F)>n. Given that the rest of the paper is carefully written and the conditional results are clearly stated, a major revision with this fix would be appropriate. It may also be worth asking the authors to double-check the proof of Corollary 3.3 and its dependence on Proposition 2.4 in other places where the reduction to x^n is used."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I read Cho–Kang–Lee with care. The paper is substantial and mostly honest. The main new thing is the stratification-plus-smoothening framework applied to stable orbital integrals, and it delivers: exact closed formulas for gl2 and gl3 in Part 1, a u2 formula in Part 2, and a lower bound for all n whose second leading term improves Yun's. The small-rank computations are worked out in detail, the appendices are real proofs, and the authors state most of their restrictions explicitly. That is genuine progress, not a repackaging of Yun.\n\nNow the soft spots, in proportion. The stress-test concern about characteristic restrictions is correct and lands on a load-bearing part of the stated theorems. Theorem 1.3(1) and Theorem 1.4(1) are announced for a local field of any characteristic, but the proof chain does not support that. The reduction to the case where the reduction of chi_gamma is x^n uses Corollary 3.3, which comes from Proposition 2.4; Proposition 2.4 is proved only for char(F)=0 or char(F)>n. At the S(gamma)=0 endpoint, the gl2 proof invokes Corollary 3.3 to get the terminal value (q-1)/q, and the gl3 proof does the same for the l=0 term. No direct computation of that endpoint is supplied. So for char(F)=2 (gl2) and char(F)=2,3 (gl3), the formulas as stated are not proved. This is an internal gap, not a charitable quibble about the measure comparison. It does not kill the rest of the paper—the conditional results and the gln lower bound are unaffected—but the headline \"any characteristic\" in the abstract overstates what is established.\n\nPart 2 is more conditional by design: the un/sp lower bounds assume char(F)=0 or char(F)>n, plus a singleton irreducible component, plus for even n in the unitary case char(kappa)>2. Those conditions are clearly flagged. The comparison between measures in Section 8.5 is delicate and the proof leans on [FLN10]/[Gor22] plus a Jacobian argument; I did not find a hole, but the assumptions are doing real work.\n\nWho this is for: people working on trace formulas, orbital integrals, and p-adic integration. They will get value even from the structure alone. The citation pattern looks appropriate: Yun, FLN, Gor22, GY00, CY20 are the natural sources, and the authors are not inflating claims about novelty.\n\nMy recommendation: this deserves a serious referee, but the authors should be asked to either prove the S=0 endpoint for small characteristic or change Theorems 1.3(1) and 1.4(1) to the characteristic range the proof actually covers. That is a substantive revision, not a cosmetic one.","headline":"Genuine new formulas and a real method, but the any-characteristic claim for the gl2/gl3 formulas is not supported by the proof chain.","tokens_in":89490,"tokens_out":1244,"would_cite":true,"duration_ms":16214,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11F72","11S80","14B05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Stable orbital integrals over p-adic fields can be computed by stratifying matrices and smoothing each stratum.","keywords":["stable orbital integrals","non-Archimedean local fields","smoothening","stratification","classical Lie algebras","Chevalley morphism","p-adic volumes","orbital integrals"],"falsifier":"Compute $SO_\\gamma$ for an explicit elliptic $\\gamma\\in\\mathfrak{gl}_4(o)$ with $\\chi_\\gamma\\equiv x^4\\pmod{\\pi}$ and a chosen $d_\\gamma$ by evaluating $\\lim_{N\\to\\infty} q^{-N(4^2-4)}\\#G_\\gamma(o/\\pi^N o)$ for a small residue field, say $q=2$ or $q=3$; if the computed value is not strictly larger than the right-hand side of Theorem 7.8, the lower bound is false.","tokens_in":88497,"feed_emoji":"🧮","tokens_out":7235,"duration_ms":67408,"temperature":0.7,"pith_summary":"This paper claims that the stable orbital integral