{"id":"5d769d30-9f58-4d53-8627-23db689cdfb9","arxiv_id":"2507.05003","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For SL(2) over local fields of any characteristic, including p=2 where Shalika's expansion fails, the paper proves an endoscopic germ expansion for orbital integrals near the center and states a conjecture for arbitrary groups.","lead":"This paper gives a way to describe orbital integrals near the identity for the group SL(2) over local fields of characteristic 2, where the classical Shalika expansion breaks down because there are too many unipotent conjugacy classes. It uses endoscopy, a transfer between groups, to supply a replacement germ expansion in all characteristics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 7.3 depends on endoscopic transfer in characteristic 2 (Thm 6.3), which is cited but not proved; the paper's split-torus transfer formula in §7 is also internally inconsistent.","rationale":"The central claim is Theorem 7.3. For elliptic tori, the proof is: Fourier inversion O_G = SO_G + O^ε, Lemma 7.1 gives the stable germ expansion via GL(2), and Lemma 7.2 gives the unstable term via endoscopic transfer. The unstable term is entirely dependent on Theorem 6.3. The paper does not prove Theorem 6.3; it invokes [LL, Lemma 2.1] in characteristic zero with the assertion that the characteristic assumption is 'seldom used' and refers to [L] for details. This is a genuine verification gap because the whole point of the paper is the p=2 case, where ordinary Shalika theory breaks down. The transfer identity is exactly where the characteristic behavior enters through Q_F and flat cohomology, so the citation cannot be waved through without checking. The reader's CONDITIONAL verdict is therefore appropriate. I additionally noticed a separate inconsistency in the split-torus part of §7: the claimed identity O_G(t,f)=Δ_E(t)^{-1}SO_G(zu_0,f) with f^E(z)=SO_G(zu_0,f) is false for functions supported near 1 but away from u_0; however Theorem 7.3's equation (1) for split tori reduces to O_G(t,f)=SO_G(t,f), which is the correct statement, so this passage is extraneous to the central theorem rather than fatal. Agreement: the reader's weakest assumption correctly identifies Theorem 6.3; the concrete test above would verify that transfer in characteristic 2.","tokens_in":6971,"tokens_out":39552,"duration_ms":484392,"concrete_test":"Take F=F_2((X)), let E/F be the Artin-Schreier quadratic extension E=F(Y)/(Y^2+Y=X^{-1}), and let f be the characteristic function of SL(2,O_F). Use the explicit transfer formula of [LL, Lemma 2.1] as adapted in [L] to define f^E on E^1. For three regular elliptic elements t_1,t_2,t_3 approaching 1, compute both sides of SO_{E^1}(ι(t_i), f^E) = Δ_E(t_i) O_G^{ε_{E/F}}(t_i, f). If equality fails in any case, Theorem 6.3 is false in characteristic 2 and Theorem 7.3 collapses. If equality holds, also test the split-torus passage by taking f supported in a neighborhood of 1 excluding u_0 and checking whether O_G(t,f) equals Δ_E(t)^{-1}SO_G(u_0,f) for split t near 1; a nonzero left side with zero right side confirms the §7 split passage is a genuine error.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central expansion for elliptic tori reduces via Fourier inversion to Lemma 7.2, which asserts that O_G^{ε_{E/F}}(t,f) = Δ_E(t)^{-1} f^E(z) for t near z. This identity is only as solid as Theorem 6.3, the existence of an endoscopic transfer f^E in characteristic p=2. Theorem 6.3 is not proved in this paper: it cites [LL, Lemma 2.1], a characteristic-zero computation, with the note that the characteristic assumption is 'seldom used', and the detailed proof is relegated to the author's lecture notes [L]. The characteristic matters exactly where ordinary Shalika theory fails: in characteristic 2, Q_F is uncountable and flat cohomology is needed. If the construction of f^E or the limiting identity f^E(z) = lim_{t→z} Δ_E(t) O^ε(t,f) does not survive p=2, Lemma 7.2 and hence Theorem 7.3 fail for elliptic tori. A separate red flag is the split-torus passage in §7: it states O_G(t,f)=Δ_E(t)^{-1}SO_G(zu_0,f) with f^E(z)=SO_G(zu_0,f); for f supported in a small neighborhood of 1 that excludes u_0, the left side is nonzero for t near 1 while the right side is zero, so the split passage is an error. Theorem 7.3 itself only uses the elliptic data term {G,1} for split tori, so this error does not by itself refute the central theorem, but it strengthens the need to verify the transfer machinery.