{"id":"47d62e17-e6fd-4ee9-aabf-368070094d7c","arxiv_id":"2508.15101","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every irreducible representation of a connected reductive group over a finite field is matched with a special Langlands parameter, with the fiber over each parameter given by irreducible representations of a finite component group.","lead":"This paper builds a Langlands correspondence for all connected reductive groups over finite fields, pairing irreducible representations with special Langlands parameters and describing the fibers via component groups. It also conjectures a precise link to the tame categorical local Langlands correspondence for p-adic fields.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proof of Theorem 5.3 constructs L_G but never proves it is surjective onto Φ^sp; the asserted bijection for every special parameter therefore needs an additional counting/bijection argument that is not supplied.","rationale":"Reading in good faith, the categorical construction is impressive and the map is carefully defined. The most load-bearing point is not the list of external preprints but the completeness half of the theorem: the proof shows the packets over the image have the right size, but the “for every φ” clause needs surjectivity. The equivalence of categories in Theorem 4.5 may make this fillable—indeed every twisted conjugacy class [g] supports an irreducible equivariant sheaf—but the manuscript does not spell out the bijection between the indexing data (o,c,β,g) and Φ^sp, nor does it connect the construction to Theorem 4.12. Because Theorem 4.12 carries a hidden good-prime assumption, the gap is not harmless. I therefore keep the reader's CONDITIONAL verdict: the central claim is plausible and likely repairable, but as written the theorem states more than the proof demonstrates. I partially agree with the reader: their [LY21]/[Sol25] concern is real but is about external foundations, whereas the surjectivity gap is internal to the proof of Theorem 5.3.","tokens_in":13907,"tokens_out":23832,"duration_ms":287839,"concrete_test":"Check surjectivity by an independent count: for a fixed G, verify that sum_{φ ∈ Φ_Qℓ(G)^sp} |Irr(A_φ)| = |Irr(G(k))|. This can be done by exhibiting a bijection between Φ_Qℓ(G)^sp and the union over F_*-stable special unipotent classes C in centralizers of semisimple elements of the sets M[A_{G^*}(g) ⊂ eA_{G^*}(g)] appearing in Theorem 4.12, then using Lusztig's count from Theorem 4.12. Run the check for G = GL_2, SL_2, PGL_2, and for a bad-prime case such as Sp_4 in characteristic 2; if the counts match, the gap is fillable, and if not, Theorem 5.3 is false as stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 5.3 asserts L_G : Irr_Qℓ(G(k)) → Φ_Qℓ(G)^sp is a natural map and that for every φ ∈ Φ^sp there is a natural bijection L_G^{-1}(φ) ≅ Irr(A_φ). Since A_φ is a finite group, Irr(A_φ) is nonempty, so the second assertion entails that L_G is surjective. The proof only constructs LΨ_G and then shows, for ψ that occur as LΨ_G(π), that the preimage of ψ is parametrized by Irr(A_ψ) (via (5.3)–(5.4)). No argument is given that every special Frobenius-semisimple parameter is in the image. The natural completion would compare the target with the indexing set of Theorem 4.12 (F_*-stable special unipotent classes in centralizers of semisimple elements), but Theorem 4.12 is proved only under the in-section assumption “p is a good prime” (Section 4, before Definition 4.11), while Theorem 5.3 is stated without that hypothesis. Thus for bad primes the surjectivity gap is not covered even by the earlier parametrization.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs a Langlands correspondence for connected reductive groups over finite fields. The main categorical result, Theorem 4.5, describes the category of finite-dimensional ℓ-adic representations of G(k) as a direct sum of equivariant sheaf categories associated to finite groups, building on Lusztig's theory and on endoscopic equivalences from [LY21] and [Sol25]. Theorem 5.3 then defines a natural map L_G from Irr_{Qℓ}(G(k)) to the set of special Frobenius-semisimple Weil–Deligne L-parameters, and claims that for every such parameter φ the fiber L_G^{-1}(φ) is in natural bijection with Irr_{Qℓ}(A_φ). The paper also gives a Lusztig-style parametrization of irreducible representations (Theorem 4.12) and formulates a conjecture relating the finite correspondence to the tame categorical local Langlands correspondence.","tokens_in":14099,"tokens_out":5572,"duration_ms":59333,"significance":"If the main theorem is correct, this is a substantial contribution: it provides a canonical, categorical construction of the finite Langlands correspondence for all connected reductive groups over finite fields, with L-packets exactly indexed by irreducible representations of the component group A_φ. The categorical formulation is a genuine strength, as it yields functoriality that a mere bijection would not supply, and it allows the treatment of disconnected centers with a rigidification by a Whittaker datum. The paper also extends Macdonald's GL_n result to a general setting and connects the construction with the categorical local Langlands program. The proof relies heavily on several recent preprints, but the