{"id":"c07dbace-2c29-438f-a81e-24f56d559eef","arxiv_id":"2507.16304","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Finite-field reductive group representation series with fixed semisimple parameter are canonically equivalent to unipotent representations of endoscopic groups with equivariant structure, including for disconnected groups.","lead":"Representations of reductive groups over finite fields are sorted into series by a semisimple parameter; this paper turns that sorting into an exact, canonical equivalence with unipotent representations of smaller endoscopic groups. The same equivalence is proved for disconnected reductive groups, the generality needed for many applications to p-adic groups.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The disconnected-group theorems (3.4, 3.5) require that the transferred π0(G)-action and the Ω∘_L-action combine into a coherent Ω_L-action; this coherence is asserted but not proved, and without it the 'same information' step in Eq. (3.7) fails.","rationale":"The reader's weakest-assumption analysis focused on the black-box dependence on [LuYu2, Cor. 12.7]. I agree that this is a genuine external dependency, but I do not regard it as the most load-bearing internal concern. Even with Corollary 12.7 fully granted, Theorem 3.4 and 3.5 make a new claim about disconnected groups: the transported π0(G)-action and the Ω∘_L-action must combine into a coherent Ω_L-action. The proof's sentence that the two equivariant structures 'contain precisely the same information' is precisely the place where the disconnected case goes beyond the connected case, and it is not supported by a coherence computation. This is a proof gap rather than a demonstrated falsehood, so the appropriate verdict is CONDITIONAL: the paper should add a coherence lemma, or a reference proving it, before the canonical equivalence for disconnected groups is accepted as fully established. The proposed O_4 example gives a concrete, finite test: if the 2-cocycle of the combined action is a coboundary, the theorem survives; if not, the central claim would need modification. Because the paper is otherwise careful and the statement is plausible, I would not reject it, but I would not accept it as-is without the missing coherence verification.","tokens_in":12134,"tokens_out":11827,"duration_ms":134592,"concrete_test":"Take G=O_4 over F_q, so G∘=SO_4 and π0(G)=C2, and choose a rank-one character sheaf L with W∘_L≅C2 (e.g. a semisimple parameter whose centralizer has two A1 factors), so that Ω_L∘≅C2 and the component group swaps the two factors. Compute, using the formulas after Eq. (3.6) and Eq. (2.2), the 2-cocycle of the transported π0(G)-action on \\bigoplus_{\\beta\\in B^\\circ_L} Rep_1(H^{\\sigma\\beta\\epsilon})^{\\Omega_L^\\circ} for a generating pair (ω,γ). If the cocycle is not a coboundary, the combined Ω_L-equivariant category is not well-defined and Theorem 3.4 fails in this case; if it is a coboundary, the coherence step is verified in a nontrivial case.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Granting [LuYu2, Cor. 12.7], the genuinely new step for disconnected groups is the passage in the proof of Theorem 3.4 from Eq. (3.7), a π0(G)W∘_L-equivariant structure on the Ω∘_L-equivariant category, to the claimed Ω_L-equivariant structure on the same direct sum. The text says these carry 'precisely the same information' because π0(G)W∘_L ≅ Ω_L/Ω∘_L, but this is only valid if the two group actions combine into a genuine 2-action satisfying coherence conditions such as the pentagon axiom. The paper defines the Ω_L-action only 'up to canonical isomorphisms' and does not verify the required cocycle or associativity coherence for the transferred action. A set-theoretic action modulo inner automorphisms does not automatically yield a well-defined category of equivariant objects; a nontrivial 2-cocycle in the relevant cohomology group would make the two equivariant categories inequivalent and would break the canonical equivalence of Theorem 3.4. The same unverified coherence step is used verbatim in Theorem 3.5. This is not an objection to the cited black box [LuYu2], but to the internal reduction that constitutes the paper's extension to disconnected groups.