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a substitute for Shalika's germ expansion for SL(2) over local fields of arbitrary characteristic. The motivating problem is that in characteristic 2 the non-trivial unipotent conjugacy classes of SL(2,F) form an uncountable set, so the usual Shalika expansion is not available. The author instead uses endoscopic transfer: Theorem 7.3 asserts that, for regular semisimple t sufficiently close to a central element z, the orbital integral O_G(t,f) can be written as a sum over elliptic endoscopic data of terms Δ_E(t)^{-1} SO_H(ι(t), f^E). The proof combines Fourier inversion over Q_F, a stable germ expansion obtained from GL(2), and an endoscopic transfer theorem (Theorem 6.3). A conjecture for arbitrary quasi-split groups is also stated.","tokens_in":7293,"tokens_out":11361,"duration_ms":132271,"significance":"If the main theorem is correct, it is a genuine and interesting advance: it gives a germ expansion in characteristic 2, where Shalika's expansion is meaningless because of the uncountably many unipotent classes. The Fourier-inversion reformulation of the problem and the use of endoscopy are elegant, and the paper is short and readable. The statement of Conjecture 8.1 is also valuable as a roadmap, and it properly acknowledges a prior error. However, the central result is conditional on an endoscopic transfer theorem whose proof for positive characteristic is not contained in this manuscript, and one passage in Section 7 contains a demonstrably false split-torus identity. These issues must be resolved before the main claim can be regarded as established.","major_comments":[{"comment":"The existence of an endoscopic transfer f^E for positive characteristic, and especially for p=2, is load-bearing for Theorem 7.3, but it is not proved here. The proof cites [LL, Lemma 2.1], a characteristic-zero computation, with the assertion that the characteristic hypothesis is 'seldom used'; that is not a verification. Since Lemma 7.2 and hence the elliptic-torus case of Theorem 7.3 depend entirely on this theorem, the manuscript must either include a complete proof valid for all characteristics, or give a precise reference to a proof in [L] that explicitly covers p=2. The difficulty is not cosmetic: in characteristic 2 the quotient Q_F is uncountable and flat cohomology is needed, so the positive-characteristic transfer requires new arguments beyond the cited characteristic-zero lemma.","section":"Section 6, Theorem 6.3"},{"comment":"The displayed identity O_G(t,f)=Δ_E(t)^{-1} f^E(z) for split tori, with f^E(z)=SO_G(zu_0,f), is inconsistent with Definition 6.2. For a split torus, Definition 6.2 gives f^E(ι(t))=Δ_E(t)O_G(t,f), so passing to the limit t→z yields f^E(z)=lim_{t→z} Δ_E(t)O_G(t,f), not O_G(t,f)=Δ_E(t)^{-1}f^E(z). Concretely, if f is supported in a small neighbourhood of the identity that excludes zu_0, then O_G(t,f) is nonzero for regular split t close to 1, while SO_G(zu_0,f)=0, so the displayed formula would give zero on the right and a nonzero value on the left. This passage should be deleted or corrected. It does not by itself refute Theorem 7.3, because the theorem only uses the elliptic datum {G,1} for split tori, but the error must be fixed.","section":"Section 7, split-torus paragraph"},{"comment":"The proof of Lemma 7.2, which states only 'This follows from 6.3', omits a non-formal limiting argument. Theorem 6.3 gives SO_H(ι(t), f^E)=Δ_E(t)O_G^κ(t,f), hence O_G^κ(t,f)=Δ_E(t)^{-1}SO_H(ι(t), f^E). To obtain Δ_E(t)^{-1}f^E(z), one must justify that SO_H(ι(t), f^E) tends to f^E(z) and that the singularity of Δ_E(t)^{-1} is matched by the behaviour of the