for the unit element of the Hecke algebra on the Lie algebras $\\mathfrak{gl}_{n}$, $\\mathfrak{u}_{n}$, and $\\mathfrak{sp}_{2n}$ over a non-Archimedean local field can be described by grouping the matrices in the stable orbit according to the sublattices they map onto, then smoothing each group by extra congruence conditions. If the description is correct, the integral becomes a finite sum of volumes of smooth schemes, each computed by Weil's point-counting formula. The paper obtains closed rational formulas for $\\mathfrak{gl}_{2}$, $\\mathfrak{gl}_{3}$, and $\\mathfrak{u}_{2}$, and lower bounds for all $n$ whose second leading term improves an earlier lower bound. It also proposes conjectures asserting that these leading terms are optimal. A reader would care because exact formulas for p-adic orbital integrals are rare and these integrals feed local factors in automorphic and Siegel-type formulas.","feed_headline":"Exact p-adic orbital integrals for gl2, gl3, and u2","feed_subtitle":"A stratification-and-smoothening method also improves lower bounds for gln, un, and sp2n.","key_machinery":"The load-bearing object is the Chevalley morphism $\\phi_n:\\mathfrak{g}\\to\\mathbb{A}^n_o$ sending a matrix to the coefficients of its characteristic polynomial, together with the stratification of its fibre $G_\\gamma(o)$ by sublattices $M$ of type $(k_1,\\ldots,k_{n-m})$ and subspaces $V$ of dimension $m-t$. The mechanism is smoothening: each stratum is realized as a scheme $L(L,M,V)$ (or $L(L,M)$) whose generic fibre is the same but whose special fibre becomes smooth after imposing congruence conditions, so that the quotient volume form $\\omega^{\\mathrm{ld}}_{\\chi_\\gamma}$ is computed by the point-counting formula of [Wei12, Theorem 2.2.5]. Counting sublattices of each type through Grassmannians over the residue field $\\kappa$, and subspaces of each dimension through the $d_t$, supplies the weights $c_{(k_1,\\ldots,k_{n-m})}$ and $d_t$ in the summation formulas.","core_discovery":"The paper's central claim is that the stable orbital integral $SO_\\gamma$ for a regular semisimple element $\\gamma\\in\\mathfrak{gl}_{n}(o)$, and under extra hypotheses for $\\mathfrak{u}_{n}$ and $\\mathfrak{sp}_{2n}$, can be computed exactly through a finite stratification. Writing $G_\\gamma(o)$ for the set of matrices in $\\mathfrak{gl}_{n}(o)$ with characteristic polynomial $\\chi_\\gamma$, every $f\\in G_\\gamma(o)$ has image a rank-$n$ sublattice $M\\subset L$, and the index $[L:M]$ is determined by the constant term of $\\chi_\\gamma$. Grouping by the isomorphism type $(k_1,\\ldots,k_{n-m})$ of $L/M$, and then by the image subspace $V$ of the induced map on $\\overline{M}$, reduces $SO_\\gamma$ to weighted sums of finer volumes $SO_{\\gamma,(k_1,\\ldots,k_{n-m}),t}$. The paper shows that after imposing suitable congruence conditions each such stratum becomes smooth over $o$, so Weil's formula computes its volume from the point count of its reduction. On this basis it proves closed formulas for $\\mathfrak{gl}_{2}$, $\\mathfrak{gl}_{3}$, and $\\mathfrak{u}_{2}$, proves the lower bound in Theorem 1.6 whose second leading term improves an earlier bound, and proposes conjectures asserting that the relevant leading terms are optimal.","pith_inferences":["Beyond the paper: the stratification step reduces the degree of the polynomial constraints by one, so the method suggests that a fully general treatment of $\\mathfrak{gl}_n$ for $n\\geq 4$ would need further geometric reductions that lower the degree by more than one.","Beyond the paper: one can test Conjecture 1.12 numerically for $n=4$ and small $q$ by computing $SO_\\gamma$ through the limit formula $\\lim_{N\\to\\infty} q^{-N(n^2-n)}\\#G_\\gamma(o/\\pi^N o)$; if the coefficient of $q^{-d}$ differs from the conjectured $\\alpha(\\overline{d}_\\gamma)$, the conjecture fails.","Beyond the paper: the gap between the lower bound and the true value should be governed by strata of types $(k_1,\\ldots,k_{n-m})$ with more than two parts, and a natural next step is to quantify how those strata contribute after the smoothening used in Theorem 