{"id":"7a480619-047d-427e-ae86-941ce43b89b8","arxiv_id":"1908.02652","paper_version":4,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Blocks with elementary abelian defect group (C2)^5 are shown to be Morita equivalent to exactly one of 34 listed block algebras, and Harada's conjecture is verified for them.","lead":"This paper classifies all Morita equivalence classes of certain algebraic structures called blocks, when the underlying defect group is elementary abelian of order 32. It gives an explicit list of 34 representative block algebras and verifies a conjecture of Harada for all these blocks.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The classification hinges on unverified GAP/Magma computations; Proposition 3.4's OrthogonalEmbeddings assertions are load-bearing for the (G,B)-local systems and are not reproducible from the preprint.","rationale":"I read the paper as a serious classification theorem built from reduction to quasisimple blocks, passage through normal subgroups via (G,B)-local systems and crossed products, and computation of Picard groups. The overall strategy is coherent, and I found no internal contradiction that would refute Theorem 1.1 independent of the computational steps. The most fragile point is exactly the one the reader identified: Proposition 3.4 uses GAP's OrthogonalEmbeddings in a way that is not reproducible from the preprint, and the proof explicitly points to the PhD thesis for detailed computations. This matters because Proposition 3.5 and the main proof rely on those local systems to handle blocks covered by a block of a normal subgroup of 2-power index. The admitted open points in Corollary 5.3 concerning inertial quotients are limitations rather than counterexamples, since Theorem 1.1 classifies Morita equivalence classes and does not assert inertial quotient invariance, although a stronger result would be desirable. The reader's CONDITIONAL verdict already reflects the right level of confidence: accept contingent on verification of the computational assertions. My stress-test does not identify a reason to change that verdict.","tokens_in":35742,"tokens_out":5180,"duration_ms":56496,"concrete_test":"Provide the GAP script and the matrices N^0 used in Proposition 3.4 for each required subgroup Q, and rerun OrthogonalEmbeddings to confirm uniqueness of the solution with the stated number of irreducible characters (32 or 16 for E=C3×C3; 32 for E=(C7⋊C3)_1). If any case admits a second extension, check whether it is NE(Q)-stable; if not, the conclusion of Proposition 3.4 fails. As a supplementary check, reproduce the subgroup diagram of Proposition 2.8 with a GAP/Magma command that lists subgroups of odd order of GL5(2) up to conjugacy.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is Theorem 1.1, and one load-bearing step is the existence of (G,B)-local systems for blocks with inertial quotient C3×C3 or (C7⋊C3)_1. Proposition 3.4 is the only argument supplying these systems, and Proposition 3.5 converts them into Morita equivalences used in step (III) of the proof (excluding normal subgroups of index 2) and in the chain argument for ℓ_e=4 or 8. The proof of Proposition 3.4, however, delegates the decisive verification to GAP: after reducing an extension of the isometry Δ_Q^0 to an equation Y^t Y = N^0, it states that OrthogonalEmbeddings shows that in each case CE(Q)=E there is a unique solution (Q=1 or Q=R=C2 for E=C3×C3; Q∈{1,C2,(C2)^2} for E=(C7⋊C3)_1), and that uniqueness implies NE(Q)-stability. No scripts, input data, or certificates are included, and the paper explicitly refers to the PhD thesis for the detailed computations. The same issue affects Proposition 2.8, whose enumeration of subgroups of odd order of GL5(2) is asserted from a Magma computation with only a diagram as evidence. Because the main theorem's completeness and the Morita equivalence claims in Proposition 3.5 both depend on these unverified computations, a wrong or incomplete computation would change the classification. This is not an internal inconsistency, but it is an unresolved dependence on computations that cannot be independently checked from the preprint alone.