internal logic is mostly coherent; the main missing piece is a surjectivity argument for the constructed map.","major_comments":[{"comment":"The proof of Theorem 5.3 constructs a map LΨ_G : Irr_{Qℓ}(G(k)) → Ψ_{Qℓ}(G)^sp and shows, for ψ in the image, that the fiber (LΨ_G)^{-1}(ψ) is parametrized by Irr_{Qℓ}(A_ψ), using (5.3) and (5.4). It never proves that LΨ_G is surjective onto Ψ_{Qℓ}(G)^sp. Since A_ψ is a finite group and Irr_{Qℓ}(A_ψ) is nonempty, the asserted bijection for every special φ is equivalent to surjectivity. No counting argument or comparison with the indexing set in Theorem 4.12 is supplied. This is a load-bearing gap in the proof of the main theorem.","section":"Theorem 5.3 and its proof"},{"comment":"Theorem 4.12, which is the only result in the paper that could provide a counting or parametrization argument for surjectivity, is proved under the explicit assumption that p is a good prime, made just before Definition 4.11. Theorem 5.3 is stated without this hypothesis, and its proof does not invoke Theorem 4.12. Consequently, even if a surjectivity argument were added using Theorem 4.12, it would cover only good primes. The paper should either extend the argument to all primes or state Theorem 5.3 under the good-prime assumption.","section":"Section 4, after Lemma 4.10 / Theorem 4.12"},{"comment":"The proof of Theorem 4.5, and hence of the main correspondence, depends on [LY21, 12.7 Corollary] and its Whittaker-datum rigidification ([LY21, 5.11]), as well as on [Sol25, (3.3),(3.4)]. These are very recent preprints, and the manuscript does not indicate which parts of these inputs are already published and which are still under review. This is not an internal inconsistency, but it is a verification risk: if any of these equivalences fails to be natural under the actions used in equation (4.2), the direct-sum decomposition of Rep(G(k)) and the construction of L_G would lack a foundation. The authors should state the dependence explicitly and, where possible, point to published versions or provide more detail.","section":"Theorem 4.5, proof"}],"minor_comments":[{"comment":"There are numerous typographical errors: 'parametrizatin', 'run thorough', 'maxmal', 'generaic', 'Joradan decomposition', 'Frobenis', 'automorphsim', 'minimum positive integern'. These should be corrected in a revision.","section":"Throughout"},{"comment":"Theorem 5.3 is stated for 'a reductive algebraic group G over k', but Section 3 defines Langlands parameters only for connected reductive groups, and the Whittaker datum in Section 2 is defined for connected quasi-split groups. Please clarify whether G is assumed connected, or explain how the definitions extend to the non-connected case.","section":"Section 5, opening"},{"comment":"The first line of the proof, 'It suffices to show Z_AH(uc)(ghβ ˙wβσq)/Z_Gc(gτβ) ≅ Ω_{c,β}', is slightly imprecise: what is needed is a natural isomorphism of the relevant groups, not merely an isomorphism of quotients. The displayed formula could be misread as defining the quotient. Please rephrase.","section":"Lemma 4.10"},{"comment":"The notation A(φ0) is used both for the quotient π0(Z_{Ĝ}(φ0)/Z(Ĝ)) and for its further quotient by the kernel from the canonical quotient A_{Z_{Ĝ}(φ(I_k))^∘}(φ(G_a)). This overload is potentially confusing; a different symbol for the second quotient would improve readability.","section":"Section 3, Definition 3.3"}],"recommendation":"major_revision","confidential_remarks":"The main theorem as stated is not fully proved because the proof of Theorem 5.3 omits a surjectivity argument for L_G. This is potentially fixable by comparing the image of the construction with the classification in Theorem 4.12 (for good primes) and by handling bad primes separately, but as written the central claim is incomplete. In addition, the paper's dependence on the author's own preprints [IV25], [BMIY24], and [Ima24], and on the very recent [LY21] and [Sol25], is heavy; the editor may wish to verify with the handling editor that these preprints are publicly accessible and sufficiently established for the claimed result."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the short version: this is a serious paper that assembles a categorical finite Langlands correspondence for connected reductive groups over finite fields, including disconnected centers. The genuinely new content is Theorem 4.5, a canonical equivalence between Rep(G(k)) and a direct sum of equivariant sheaf categories on finite groups, and the use of that equivalence to define a natural L-map with A_phi packets. The bijection at the isomorphism-class level is mostly a repackaging of Lusztig, which the author honestly credits. The paper deserves a careful referee, but the main theorem has a real proof gap.\n\nWhat the paper does well: the categorical equivalence is put together carefully from Lusztig's categorical center, Bezrukavnikov-Finkelberg-Ostrik, Ostrik, and the corrected Lusztig-Yun endoscopy paper. Extending to disconnected groups via Solleveld is a genuine addition. The author is scrupulous in citing [LY21] rather than the flawed [LY20]. The argument is compressed but I found no internal contradiction.