{"id":"5cf34128-d799-46a7-8bf3-684ad14e2816","arxiv_id":"2504.17225","paper_version":1,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For simple adjoint groups over non-Archimedean local fields, every depth-zero supercuspidal representation is shown to come from an enhanced L-parameter, with bijectivity in many cases.","lead":"This paper constructs a map matching a class of Galois-type data (enhanced L-parameters) to the building-block representations of p-adic simple adjoint groups, and shows the matching is one-to-one for many group types. A generalist reader might care because such a 'local Langlands correspondence' for these building blocks is a central step in a major program in number theory and representation theory.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"E6/E7 bijectivity depends on an unstated theorem from an unpublished preprint and an unverified diagram case list, so the central bijectivity claim is not checkable as written.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing point: the E6/E7 bijectivity proof depends on an unpublished theorem and an asserted diagrammatic case analysis. I read the construction of LLC in Sections 3 and 4 as a plausible chain: the use of Lusztig's Jordan decomposition and the FOS20 correspondence for unipotent supercuspidal representations gives a natural route to a surjective map, and the paper is explicit that the map depends on non-canonical choices. I found no fatal flaw in the surjectivity argument itself; the well-definedness issue is acknowledged and the existence claim is weaker than canonicity. The formal-degree theorem is conditional on bijectivity and follows by comparing formal degrees and adjoint gamma factors once the correspondence is bijective. The unresolved dependency on [Kal21, Theorem 2.7.7] is therefore the most serious obstruction to the central claim as stated. Because the reader already assigned CONDITIONAL with low confidence, my stress-test does not change that verdict; it confirms that the main outstanding question is whether the E6/E7 counting argument can be substantiated without appeal to an inaccessible or possibly inapplicable theorem.","tokens_in":30766,"tokens_out":23663,"duration_ms":250280,"concrete_test":"Independently enumerate all Frob-stable subsets Delta_F of the extended E6 and E7 affine Dynkin diagrams for which the parahoric quotient has disconnected center and Omega_{G,F} is nontrivial, compute the root system of the resulting parahoric quotient for each case, and verify that the list in Section 6 is complete. In parallel, extract the exact statement of [Kal21, Theorem 2.7.7] and confirm that it is a finite-group statement applicable to G(k) = ~G(k)_s and S'(k); if it is instead a p-adic statement, replace the counting argument by a direct derivation of the displayed equality for each listed case.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The bijectivity assertion for types E6 and E7 in Theorem 1.1(1), proved in Theorem 6.1, rests on two inputs that the paper neither states nor verifies. First, the key counting step applies [Kal21, Theorem 2.7.7] to obtain a bijection Irr(~G(k)_s)_s -> Irr(N_{~G(k)_s}(S'), theta). This theorem is from an unpublished preprint, and no statement is quoted, so the reader cannot check its hypotheses or conclusion; in particular, it is not evident that a theorem from a paper on supercuspidal L-packets is a statement about finite groups of Lie type in the form needed here. If the theorem is inapplicable, the equality #Irr(~G(k)_s)_s = |~G(k)_s|/|G(k)| * #Irr(G(k))_s does not follow. Second, the reduction to products of type A_n groups relies on an asserted enumeration of possible Frob-stable subsets Delta_F of the extended E6 and E7 affine diagrams, presented only by diagrams and not by a systematic list or algorithm. The paper says 'the possible choices ... are as follows' and then 'In all of those cases, G is a product ...'; an omitted case would break the argument. These two gaps are the weakest points of the central bijectivity claim. The surjectivity part of Theorem 1.1 is not affected by this concern.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs, for a simple adjoint group G over a non-Archimedean local field F that splits over an unramified extension, a map LLC from conjugacy classes of depth-zero cuspidal enhanced L-parameters to depth-zero supercuspidal representations. The construction proceeds by associating to a depth-zero discrete L-parameter a smaller unramified reductive group H, embedding apartments of inner twists of H into those of G, and then using Lusztig's Jordan decomposition together with the unipotent supercuspidal correspondence of FOS20. The paper proves that LLC is surjective in general, verifies bijectivity in a list of cases (type A_n, E_6, E_8, F_4, G_2, inner forms of 3D_4, split C_n and E_7, and, for odd residual characteristic, type B_n and quasi-split types 2D_2n and D_(2n+1)), and proves the Hiraga–Ichino–Ikeda formal-degree formula conditionally on bijectivity.","tokens_in":30999,"tokens_out":5679,"duration_ms":53331,"significance":"If the construction is correct, the paper gives a unified extension of the DeBacker–Reeder and Feng–Opdam–Solleveld correspondences and yields new bijectivity results, especially for