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a general reduction technique for classifying p-blocks with elementary abelian defect groups and applies it to blocks with defect group (C2)^5 over a complete discrete valuation ring with residue field of characteristic two. The main result, Theorem 1.1, asserts that every such block is Morita equivalent to the principal block of one of thirty-one explicitly listed groups or to a nonprincipal block of one of three further groups, and that the elementary abelian defect group is invariant under Morita equivalence. The proof proceeds by a minimal-counterexample argument: it first establishes quasprimitivity and excludes normal subgroups of index 2 using (G,B)-local systems and a result of Watanabe, then analyzes the generalized Fitting subgroup through a chain of normal subgroups, using crossed products and Picard groups for odd index extensions. A corollary verifies Harada's conjecture on the listed representatives. The paper explicitly relies on GAP and Magma computations in several places and openly states two unresolved cases concerning whether inertial quotients are Morita invariants.","tokens_in":36015,"tokens_out":6488,"duration_ms":66848,"significance":"If the classification is correct, it is a substantial contribution: it gives the first explicit Morita equivalence classification for blocks with elementary abelian defect groups of order 32, verifies Harada's conjecture for these blocks, and extends the reach of the (G,B)-local-system method and the crossed-product/Picard-group machinery of Eaton and Livesey. The paper is genuinely theorem-driven, with no fitted parameters and no circular dependence on its own conclusion; its main structural proof is a carefully organized minimal-counterexample reduction. The explicit list of thirty-four classes and the detailed block-chain analysis provide a useful reference. The honesty of the author in flagging the computational dependencies and the two unresolved inertial-quotient cases is commendable, though those same points are the main obstacles to accepting the completeness claims as they stand.","major_comments":[{"comment":"The existence of (G,B)-local systems for inertial quotients E = C3×C3 and E = (C7⋊C3)_1 is the load-bearing step needed for Proposition 3.5 and hence for excluding normal subgroups of index 2 in step (III) of the proof of Theorem 1.1. The proof of Proposition 3.4, however, delegates the decisive verification to the GAP command OrthogonalEmbeddings, stating only that there is a unique solution in each required case and that uniqueness implies NE(Q)-stability. No GAP code, input matrices N^0, output logs, or certificates are provided, and the detailed computations are referred to the author's PhD thesis. A reader cannot independently verify this crucial step from the preprint alone. Please provide the computational data, scripts, and certificates, or replace the computational assertion by a complete proof.","section":"§3, Proposition 3.4"},{"comment":"Proposition 2.8 determines the complete list of possible inertial quotients by asserting that 'an explicit computation (using Magma)' yields the displayed diagram of subgroups of GL5(2) of odd order. This enumeration is the starting point for all subsequent case distinctions, including the subsection analysis in the same proposition. The evidence offered is only a diagram with no code or machine-readable output. This is a second load-bearing computational dependence that cannot be checked from the manuscript. Please include the Magma code or an independent derivation of the subgroup classification and the actions on D.","section":"§2, Proposition 2.8"},{"comment":"Corollary 5.3 explicitly leaves open the possibility that a block with inertial quotient (C7⋊C3)_2 is Morita equivalent to the class labeled (v), and that a block with inertial quotient (C7⋊C3)×C3 and 15 simple modules is Morita equivalent to (xi) or (xii). This means that the labels 'i.q. C5', 'i.q. C15', and 'i.q. (C7⋊C3)×C3' in Theorem 1.1 are properties of the chosen representatives, not established invariants of the Morita equivalence classes. Since the theorem presents these labels parenthetically in the list, the current wording could mislead a reader into thinking that the inertial quotient is part of the Morita invariant. Please reformulate the statement so that the labels are explicitly representative-dependent, and discuss the consequences for the completeness claim, including why the two unresolved possibilities do not affect Theorem 1.1 as stated.","section":"§5, Corollary 5.3"},{"comment":"The verification of Harada's conjecture is described as a Magma computation checking, for each of the listed classes, that no proper subset J with fewer than k(B)/2 elements satisfies equation (†). No code, input data, or output is supplied, and the decomposition matrices and character tables used are not listed. Since the claimed verification is a finite but nontrivial computational check across all subsets for several groups, please provide the scripts or a detailed table of the data and verification results.","section":"§6, Harada's conjecture"}],"minor_comments":[{"comment":"The abbreviation 'i.q.' is used