orbital integral. This is not automatic from the transfer identity alone and should either be proved in Lemma 7.2 or incorporated into the statement and proof of Theorem 6.3.","section":"Section 7, Lemma 7.2"}],"minor_comments":[{"comment":"There is a typo in 'charateristic' in the first paragraph.","section":"Section 2"},{"comment":"The notation SU G is used both for the set of stable unipotent classes and for the vector space of stable distributions with true unipotent support; these should be given distinct notations to avoid confusion.","section":"Section 8"},{"comment":"The constant c in the transfer factor is said to be irrelevant, but the final expansion in Theorem 7.3 must be independent of c, τ, and the choice of Haar measures; this independence is asserted but not demonstrated.","section":"Definition 6.1"},{"comment":"The statement that 'Assumptions of Shalika's theorem [Sh] are fulfilled for eG(F) in any characteristic' needs support: Shalika's theorem is usually stated for characteristic zero, and while the GL(2) case can be checked directly, a reference or a short argument for positive characteristic should be supplied.","section":"Lemma 7.1"}],"recommendation":"major_revision","confidential_remarks":"The main theorem depends on a transfer theorem whose positive-characteristic proof is deferred to the author's lecture notes [L]. If the journal is willing to accept a reference to a public arXiv preprint for a key technical step, this could be acceptable; otherwise the author should include the proof. The split-torus paragraph is a clear mathematical error and should be removed. After these revisions, the paper would be publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is Conjecture 8.1, and even that is a carefully hedged conjecture rather than a theorem. The SL(2) germ expansion in Theorem 7.3 is, as the introduction admits, a restatement of results proved in the author's lecture notes [L]. So the paper's value is as a compact research announcement plus a conjecture, not as a self-contained proof.\n\nThe paper does several things well. It explains clearly why Shalika's germ expansion breaks down in characteristic 2 (uncountably many unipotent classes) and why an endoscopic expansion is a natural substitute. The conjecture is thoughtfully formulated, with the notion of \"true unipotent\" introduced to avoid the primitive-element obstruction the author found in PGL(2). The acknowledgement to Waldspurger for catching a major mistake in an earlier version tells me the conjecture has already been stress-tested. That kind of honesty deserves credit.\n\nThe soft spots are real but not uniformly serious. The load-bearing step is Theorem 6.3, the existence of endoscopic transfer in characteristic 2. That theorem is not proved here; it is cited from [LL] with the remark that the characteristic assumption is \"seldom used,\" and from [L]. If you are going to accept Theorem 7.3, you need to trust [L] for exactly the case where Shalika's theory fails. That is a genuine dependency, and the paper should be read as conditional on it.\n\nThere is also a clear error in the split-torus passage. The text claims that for t close to z in a split torus, O_G(t,f) = Δ_E(t)^{-1} SO_G(zu_0,f), with f^E(z) equal to that stable unipotent orbital integral. But if f is supported in a small neighborhood of 1 that avoids u_0, the left side is nonzero for t near 1 while the right side is zero. The identity is false as stated. It does not undermine Theorem 7.3, because split tori are handled by the {G,1} term where the assertion reduces to O = SO, but it is an error that should be corrected before publication.\n\nWho is this for? Specialists in local harmonic analysis and endoscopy in positive characteristic, and anyone working on the Tamagawa numbers problem. The conjecture is a reasonable target for future work. I would not cite this paper in my own writing over the next year; I would cite [L] for the SL(2) theorem and cite the conjecture separately if I needed it. But the conjecture is stated clearly enough to be worth public record.