exceptional groups. The formal-degree theorem is a useful conditional contribution that connects the constructed parametrization to a standard expected property. The paper is also commendably explicit about the non-canonical choices in the construction, and it includes a concrete worked example. The main weaknesses are that the bijectivity claims for E_6 and E_7 rest on an unpublished theorem whose statement is not quoted, and that the asserted well-definedness of LLC on conjugacy classes is not proved in detail.","major_comments":[{"comment":"The passage beginning 'It is clear that, for ϕ′ = Ad(g)ϕ, ξ′ = Ad(g)ξ and g∈Ĝ, we have LLC[ϕ′],ξ′∘Ad(g) = LLC[ϕ],ξ for suitable choices of ~J′' asserts, rather than proves, the compatibility of the constituent maps under conjugation. Since Theorem 1.1 and Theorem 4.11 are statements about conjugacy classes in Φe(G′)0,cusp/∼, a well-defined map on the quotient requires a simultaneous choice of ~J′ and of the auxiliary embeddings for every class that is equivariant under conjugation. Remark 4.12 concedes that these choices are non-canonical, so the quotient map is not established by the text as it stands. This issue does not affect the surjectivity argument viewed on representatives, but it is load-bearing for the stated form of the theorem.","section":"§4, proof of Theorem 4.11"},{"comment":"The counting step for the E_6 and E_7 cases applies [Kal21, Theorem 2.7.7] to obtain the bijection Irr(~G(k)s)s → Irr(N~G(k)s(S′),θ), and then concludes #Irr(~G(k)s)s = (|~G(k)s|/|G(k)|)·#Irr(G(k))s. This theorem is from an unpublished preprint and its statement is not reproduced, so the reader cannot verify its hypotheses or its applicability to the finite groups of Lie type appearing here. The displayed equality is the load-bearing step for condition (B), so if [Kal21, Theorem 2.7.7] is unavailable or inapplicable, the bijectivity claims for E_6 and E_7 in Theorem 6.1 are unsupported.","section":"§6, E6/E7 bijectivity"},{"comment":"The sentence 'the possible choices ... are as follows' is supported only by diagrams, and the subsequent sentence 'In all of those cases, G is a product of reductive groups of type 1A_n' is not accompanied by a systematic enumeration or a proof that no other Frob-stable subsets ΔF can occur. An omitted case would break the reduction to products of type A_n and hence the verification of condition (B). This is a concrete and checkable gap in the bijectivity proof for E_6 and E_7.","section":"§6, E6/E7 diagram case list"}],"minor_comments":[{"comment":"The notation oscillates between Φ(LG)0,disc and Φ(G)0,disc; please make the domain of the equivalence relation 'w∼' explicit throughout.","section":"§3, Definition 3.1"},{"comment":"The sentence 'H = S is compact' should read 'H(F) is compact', since S is a torus over F and it is its group of F-points that is compact.","section":"§4, Remark 4.13"},{"comment":"The notation 'type 1E6' in the proof of Theorem 6.1 is not defined, while Theorem 1.1 simply says 'type E6'; please clarify which inner or outer forms are covered by this notation.","section":"§6, E6/E7 subsection"},{"comment":"The phrase 'Examining the extended Dynkin diagram' appears twice without an explicit statement of the resulting cases; a short table for the exceptional cases would make the verification easier to follow.","section":"§5, proof of Theorem 5.1"},{"comment":"In the sentence 'The representation ρ0 is an irreducible component of the Deligne–Lusztig character RG_S′(θ)', the notation would benefit from specifying which parabolic or which Deligne–Lusztig induction is meant.","section":"§6, Example 6.3"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the paper's strongest bijectivity claims for E_6 and E_7 depend on an unpublished preprint theorem that is not stated, and on a diagrammatic case enumeration that is not proved complete. These are local, but load-bearing, issues. The surjectivity part appears structurally sound and could be published separately if the bijectivity claims cannot be fully supported. I therefore recommend major_revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read. The paper's real contribution is a uniform surjective map LLC from depth-zero cuspidal enhanced L-parameters to depth-zero supercuspidal representations for simple adjoint groups, covering the mixed cases (neither regular nor unipotent) that DeBacker-Reeder and Feng-Opdam-Solleveld left open. The construction via Lusztig's Jordan decomposition and the auxiliary group H is natural, and the surjectivity proof in Theorem 4.11 reduces cleanly to the known unipotent correspondence plus the Jordan decomposition. I could not find a gap in that part. The verification of condition (B) for types A_n, E8, F4, G2, 3D4, split C_n and E7 (and B_n/quasi-split D_n for odd p) is lengthy but plausible; Theorem 5.1 on lifting fundamental-group elements while preserving pinning is a solid piece of root-system combinatorics. The formal-degree result, Theorem 7.2, is a direct extension of FOS20's theorem, but it is correctly derived and useful.