throughout the theorem and later sections but is not defined in the introduction; it would help to define it explicitly as 'inertial quotient' at first use.","section":"§1, Introduction"},{"comment":"The diagram of subgroups of GL5(2) is difficult to read as typeset, with arrows appearing broken or overlapping; please redraw it with a clearer layout or replace it with a table of inclusions and actions.","section":"§2, Proposition 2.8"},{"comment":"The notation O∗L̂ and LK(ĈL(Y)) is used without a full definition of the twisted group algebra and its character group; a brief explanation or references to [48, §5.12] and [58, §1.2] would improve readability.","section":"§3, Proposition 3.4"},{"comment":"The two paragraphs discussing the pairs (16,5) and (32,15) are dense and would benefit from being split into clearly separated cases or a table summarizing which classes can have which inertial quotients.","section":"§5, Corollary 5.3"},{"comment":"Reference [2] is cited as an arXiv preprint from 2019; if it has appeared in a journal, the published version should be cited.","section":"References"},{"comment":"There are several typographical issues such as 'abeli an' in the abstract and 'isometrics' for 'isometries'; a careful proofread is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper presents an elaborate and largely convincing reduction argument, and the overall strategy is sound. My main concern is reproducibility: the classification depends at several critical points on GAP and Magma computations that are not included, and the author explicitly refers to the PhD thesis for details. In a classification theorem, the computational certificates or scripts should be part of the archival record. The two admitted open cases in Corollary 5.3 are not fatal to Theorem 1.1, but they should be prominently framed as limitations of the inertial-quotient part of the classification. I would support publication after the computational evidence is made available or replaced by proofs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe short version: this is a real classification result, the first explicit one for elementary abelian defect group (C2)^5, and it verifies Harada's conjecture for those blocks. It deserves a serious referee. The main caveat is not the math but the bookkeeping: several load-bearing steps are handed to GAP and Magma without scripts or certificates, so a referee cannot fully check them from the preprint.\n\nWhat's new: the list of 31 principal and 3 nonprincipal classes (Theorem 1.1) and the proof that defect groups are invariant under Morita equivalence for this defect group. That is a genuine advance; the literature stops at (C2)^4. The overall strategy is a systematic assembly of known tools—Fong–Reynolds, Fong's second reduction, Puig–Usami local systems, crossed products and Picard groups—rather than a new conceptual framework, but for a classification paper that's normal and the assembly is careful. The verification of Harada's conjecture is a nice bonus, and checking it class-by-class makes sense.\n\nWhere it gets soft: Proposition 3.4, which produces (G,B)-local systems for the two non-cyclic inertial quotients, reduces the decisive extension problem to `OrthogonalEmbeddings` in GAP and reports that the solution is unique and NE(Q)-stable. That uniqueness is load-bearing for the Morita equivalences in Proposition 3.5 and for step (III) of the main proof. No scripts, input data, or certificates are in the preprint; the paper points to the PhD thesis for details. Proposition 2.8's enumeration of odd-order subgroups of GL5(2) has the same issue, though that is less worrying because the list of possible inertial quotients matches what one expects and is checked against known classifications. I'm not saying the computations are wrong; I'm saying they are unverifiable from the preprint alone. A referee can handle that by asking for the computational artifacts or an independent run.\n\nThe other caveat is explicit in Corollary 5.3: the paper leaves open whether classes (v), (xi), (xii) might contain blocks with different inertial quotients. The author correctly notes that any such example would be a counterexample to Broué's conjecture, which is a strong reason to expect they don't exist. That is a disclosed limitation, not a hidden flaw.