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves canonical categorical analogues of Lusztig's Jordan decomposition for finite groups of Lie type. For a connected reductive group G over a finite field, Theorem 2.1 assembles the cell-wise endoscopy equivalences of Lusztig and Yun into a canonical equivalence between the geometric series Rep_{WL}(G) and a direct sum of unipotent representation categories of endoscopic forms enriched with Ω_L-equivariant structures. Theorem 2.4 uses a regular embedding to obtain a canonical equivalence for rational series, adding canonicity to Lusztig's Jordan decomposition. The main new results are Theorems 3.4 and 3.5, which extend these categorical equivalences to smooth group schemes G with reductive neutral component G^∘, allowing π_0(G) to be infinite. The proof in the disconnected case proceeds by restricting to G^∘, applying the connected results, and then combining a π_0(G)W^∘_L-action with an Ω^∘_L-action into an Ω_L-action. The paper also provides a parametrization of rational series by pairs (L, β) and discusses applications to depth-zero supercuspidal representations of p-adic groups.","tokens_in":1658,"tokens_out":2277,"duration_ms":146966,"significance":"If the disconnected-group theorems are correct, the paper delivers a canonical, functorial Jordan decomposition for representations of finite groups of Lie type, including disconnected reductive groups, which is a natural and useful extension of the bijections of Lusztig and the categorical results of Lusztig and Yun. The connected case is a clean and useful repackaging of [LuYu2], and the parametrization of rational series in Lemma 2.3 is a valuable contribution. The promised applications to depth-zero representations of p-adic groups give the results clear potential impact. However, the genuinely new step for disconnected groups — the coherence of the combined Ω_L-action — is asserted rather than proved, so the significance of the main theorems is currently conditional on a gap being filled.","major_comments":[{"comment":"The proof of Theorem 3.4 contains a load-bearing gap. After establishing the chain of equivalences Rep_{WL}(G_ϵ) ≅ Rep_{W^∘L}(G^∘_ϵ)^{π_0(G)W^∘_L} ≅ ((⊕_{β∈B^∘_L} Rep_1(H^{σ_β ϵ}))^{Ω^∘_L})^{π_0(G)W^∘_L}, the text asserts that 'a π_0(G)W^∘_L-equivariant structure on top of a Ω^∘_L-equivariant structure contains precisely the same information as an Ω_L-equivariant structure' because π_0(G)W^∘_L ≅ Ω_L/Ω^∘_L (3.5). This is not automatic for weak actions on categories: the actions are defined only up to canonical isomorphisms, and one must prove that the π_0(G)W^∘_L-action on the Ω^∘_L-equivariant category and the Ω^∘_L-action combine into a coherent Ω_L-action satisfying the relevant cocycle and associativity (pentagon) conditions. A nontrivial 2-cocycle would make the iterated equivariant category inequivalent to the Ω_L-equivariant category. The proof does not construct the comparison functor or verify coherence, so Theorem 3.4 is not established as written. Since Theorem 3.5 invokes the same 'combine' step with the same justification, it inherits the same gap.","section":"§3, proof of Theorem 3.4, Eq. (3.7)"},{"comment":"The definition of the Ω_L-action on ⊕_{β∈B^∘_L} Rep_1(H^{σ_β ϵ}) is not sufficiently precise to support the equivalence claimed in Theorem 3.4. The text says that Ω_L acts 'by the same constructions as in the case of a connected G' and that the action is 'well-defined up to canonical isomorphisms,' but it does not prove that this action is compatible with the Ω^∘_L-action and the transferred π_0(G)W^∘_L-action. In particular, one needs to know whether the Ω_L-action restricts to the Ω^∘_L-action on each summand, whether the transferred π_0(G)W^∘_L-action agrees with the conjugation action induced by Ω_L on the category of Ω^∘_L-equivariant objects, and whether the resulting combined action satisfies the coherence conditions needed for a well-defined category of equivariant objects. These compatibilities are the substance of the passage from (3.7) to the theorem's conclusion, and they cannot be assumed without proof.","section":"§3, definition of the Ω_L-action before Theorem 3.4"}],"minor_comments":[{"comment":"The formula for the action of Ω_L on B_L is hard to parse: 'Ad ϵ(ω)w = ωwϵ(ω)−1' should be typeset with explicit parentheses and a clarification of whether ϵ(ω) denotes the image of the representative of ω under the Frobenius action, and where the inverse is taken.","section":"§2, Eq. (2.1)"},{"comment":"Theorem 2.1 invokes [LuYu2, Corollary 12.7] as a black box without stating its hypotheses. Please state the precise conditions on the rank-one character sheaf L under which the cell-wise equivalences of [LuYu2, Corollary 12.7] hold, and confirm that those conditions are satisfied for