1.6."],"forward_implications":["The closed formulas for $\\mathfrak{gl}_2$, $\\mathfrak{gl}_3$, and $\\mathfrak{u}_2$ make the stable orbital integral an explicit rational function of the residue cardinality $q$ and the Serre invariant $S(\\gamma)$.","The lower bound for $\\mathfrak{gl}_n$ improves the second leading term of the earlier bound, so for large $q$ the true orbital integral is pinned down to one further power of $q^{-1}$.","The conjectured optimality of the second leading term for $\\mathfrak{gl}_n$ and first leading term for $\\mathfrak{u}_n$ and $\\mathfrak{sp}_{2n}$ would identify the precise asymptotic size of these stable orbital integrals.","The method applies to $\\mathfrak{u}_n$ and $\\mathfrak{sp}_{2n}$ only under hypotheses on the factorization of the characteristic polynomial and, for even $n$ in the unitary case, a condition on the residue characteristic.","The same stratification and smoothening recipe is presented as promising for other orbital problems, including local densities and Siegel series."],"supporting_citations":[{"why":"Supplies the quotient-measure normalization, the parabolic descent formula, the Serre invariant $S(\\gamma)$, and the earlier lower bound whose second leading term is improved.","marker":"[Yun13]"},{"why":"Gives the definition of the stable orbital integral and the measure comparison used to translate between geometric and quotient measures.","marker":"[FLN10]"},{"why":"Provides the refined comparison of measures and the identification of regular semisimple elements through the discriminant.","marker":"[Gor22]"},{"why":"Supplies Weil's formula for volumes of smooth schemes, used after each stratum is smoothed.","marker":"[Wei12]"},{"why":"Gives the smoothness of the Chevalley morphism at regular elements, so the smooth strata have the right dimension.","marker":"[Hum95]"},{"why":"Introduces the congruence-condition smoothening technique and quotient volume forms that the paper adapts.","marker":"[GY00]"},{"why":"Provides the volume-form lemma that computes such integrals as limits of point counts.","marker":"[CY20]"},{"why":"Supplies the ft-Néron models of centralizer tori whose residue counts enter the unitary and symplectic comparisons.","marker":"[KP23]"},{"why":"Fixes the quotient-measure normalization for the unitary and symplectic cases.","marker":"[Yun16]"}],"fun_headline_variants":["Exact orbital integrals for gl2, gl3, u2 via smoothing","Sharper lower bounds for stable orbital integrals on gl_n, u_n, sp_2n","New closed forms for p-adic orbital integrals","Stratify and smooth: exact p-adic orbital integrals","Exact formulas and conjectured optimal bounds for orbital integrals"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The quantitative claims rely on the comparison between the geometric measure and the quotient measure, which the paper proves only when $\\mathrm{char}(F)=0$ or $\\mathrm{char}(F)>n$.","fun_headline_variants_meta":{"raw":{"variants":["Exact orbital integrals for gl2, gl3, u2 via smoothing","Sharper lower bounds for stable orbital integrals on gl_n, u_n, sp_2n","New closed forms for p-adic orbital integrals","Stratify and smooth: exact p-adic orbital integrals","Exact formulas and conjectured optimal bounds for orbital integrals"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001052,"raw_usage":{"total_tokens":4496,"prompt_tokens":1099,"completion_tokens":3397,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":715,"completion_tokens_details":{"reasoning_tokens":3305}},"tokens_in":715,"tokens_out":3397,"duration_ms":21623,"temperature":1.0,"reasoning_tokens":3305,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:36:07.923177+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $SO_\\gamma$ for an explicit elliptic $\\gamma\\in\\mathfrak{gl}_4(o)$ with $\\chi_\\gamma\\equiv x^4\\pmod{\\pi}$ and a chosen $d_\\gamma$ by evaluating $\\lim_{N\\to\\infty} q^{-N(4^2-4)}\\#G_\\gamma(o/\\pi^N o)$ for a small residue field, say $q=2$ or $q=3$; if the computed value is not strictly larger than the right-hand side of Theorem 7.8, the lower bound is false.","supporting_citations":[],"review_version":1}