\n\nRecommendation: send it to peer review. The referee should verify the transfer machinery in [L] and require the split-torus formula to be fixed. With those two conditions, the note is publishable as a research announcement.","headline":"The SL(2) germ expansion is a deferred summary of the author's own lecture notes, and the genuinely new item is Conjecture 8.1; a plausible paper with one real gap and one local error in the split-torus passage.","tokens_in":7868,"tokens_out":4938,"would_cite":false,"duration_ms":54974,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["22E35","11F70","20G25","11S37"],"pacs":[],"model":"deepseek-v4-flash","headline":"Orbital integrals on SL(2) have a germ expansion in every characteristic.","keywords":["SL(2)","germ expansion","orbital integrals","endoscopy","Shalika germs","positive characteristic","unipotent conjugacy classes","local fields"],"falsifier":"Take F = F_2((X)), G = SL(2), and a test function f supported close to the identity; compute O_G(t_n,f) for a sequence of regular elliptic t_n tending to z and compare with the right side of Theorem 7.3, constructing f^E by the explicit formula cited from [LL, Lemma 2.1]. If the difference does not tend to zero for some f, the transfer identity fails in characteristic 2 and Theorem 7.3 is false.","tokens_in":6726,"feed_emoji":"🧮","tokens_out":10986,"duration_ms":103895,"temperature":0.7,"pith_summary":"The paper gives a substitute for Shalika's germ expansion of orbital integrals that works for SL(2) over any non-archimedean local field, including characteristic 2, where Shalika's expansion does not exist because the non-trivial unipotent conjugacy classes form an uncountable set. The substitute expresses the orbital integral of a regular semisimple element near the center as a sum over elliptic endoscopic data: for each quadratic extension E/F or the split data, a stable orbital integral on an endoscopic group, multiplied by an explicit transfer factor. For p different from 2 this endoscopic expansion is equivalent, up to a Fourier transform, to the classical Shalika expansion; for p=2 it is new. The paper also states a conjecture extending this kind of stable germ expansion to arbitrary quasi-split reductive groups.","feed_headline":"Germ expansion for SL(2) survives characteristic 2","feed_subtitle":"Shalika's expansion fails at p=2; an endoscopic sum replaces it with one that still works.","key_machinery":"The central machinery is the endoscopic expansion of orbital integrals via κ-orbital integrals and Fourier inversion over the compact quotient Q_F = (F^×)^2 \\ F^×. Characters κ of Q_F parametrize elliptic endoscopic data; the κ-orbital integral O_G^κ(t,f) vanishes except for κ = 1 or κ = ε_{E/F}, and Fourier inversion gives O_G(t,f) = Σ_κ O_G^κ(t,f). The transfer factor Δ_E(t) = c ε_{E/F}((ι(t)-ι(t))/(ι(τ)-ι(τ))) |ι(t)-ι(t)|_E and the existence of an endoscopic transfer f^E satisfying SO_H(ι(t),f^E) = Δ_E(t) O_G^κ(t,f) carry the reduction to stable orbital integrals on GL(2) and endoscopic tori, where ordinary Shalika germs are available in all characteristics.","core_discovery":"Theorem 7.3 asserts that for G = SL(2) over a non-archimedean local field F of arbitrary characteristic, if t is regular semisimple and close enough to a central point z, then O_G(t,f) = sum over elliptic endoscopic data E = {H,kappa} of Φ_E(t), where Φ_E(t) = Δ_E(t)^{-1} SO_H(ι(t), f^E) if the torus of t embeds in H and Φ_E(t) = 0 otherwise. The elliptic data are {G,1} and {T_{E/F}, ε_{E/F}} for separable quadratic extensions E/F. This yields a germ expansion in every characteristic. When p ≠ 2 the endoscopic sum is equivalent, up to a Fourier transform on Q_F, to Shalika's classical germ expansion; when p = 2, where the unipotent conjugacy classes are uncountable and Shalika's expansion is undefined for elliptic elements, it is a new expansion.","pith_inferences":["The paper's stated motivation is to provide a uniform local ingredient for a characteristic-independent proof of Weil's conjecture on Tamagawa numbers; if the endoscopic germ expansion