the L considered in Theorem 2.1.","section":"§2, Theorem 2.1"},{"comment":"In Theorem 3.4 the hypothesis is written as 'W^∘L is ϵ-stable,' but since W^∘ is a group, what is presumably meant is that the W^∘-orbit of L is ϵ-stable. The wording should be corrected for clarity.","section":"§3, Theorem 3.4 statement"},{"comment":"The definition of Rep_{W(L,β)}(G_ϵ) is followed by the remark that it 'actually corresponds to a set of semisimple elements in G^{∘∨}_ϵ larger than the rational conjugacy class of s_β.' This is vague; please give a precise description of the parameter set, perhaps using the action of π_0(G) on the pairs (L, β) as in (3.8).","section":"§3, rational series for disconnected groups"}],"recommendation":"major_revision","confidential_remarks":"The paper is well written and the connected case is a solid repackaging of Lusztig-Yun. The main concern is whether the disconnected extension is actually proved: the coherence step in Theorems 3.4 and 3.5 is asserted rather than demonstrated. If the author can supply a rigorous proof of the equivalence between the iterated equivariant category and the Ω_L-equivariant category (or a reference establishing it), the paper would be suitable for publication. The referee report should focus on this gap and request that the proof be made complete."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this paper does something useful and labels its debts honestly. The connected-group part is a cell-free reformulation of [LuYu2, Cor. 12.7]; the genuinely new content is the canonical rational-series statement (Theorem 2.4) and the extension to disconnected reductive groups (Theorems 3.4 and 3.5). The exposition is clear, the citation pattern is clean, and the paper is open about the fact that the geometric connected case largely restates Lusztig-Yun. The reader's circularity score of 1 is right: there is no parameter-fitting or conclusion-normalizing here.\n\nThe soft spot is exactly where the stress-test note lands. In the proof of Theorems 3.4 and 3.5, Eq. (3.7) passes from an Omega_L^0-equivariant structure plus a pi_0(G)W_L^0-equivariant structure to a single Omega_L-equivariant structure. The text says these carry 'precisely the same information' because pi_0(G)W_L^0 is isomorphic to Omega_L/Omega_L^0. As a statement about groups that is true, but as a statement about categories it is not automatic. Combining the two actions requires that the chosen lifts of Omega_L act coherently on the direct sum of unipotent categories. The paper defines the Omega_L-action only 'up to canonical isomorphisms' and does not verify the required associativity or pentagon-type coherence. Without that check, the right side of (3.7) could define a different equivariant category, and the main theorems would not follow. I suspect the coherence can be extracted from the explicit morphisms in Section 2, but the paper needs to show it rather than assert it.\n\nTwo smaller issues. The proof of Theorem 2.4 says 'from [LuYu2, Section 12.10] one can see' how the restriction functor behaves; that is a handwave in the central connected statement. And the whole edifice rests on the black box [LuYu2, Cor. 12.7]; if that result carries hypotheses not satisfied by Condition 3.1 groups, the conclusions would not follow. I do not think either of these is fatal; they are exactly what a referee should ask to be pinned down.\n\nThe audience is representation theorists working on finite groups of Lie type and on depth-zero p-adic supercuspidals. This paper deserves a serious referee. I would not desk-reject; I would send it out with the explicit request to prove the coherence claim in Theorems 3.4 and 3.5. I personally would not cite it in my own work until that step is settled.","headline":"A transparent and mostly well-executed package: the connected part is a clean repackaging of Lusztig-Yun, the disconnected extension is the real novelty but depends on a coherence step that is asserted rather than proved.","tokens_in":12996,"tokens_out":4356,"would_cite":false,"duration_ms":51224,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20C33","20G40"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes canonical equivalences of categories that make the Jordan decomposition for finite reductive groups canonical and extend it to disconnected reductive groups.","keywords":["Jordan decomposition","Lusztig series","endoscopy","disconnected reductive groups","character sheaves","unipotent representations","canonical equivalences","finite groups of