\n\nThe load-bearing gap, as a stress-test note correctly observed, is in the proof of Theorem 5.3. The proof constructs LΨ_G on Irr and then shows that for any ψ in the image of LΨ_G, the fiber is naturally Irr(A_ψ). But it never proves that every special Frobenius-semisimple parameter appears in the image. The theorem states a bijection for all φ ∈ Φ(G)^sp; without surjectivity, what is actually proved is only a fiber description over the image. A natural fix would be a counting argument comparing the image with the indexing set of Theorem 4.12, but that theorem is proved only under a good-prime assumption (introduced before Definition 4.11), while Theorem 5.3 is stated unconditionally. So even that route would not cover bad primes. This is not a fatal flaw—the statement is likely true—but it is a missing argument in a central claim.\n\nA minor concern: the proof leans on recent preprints ([LY21] v3 and [Sol25]), so correctness is conditional on those results. That is acceptable in this field, but a referee should check that the recorded versions match the usage.\n\nThis paper is for specialists in finite reductive groups and categorical Langlands. It has a substantial new categorical result and a plausible main theorem, but the surjectivity issue needs to be addressed before the correspondence can be considered fully proved. I'd recommend sending it to peer review with a request to supply the missing surjectivity argument and to clarify the prime hypotheses in the main theorem. A serious referee will find the paper worth engaging with.","headline":"The paper builds a categorical finite Langlands correspondence, but the proof of Theorem 5.3 never establishes surjectivity of the L-map, so the main bijection is not fully proved as written.","tokens_in":14646,"tokens_out":6810,"would_cite":true,"duration_ms":75023,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11F70","20C33","22E50"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper constructs the finite Langlands correspondence: every irreducible ℓ-adic representation of a connected reductive group over a finite field is attached to a special Frobenius-semisimple Langlands parameter, and the preimage of eac","keywords":["finite Langlands correspondence","Langlands parameters over finite fields","reductive groups over finite fields","categorical equivalence","unipotent representations","equivariant sheaves","Lusztig series","categorical local Langlands"],"falsifier":"Take a small reductive group over F_q with disconnected center, such as SL_2 or a quotient with adjoint center, enumerate the special Frobenius-semisimple Langlands parameters φ and the irreducible ℓ-adic representations of G(k), and check that the fiber over every φ has size |Irr(A_φ)|. A single mismatch would falsify Theorem 5.3; a more structural check is to compute the block category on the right side of Theorem 4.5 for such a group and see whether its Grothendieck group matches the representation ring.","tokens_in":13708,"feed_emoji":"🧮","tokens_out":10888,"duration_ms":115264,"temperature":0.7,"pith_summary":"The paper's goal is to construct the Langlands correspondence for every connected reductive group over a finite field, not just for the general linear group. It assigns to each irreducible ℓ-adic representation of G(k) a special Frobenius-semisimple Langlands parameter, and it shows that the set of representations over a fixed parameter is naturally bijective to the irreducible representations of an associated finite component group A_φ. The proof works by first establishing a stronger categorical statement: the whole category of representations of G(k) decomposes into blocks of equivariant sheaves on finite groups, one block for each semisimple and unipotent parameter. This categorical rigidity makes the parametrization canonical and extends it to groups with disconnected center. If correct, the result gives a uniform finite-field analogue of the local Langlands correspondence and a finite shadow of the categorical local correspondence.","feed_headline":"Every irreducible finite-field group rep gets a Langlands parameter","feed_subtitle":"New theorem attaches each irreducible representation to a special parameter and indexes each preimage by a finite group.","key_machinery":"The load-bearing mechanism is the categorical equivalence of Theorem 4.5, which decomposes the entire representation category of G(k) into blocks indexed by semisimple and unipotent parameters. Each block is the category of G_c-equivariant sheaves on a finite group G_c with a τ_β-twisted conjugation action and an additional Ω_{c,β}-equivariant structure. From this block category one reads off both the Langlands parameter φ and the finite component group A_φ whose irreducible representations index the fiber over φ. A Whittaker datum is fixed to make the equivalence and the resulting map natural.","core_discovery":"The central discovery is Theorem 5.3: after fixing a Whittaker datum, there is a natural map L_G from the irreducible ℓ-adic representations of G(k) to the set of equivalence classes of special Frobenius-semisimple Langlands parameters of Weil–Deligne type, and for each parameter φ the fiber L_G^{-1}(φ) is naturally in bijection with the irreducible representations of A_φ, the finite component group attached to φ. The correspondence is built from Theorem 4.5, a categorical equivalence identifying Rep(G(k)) with a