\n\nThe soft spot is exactly where the stress-test puts it: the bijectivity proof for E6 and E7. It applies [Kal21, Theorem 2.7.7] — an unquoted theorem from an unpublished preprint (arXiv:1912.03274) — to get the key counting bijection #Irr(~G(k)_s)_s = |~G(k)_s|/|G(k)| * #Irr(G(k))_s. Since the statement and hypotheses are not given, the referee cannot check whether that theorem applies to the finite-group situation here. The reduction to products of type A_n also rests on an enumeration of possible Frob-stable subsets Delta_F of the extended E6/E7 diagrams that is presented as pictures plus the assertion \"in all of those cases...\". An omitted case would sink the argument. These issues do not affect the surjectivity or the other bijectivity cases, but they do mean the headline claim \"bijective for E6 and E7\" is not checkable as written. The non-canonical choices in LLC are acknowledged, and I do not hold that against the paper, though \"suitable choices\" in Remark 4.12 is doing some work.\n\nWho should read it? Anyone working on depth-zero local Langlands. It is a serious piece of work that deserves a serious referee. My recommendation: send to peer review, but insist that the E6/E7 argument be made self-contained — either state and prove the required finite-group theorem (or replace it with a published reference) and give a complete algorithmic list of Delta_F cases. The surjectivity part is strong enough to justify publication even if the E6/E7 bijectivity needs revision.","headline":"Useful surjectivity result for all depth-zero supercuspidals of simple adjoint groups, but the E6/E7 bijectivity claim is uncheckable as written because it depends on an unquoted theorem from an unpublished preprint and a pictured case enumeration.","tokens_in":31568,"tokens_out":3450,"would_cite":true,"duration_ms":31564,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["22E50","20C33","11S37"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper constructs a surjective parametrization of depth-zero supercuspidal representations of simple adjoint p-adic groups by depth-zero cuspidal enhanced L-parameters, and proves bijectivity and the formal-degree formula in a range of…","keywords":["depth-zero supercuspidal representations","enhanced L-parameters","local Langlands correspondence","simple adjoint groups","unipotent supercuspidal representations","Jordan decomposition for finite groups of Lie type","formal degree conjecture","parahoric subgroups"],"falsifier":"Compute, for one of the E6 or E7 parahoric quotients listed in Section 6, the number of cuspidal representations of ~G(k) lying above a single cuspidal representation of G(k); condition (B) fails exactly when this number exceeds one, and a single such example would disprove Theorem 6.1.","tokens_in":30511,"feed_emoji":"🔗","tokens_out":9785,"duration_ms":85804,"temperature":0.7,"pith_summary":"For a simple adjoint group over a non-Archimedean local field that splits over an unramified extension, the paper constructs a surjective map from depth-zero cuspidal enhanced L-parameters to depth-zero supercuspidal representations. This is a concrete piece of the local Langlands correspondence, restricted to the supercuspidal building blocks from which many other representations are induced. The map extends two known correspondences, one for regular supercuspidal representations and one for unipotent supercuspidal representations, filling in the mixed cases. In many root systems the map is shown to be bijective, and whenever it is bijective the formal degree of the representation matches the predicted adjoint gamma-factor expression.","feed_headline":"New map matches supercuspidal reps to L-parameters","feed_subtitle":"Surjective for all simple adjoint groups, bijective in many types, with the formal-degree formula proved when bijective.","key_machinery":"The bridge object is the centralizer group H[φ] obtained from a depth-zero discrete L-parameter φ by restricting φ to tame inertia, taking the resulting semisimple element, and forming its connected centralizer in the dual group; H[φ] is then realized as an unramified group whose L-group embeds into that of G. An embedding of apartments of H into apartments of G places parahoric subgroups of H inside those of G, so on reductive quotients the Jordan decomposition for finite groups of Lie type converts cuspidal representations of G(k) into unipotent cuspidal representations of a related finite group. The known bijection for unipotent supercuspidal representations transfers along this bridge to define LLC. Bijectivity is governed by condition (B): for any cuspidal representation of ~G(k), the number of irreducible components lying in a fixed Gad(k)-orbit is at most one. A separate result on fundamental-group actions and Frobenius-stable pinnings of parahoric quotients makes condition (B) checkable from extended Dynkin diagrams in the listed cases.","core_discovery":"The central claim is that every depth-zero supercuspidal representation of such a group arises from a depth-zero cuspidal enhanced L-parameter, and that in types A_n, E_6, E_8, F_4, G_2, inner forms of 3D_4, split C_n and E_7, and for odd residual characteristic type B_n and quasi-split 2D_{2n} and D_{2n+1}, the correspondence is one-to-one. The construction associates to each parameter a smaller unramified group H whose unipotent supercuspidal representations are already parametrized; using parahoric subgroups and the Jordan decomposition for finite groups