Lie type"],"falsifier":"Compute, for a small disconnected group satisfying Condition 3.1—for example $G = O_2(\\mathbb{F}_q)$ or the normalizer of a maximal torus in $\\mathrm{PGL}_2$—the category $\\mathrm{Rep}_s(G)$ and the claimed endoscopic category $\\mathrm{Rep}_1(H^\\epsilon)^{\\Omega_L^\\epsilon}$ for every semisimple parameter $s$, and compare the number and internal structure of irreducible objects; any mismatch would disprove Theorem 3.5, while verifying the hypotheses of [LuYu2, Corollary 12.7] for those groups would test the load-bearing input.","tokens_in":11915,"feed_emoji":"","tokens_out":9175,"duration_ms":86672,"temperature":0.7,"pith_summary":"This paper proves that the Jordan decomposition for representations of finite reductive groups can be made canonical: rather than a bijection between irreducible representations with a fixed semisimple parameter and unipotent representations, there is an equivalence of whole categories. The main theorems extend this from connected groups to smooth group schemes whose neutral component is reductive, allowing the group of connected components to be infinite. If the theorems are correct, every irreducible representation of such a group is canonically described by a unipotent representation of a smaller endoscopic group together with an equivariant structure for a component group. This upgrades a classical classification into a functorial statement, and the author notes that the canonical form is what applications to depth-zero supercuspidal representations of $p$-adic groups require.","feed_headline":"Jordan decomposition made canonical for disconnected groups","feed_subtitle":"Fixed-parameter representations match unipotent representations of endoscopic groups, up to canonical equivalence.","key_machinery":"The carrying object is the cell-by-cell categorical endoscopy equivalence from [LuYu2, Corollary 12.7], which identifies each cell subcategory of a geometric series with a direct sum of unipotent representation categories of forms of an endoscopic group, equipped with equivariant structures for stabilizer groups. The paper's work is to assemble these cell-wise equivalences into whole-series equivalences, to pass from geometric to rational series by parametrizing rational conjugacy classes through the quotient $B^\\circ_L/\\mathrm{Ad}^\\epsilon(\\Omega_L)$ using Lang's theorem and Steinberg's description of component groups, and to handle disconnectedness through the equivalence $\\mathrm{Rep}(G^\\epsilon) \\cong \\mathrm{Rep}(G^{\\circ,\\epsilon})^{\\pi_0(G)}$ of Lemma 3.2, which trades a representation of the disconnected group for a representation of its neutral component with a $\\pi_0(G)$-equivariant structure. The canonical nature of the final equivalences depends on all these identifications being compatible up to canonical isomorphisms.","core_discovery":"On the paper's own terms, the central discovery is the existence of canonical equivalences of categories between representation series of (possibly disconnected) reductive groups over finite fields and unipotent representation categories of endoscopic groups. For a connected reductive group $G^\\epsilon$, Theorem 2.4 states that for every stable pair $(L,T)$ corresponding to a semisimple element $s$ in the dual group, $\\mathrm{Rep}_s(G^\\epsilon)$ is canonically equivalent to $\\mathrm{Rep}_1(H^\\epsilon)^{\\Omega_L^\\epsilon}$, where $H$ is the endoscopic group determined by $L$ and the superscript denotes enrichment by $\\Omega_L^\\epsilon$-equivariant structures. For a smooth group scheme with reductive neutral component satisfying Condition 3.1, Theorem 3.4 gives the analogous equivalence for geometric series, $\\mathrm{Rep}_{W(G,T)L}(G) \\cong (\\bigoplus_{\\beta \\in B^\\circ_L} \\mathrm{Rep}_1(H^{\\sigma_\\beta \\epsilon}))^{\\Omega_L}$, and Theorem 3.5 gives the rational-series version $\\mathrm{Rep}_s(G) \\cong \\mathrm{Rep}_1(H^\\epsilon)^{\\Omega_L^\\epsilon}$. These results make Lusztig's Jordan decomposition, previously a bijection, into a canonical equivalence of categories and extend it to the disconnected setting.","pith_inferences":["The categorical, rather than bijective, nature suggests that natural operations on representation series, such as Deligne-Lusztig induction or restriction, can be made compatible with the Jordan decomposition, something a parametrization cannot express.","The equivariant-structure viewpoint may give a uniform language for representation rings of disconnected groups with nontrivial