direct sum of categories of equivariant sheaves on finite groups G_c with twisted conjugation. Unipotent representations of connected groups are described first (Proposition 4.1) as","pith_inferences":["The paper leaves implicit that the naturality of the map L_G should make it compatible with central character twisting and with base change or restriction of scalars; verifying such functoriality is a natural next step.","Because the proof is categorical, the correspondence should come with functoriality under morphisms of reductive groups once the underlying endoscopy equivalences are known to be functorial; the paper does not prove such functoriality.","A concrete testable extension is to compute the block categories in Theorem 4.5 for a small disconnected-center group, such as a non-split orthogonal group over F_q, and compare the resulting fibers with the character table of G(k), which would exercise the Ω_{c,β}-equivariant structure rather than just the underlying set bijection."],"forward_implications":["Every irreducible ℓ-adic representation of a connected reductive group over a finite field is assigned a well-defined special Frobenius-semisimple Langlands parameter.","The fiber over a fixed parameter φ is exactly the set of irreducible representations of the finite component group A_φ, giving a clean description of L-packets.","The categorical decomposition rigidifies the parametrization, so groups with disconnected center receive a canonical Lusztig-style parametrization rather than a merely set-theoretic one.","The general theorem encompasses the earlier GL_n and SL_n finite Langlands correspondences as special cases.","Section 6 conjectures that the finite correspondence controls the tame categorical local Langlands correspondence: the coherent sheaf attached to a representation should be supported on a specified closed set and restrict to a vector bundle built from the representation of A_φ."],"supporting_citations":[{"why":"Supplies the definition of Langlands parameters for reductive groups over finite fields and the canonical quotient group A_φ used in the correspondence.","marker":"[IV25]"},{"why":"The endoscopic categorical equivalence that identifies the relevant part of Rep(G(k)) with a sum of unipotent representation categories; the paper explicitly uses the corrected version [LY21].","marker":"[LY21, 12.7 Corollary]"},{"why":"Rigidifies the categorical equivalence by a fixed Whittaker datum, which is needed for naturality of the correspondence.","marker":"[LY21, 5.11]"},{"why":"Endoscopy for disconnected reductive groups over finite fields, used to handle the action of π_0(G) in the decomposition.","marker":"[Sol25, (3.3),(3.4)]"},{"why":"Provides the decomposition of Rep(G(k)) into Lusztig series indexed by semisimple parameters.","marker":"[DM20, Proposition 11.3.2]"},{"why":"Identifies the category of unipotent representations belonging to a two-sided cell with a relative categorical center.","marker":"[Lus15, 6.3 (a)]"},{"why":"Computes the relative categorical center as a category of twisted equivariant sheaves on the finite group G_c.","marker":"[BV12, Theorem 5.12, Theorem 6.5]"},{"why":"Gives the bijection between SL_2-type and Weil–Deligne Langlands parameters used to convert the constructed map into the final correspondence.","marker":"[BMIY24, Theorem 6.16]"}],"fun_headline_variants":["Finite field group reps linked to Langlands parameters","New theorem: finite Langlands correspondence is real","Every finite-field rep gets a unique Langlands parameter","Langlands correspondence over finite fields proved","Categorical Langlands and finite fields unify"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The whole construction depends on a prior categorical endoscopy equivalence behaving naturally and functorially when a finite component group acts on the decomposition; if that compatibility fails, the direct-sum decomposition at the center of the argument does not hold.","fun_headline_variants_meta":{"raw":{"variants":["Finite field group reps linked to Langlands parameters","New theorem: finite Langlands correspondence is real","Every finite-field rep gets a unique Langlands parameter","Langlands correspondence over finite fields proved","Categorical Langlands and finite fields unify"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000168,"raw_usage":{"total_tokens":992,"prompt_tokens":536,"completion_tokens":456,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":280,"completion_tokens_details":{"reasoning_tokens":385}},"tokens_in":280,"tokens_out":456,"duration_ms":5542,"temperature":1.0,"reasoning_tokens":385,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T18:07:00.838392+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a small reductive group over F_q with disconnected center, such as SL_2 or a quotient with adjoint center, enumerate the special Frobenius-semisimple Langlands parameters φ and the irreducible ℓ-adic representations of G(k), and check that the fiber over every φ has size |Irr(A_φ)|. A single mismatch would falsify Theorem 5.3; a more structural check is to compute the block category on the right side of Theorem 4.5 for such a group and see whether its Grothendieck group matches the representation ring.","supporting_citations":[],"review_version":1}