of Lie type, unipotent supercuspidal representations of H become depth-zero supercuspidal representations of G. The construction involves some non-canonical choices, but for parameters with trivial SL2(C) part it is unambiguous and agrees with the known regular supercuspidal correspondence. When the map is bijective, the paper proves the formal-degree formula fdeg(π) = dim(ε)/|S_φ| · |γ(0, φ, Ad, ψ)| for depth-zero supercuspidal representations.","pith_inferences":["The reduction to condition (B) suggests that bijectivity for any remaining simple adjoint type reduces to a finite check on parahoric quotients and fundamental-group actions; the Section 6 diagrammatic case analysis could likely be automated.","The unpublished theorem used for the E6 and E7 bijectivity cases is probably replaceable: a direct proof of the needed character-extension statement would remove the preprint dependency without changing the rest of the construction.","The formal-degree proof compares volumes and gamma factors before invoking bijectivity, so the formula may hold for the surjective map whenever the relevant fibers have no multiplicity, even before full bijectivity is established.","The same apartment-embedding and Jordan-decomposition strategy could potentially be adapted to quasi-split non-adjoint groups by tracking fundamental-group homomorphisms and central characters."],"forward_implications":["Every depth-zero supercuspidal representation of a simple adjoint group splitting over an unramified extension is accounted for by some depth-zero cuspidal enhanced L-parameter.","In the listed types the parametrization is one-to-one, so the depth-zero supercuspidal representations of those groups have exactly the L-packet structure predicted by the local Langlands correspondence.","Where LLC is bijective, the formal-degree conjecture holds for all depth-zero supercuspidal representations, giving an explicit value in terms of the enhanced L-parameter.","For parameters with trivial SL2(C) part, the new map agrees with the regular supercuspidal correspondence, and it extends the unipotent supercuspidal correspondence as well."],"supporting_citations":[{"why":"Supplies the bijective correspondence for unipotent supercuspidal representations on which the whole construction rests.","marker":"[FOS20, Theorem 2]"},{"why":"Gives the formal-degree equality for unipotent supercuspidal representations that Theorem 7.2 transfers to depth zero.","marker":"[FOS20, Theorem 3]"},{"why":"Provides the regular supercuspidal correspondence that LLC extends and agrees with on parameters with trivial SL2 part.","marker":"[DR09, Theorem 4.5.3]"},{"why":"Provides the Jordan decomposition bijection between a semisimple Lusztig series and unipotent representations of a centralizer.","marker":"[Lus88, 12]"},{"why":"Gives the cuspidality preservation under Jordan decomposition, used in transferring cuspidal representations.","marker":"[GM20, Theorem 3.2.22]"},{"why":"Unpublished extension theorem used to verify condition (B) and hence bijectivity for E6 and E7.","marker":"[Kal21, Theorem 2.7.7]"},{"why":"Controls the outer automorphism action on cuspidal series for orthogonal groups, used for type Bn and quasi-split Dn.","marker":"[LS20, Proposition 7.10 (ii)]"},{"why":"Identifies root systems of parahoric quotients, grounding the apartment embedding and the definition of H.","marker":"[KP23, Theorem 8.4.10]"},{"why":"Equivariance of Jordan decomposition for connected center, used to analyze the component-group action.","marker":"[CS13, Theorem 3.1]"}],"fun_headline_variants":["Surjective map from L-parameters to supercuspidal reps","Bijectivity for many types, formal degree proven","Supercuspidal correspondence for simple adjoint groups","Depth-zero supercuspidals parametrized by L-parameters","Formal-degree conjecture holds when map is bijective"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"For the E6 and E7 bijectivity claims, the argument depends on an unpublished theorem about extending cuspidal characters together with a finite list of extended-Dynkin-diagram cases; if that theorem or the case list is wrong, the bijectivity for those types fails.","fun_headline_variants_meta":{"raw":{"variants":["Surjective map from L-parameters to supercuspidal reps","Bijectivity for many types, formal degree proven","Supercuspidal correspondence for simple adjoint groups","Depth-zero supercuspidals parametrized by L-parameters","Formal-degree conjecture holds when map is bijective"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000185,"raw_usage":{"total_tokens":1284,"prompt_tokens":867,"completion_tokens":417,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":483,"completion_tokens_details":{"reasoning_tokens":335}},"tokens_in":483,"tokens_out":417,"duration_ms":3889,"temperature":1.0,"reasoning_tokens":335,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:47:25.935673+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for one of the E6 or E7 parahoric quotients listed in Section 6, the number of cuspidal representations of ~G(k) lying above a single cuspidal representation of G(k); condition (B) fails exactly when this number exceeds one, and a single such example would disprove Theorem 6.1.","supporting_citations":[],"review_version":1}