component-group actions, possibly connecting to cohomological induction for such groups.","A concrete numerical consequence of the main theorem would be identities between sizes of Lusztig series and sums of sizes of unipotent series; checking these for small $q$ would be an easy consistency test.","Because the construction inherits its canonicity from the cell-by-cell input, making that input more explicit (or replacing it) would directly control how constructive the whole equivalence is."],"forward_implications":["For any fixed semisimple parameter, the full category of representations is canonically equivalent to a unipotent category of an endoscopic group, so categorical invariants such as characters, cohomology, and natural transformations can be transferred between series.","The geometric series of a disconnected group splits into rational series parametrized by $B^\\circ_L/\\mathrm{Ad}^\\epsilon(\\Omega_L)$, one for each rational form $H^{\\sigma_\\beta \\epsilon}$ of the endoscopic group.","Representations of groups with reductive neutral component, including infinite component groups, are reduced to unipotent representations of connected endoscopic groups plus equivariant structures.","The canonical form of the equivalence is what makes the depth-zero supercuspidal application viable, since the construction is now functorial rather than a mere parametrization."],"supporting_citations":[{"why":"Supplies the cell-by-cell canonical equivalences between cell subcategories and unipotent categories from which all main theorems are assembled.","marker":"[LuYu2, Corollary 12.7]"},{"why":"Develops the geometric endoscopy framework for Hecke categories and character sheaves that the paper translates into representation-theoretic terms.","marker":"[LuYu1]"},{"why":"Introduces Deligne-Lusztig characters and the partition of irreducible representations into geometric series that anchors the definition of the series studied here.","marker":"[DeLu]"},{"why":"Proves the Jordan decomposition bijection for representations of reductive groups with disconnected centre, the statement this paper upgrades to a canonical equivalence.","marker":"[Lus2]"},{"why":"Describes component groups of centralizers of semisimple elements, used to identify the equivariance groups $\\Omega_L$ with $\\pi_0(Z_{G^\\vee}(s))$.","marker":"[Ste]"},{"why":"Provides Lang's theorem, regular embeddings, and the duality correspondence used to pass between the character-sheaf data $(L,T)$ and the semisimple parameter $s$.","marker":"[GeMa]"}],"fun_headline_variants":["Canonical Jordan decomposition for disconnected groups","Disconnected groups get canonical representation equivalence","Jordan decomposition now a canonical equivalence","Endoscopic groups offer canonical Jordan decomposition","Fixed-parameter representations match endoscopic unipotents"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper rests on the black-box cell-by-cell canonical endoscopy equivalence of [LuYu2, Corollary 12.7], with the second fragile point being that the compatibility of the $\\Omega_L$ and $\\pi_0(G)$ actions is only described 'up to canonical isomorphisms'.","fun_headline_variants_meta":{"raw":{"variants":["Canonical Jordan decomposition for disconnected groups","Disconnected groups get canonical representation equivalence","Jordan decomposition now a canonical equivalence","Endoscopic groups offer canonical Jordan decomposition","Fixed-parameter representations match endoscopic unipotents"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000258,"raw_usage":{"total_tokens":1574,"prompt_tokens":932,"completion_tokens":642,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":548,"completion_tokens_details":{"reasoning_tokens":579}},"tokens_in":548,"tokens_out":642,"duration_ms":6584,"temperature":1.0,"reasoning_tokens":579,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:13:45.079958+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a small disconnected group satisfying Condition 3.1—for example $G = O_2(\\mathbb{F}_q)$ or the normalizer of a maximal torus in $\\mathrm{PGL}_2$—the category $\\mathrm{Rep}_s(G)$ and the claimed endoscopic category $\\mathrm{Rep}_1(H^\\epsilon)^{\\Omega_L^\\epsilon}$ for every semisimple parameter $s$, and compare the number and internal structure of irreducible objects; any mismatch would disprove Theorem 3.5, while verifying the hypotheses of [LuYu2, Corollary 12.7] for those groups would test the load-bearing input.","supporting_citations":[],"review_version":1}