extends to inner forms, it would supply exactly that ingredient for the global argument.","A concrete next test of Conjecture 8.1 is PGL(2) over F_2((X)), where the paper itself notes that a primitive rational element becomes unipotent only after a quadratic extension; the finite-dimensionality of stable distributions with true unipotent support could be checked there.","If the transfer identity in Theorem 6.3 is the only missing piece, an independent proof of it in characteristic 2, not relying on the cited characteristic-zero lemma, would make Theorem 7.3 fully self-contained; the paper leaves that proof as cited work."],"forward_implications":["For any regular semisimple t near z, the orbital integral decomposes as a sum over elliptic endoscopic data; for an elliptic torus attached to E/F it is explicitly SΓ_z^G(t)f(z) + SΓ_{z\\tilde u0}^G(t)SO_G(zu0,f) + Δ_E(t)^{-1}f^E(z).","At p ≠ 2 the classical Shalika expansion is recovered up to a Fourier transform on Q_F, so the endoscopic germ expansion gives a common framework covering all characteristics.","At p = 2 the result supplies the missing local germ expansion, controlling the unipotent contribution to harmonic analysis on SL(2) even though no Shalika expansion exists.","The proof strategy, Fourier inversion over Q_F plus transfer to GL(2), provides a template for the conjecture that stable germs exist for arbitrary quasi-split groups in all characteristics."],"supporting_citations":[{"why":"Supplies Shalika's germ expansion, the classical expansion whose failure at p=2 motivates the endoscopic substitute, and the theorem used for GL(2) in Lemma 7.1.","marker":"[Sh]"},{"why":"Gives the explicit endoscopic transfer identity for SL(2) in characteristic zero, which the paper asserts remains valid in all characteristics and uses as Theorem 6.3.","marker":"[LL, Lemma 2.1]"},{"why":"The author's lecture notes provide the detailed proof of the transfer and the germ expansion for SL(2) in arbitrary characteristics, supporting Theorems 6.3 and 7.3.","marker":"[L]"},{"why":"The explicit computation borrowed in the proof of the transfer identity supplies the method for Theorem 6.3.","marker":"[JL, lemma 7.3.2]"}],"fun_headline_variants":["Endoscopic expansion rescues SL(2) germ expansion at p=2","SL(2) germ expansion works for all characteristics via endoscopy","New endoscopic germ expansion for SL(2) covers p=2","Shalika's expansion fails at p=2; endoscopic variant succeeds","SL(2) orbital integrals get germ expansion in characteristic 2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result rests on two cited existence statements: that the endoscopic transfer f^E exists in characteristic 2 with the required identities, and that Shalika's germ expansion applies to GL(2) in every characteristic; neither is proved in this paper.","fun_headline_variants_meta":{"raw":{"variants":["Endoscopic expansion rescues SL(2) germ expansion at p=2","SL(2) germ expansion works for all characteristics via endoscopy","New endoscopic germ expansion for SL(2) covers p=2","Shalika's expansion fails at p=2; endoscopic variant succeeds","SL(2) orbital integrals get germ expansion in characteristic 2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000247,"raw_usage":{"total_tokens":1523,"prompt_tokens":906,"completion_tokens":617,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":522,"completion_tokens_details":{"reasoning_tokens":523}},"tokens_in":522,"tokens_out":617,"duration_ms":6270,"temperature":1.0,"reasoning_tokens":523,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:35:31.814483+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take F = F_2((X)), G = SL(2), and a test function f supported close to the identity; compute O_G(t_n,f) for a sequence of regular elliptic t_n tending to z and compare with the right side of Theorem 7.3, constructing f^E by the explicit formula cited from [LL, Lemma 2.1]. If the difference does not tend to zero for some f, the transfer identity fails in characteristic 2 and Theorem 7.3 is false.","supporting_citations":[],"review_version":1}