\n\nWho this is for: anyone working on Donovan's conjecture or explicit block classification. It moves the boundary one step up and gives a clean target list for (C2)^5. My recommendation is to send it to a good referee, with the clear instruction that the GAP/Magma computations need to be supplied or independently checked before the classification can be signed off. As a desk decision, this is absolutely not a reject; it's a revise-and-resubmit with teeth.","headline":"A solid, genuinely new classification for (C2)^5 blocks, held back only by unarchived GAP/Magma computations that a referee should demand.","tokens_in":36551,"tokens_out":2166,"would_cite":true,"duration_ms":21878,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["20C20","16D90","20C05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that every block with defect group $(C_2)^5$ is Morita equivalent to exactly one of 34 explicit classes, and that the defect group is invariant under such equivalences.","keywords":["Morita equivalence","block theory","defect group","elementary abelian 2-group","Donovan's conjecture","Harada's conjecture","(G,B)-local systems","Picard groups"],"falsifier":"Run the GAP OrthogonalEmbeddings computation described in Proposition 3.4 for the two inertial quotients $E=C_3\\times C_3$ and $E=(C_7\\rtimes C_3)_1$; if for any subgroup $Q$ with $\\mathrm{C}_E(Q)=E$ the matrix $N^0$ admits more than one extension up to permutation and sign of rows, the proof's claim of a unique $\\mathrm{N}_E(Q)$-stable $\\Delta_Q$ fails. Alternatively, exhibit any block with defect group $(C_2)^5$ whose Morita class is not on the list.","tokens_in":35490,"feed_emoji":"🧮","tokens_out":9206,"duration_ms":89415,"temperature":0.7,"pith_summary":"Two blocks of finite group algebras are Morita equivalent when they have equivalent module categories, so a Morita class is a module-theoretic fingerprint. This paper establishes that every block whose defect group is the elementary abelian 2-group $(C_2)^5$ is Morita equivalent to exactly one representative on an explicit list of 34 classes: 31 principal blocks and 3 nonprincipal blocks of explicit finite groups. No such explicit list was previously known for this defect group, and the result implies that the defect group itself is invariant under Morita equivalence in this case. A corollary is that Harada's conjecture holds for every block with this defect group.","feed_headline":"Blocks with 32-order defect groups reduce to 34 Morita classes","feed_subtitle":"The paper lists every possible class and confirms the character-sum conjecture for these blocks.","key_machinery":"The engine is the theory of $(G,B)$-local systems: a coherent family of perfect isometries $\\Delta_Y$ between the local twisted group algebra determined by the inertial quotient and the blocks of centralizers $\\mathrm{C}_G(Y)$ for $Y$ in an upwardly closed, $E$-stable set of subgroups of the defect group, each equivariant for the operation $\\lambda\\ast\\chi$. Existence of such a system lets the paper apply a splitting lemma to pass from $k$ to $O$ and to decompose blocks as tensor products with $OQ$. The second mechanism is crossed-product structure: when a block covers a $G$-stable block $b$ of a normal subgroup of odd index, the block is Morita equivalent to a crossed product whose possible actions are controlled by the Picard group of $b$; the relevant Picard groups are computed through exact sequences for trivial-source and endopermutation-source auto-equivalences. A minimal-counterexample reduction shows that a counterexample would have to be quasiprimitive, with no normal subgroup of index 2, and reduces to a short list of quasisimple starting cases.","core_discovery":"Let $O$ be a complete discrete valuation ring with algebraically closed residue field of characteristic 2, let $G$ be a finite group, and let $B$ be a block of $OG$ whose defect group $D$ is isomorphic to $(C_2)^5$. The central claim, Theorem 1.1, is that $B$ is Morita equivalent to the principal block of precisely one of 31 listed groups, or to a nonprincipal block of one of three listed groups $(a),(b),(c)$; the list is exhaustive and the classes are pairwise distinct. The representatives range from the abelian group $(C_2)^5$ itself through products and extensions built from $A_4$, $A_5$, $\\mathrm{SL}_2(8)$, $\\mathrm{SL}_2(16)$, $\\mathrm{SL}_2(32)$, the sporadic group $J_1$, and their automorphism groups, with extraspecial $3$-groups appearing for the nonprincipal classes. The theorem also states that the defect group is invariant under Morita equivalence. The classification fixes the numerical invariants $k(B)$ and $l(B)$ for each class, with a small list of explicitly identified ambiguities, and, because the property in Harada's conjecture is Morita invariant, it suffices to check the listed representatives computationally, which the paper does.","pith_inferences":["The same block-chain and crossed-product machinery could plausibly be pushed to larger elementary abelian 2-groups, but the branching over subgroups of $\\mathrm{GL}_n(2)$ and the Picard-group computations grow rapidly, so order 64 would require new bounds rather than routine application.","Because the classification is stated over $O$ rather than merely over $k$, the list can serve as a concrete testbed for lifting questions and for comparing $k$-Morita versus $O$-Morita equivalence.","A natural next step the paper leaves open is to decide the derived equivalence and source-algebra equivalence relations among the 34 classes, especially for the nonprincipal blocks (b) and (c), whose derived equivalences are not settled here."],"forward_implications":["No further Morita equivalence classes of blocks with defect group $(C_2)^5$ exist: the 31 principal and 3 nonprincipal representatives are exhaustive.","Any block Morita equivalent to one of these has an isomorphic defect group, so the defect group's isomorphism class survives Morita equivalence in this case.","Every block with this defect group satisfies Harada's conjecture, because the defining property transfers across Morita equivalence and each listed representative was checked.","The exact invariants $k(B)$ and $l(B)$ distinguish almost all classes; the ambiguous cases are explicitly identified, and any block realizing such an ambiguity would provide a counterexample to Broué's abelian defect group conjecture."],"supporting_citations":[{"why":"Supplies the classification of blocks of quasisimple groups with abelian defect groups, the starting case for the minimal-counterexample reduction.","marker":"[15]"},{"why":"Provides the classification for defect group $(C_2)^4$ used to rule out normal subgroups of index 2 and to identify several crossed-product representatives.","marker":"[17]"},{"why":"Gives the splitting theorem over $k$ that the paper extends to $O$ using local systems.","marker":"[33]"},{"why":"Provides the $O$-splitting criterion that turns a perfect isometry invariant under the star construction into a Morita equivalence.","marker":"[61]"},{"why":"Develops $(G,B)$-local systems and gives the inductive extension criterion used in Proposition 3.4.","marker":"[49]"},{"why":"Handles the cyclic-inertial-quotient cases by producing the required perfect isometries.","marker":"[62]"},{"why":"Yields the exact sequences for Picard groups that bound the possible crossed-product actions in Method 4.5.","marker":"[5]"},{"why":"Describes blocks with normal defect group as twisted group algebras, used to classify the normal-defect-group case.","marker":"[35]"},{"why":"Gives the corollary that the defect group is invariant under Morita equivalence for the final statement.","marker":"[41]"}],"fun_headline_variants":["Morita classes for 32-element defect groups: full list, 34 total","34 Morita classes settle 32-order defect block puzzle","Defect groups of size 32: all Morita classes identified","Harada's conjecture verified via Morita classification"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification rests on computer calculations inside GAP, invoked through the OrthogonalEmbeddings command, that assert uniqueness and stability of certain isometries in Proposition 3.4; those outputs are not supplied as scripts or certificates in the paper, so if any of those computational assertions is wrong, the list of Morita classes could be incomplete or contain duplicates.","fun_headline_variants_meta":{"raw":{"variants":["Morita classes for 32-element defect groups: full list, 34 total","34 Morita classes settle 32-order defect block puzzle","Defect groups of size 32: all Morita classes identified","Harada's conjecture verified via Morita classification"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000432,"raw_usage":{"total_tokens":2163,"prompt_tokens":862,"completion_tokens":1301,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":478,"completion_tokens_details":{"reasoning_tokens":1229}},"tokens_in":478,"tokens_out":1301,"duration_ms":9227,"temperature":1.0,"reasoning_tokens":1229,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:38:11.273601+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the GAP OrthogonalEmbeddings computation described in Proposition 3.4 for the two inertial quotients $E=C_3\\times C_3$ and $E=(C_7\\rtimes C_3)_1$; if for any subgroup $Q$ with $\\mathrm{C}_E(Q)=E$ the matrix $N^0$ admits more than one extension up to permutation and sign of rows, the proof's claim of a unique $\\mathrm{N}_E(Q)$-stable $\\Delta_Q$ fails. Alternatively, exhibit any block with defect group $(C_2)^5$ whose Morita class is not on the list.","supporting_citations":[{"cited_title":"Eaton, R","cited_arxiv_id":null,"evidence_quote":"Supplies the classification of blocks of quasisimple groups with abelian defect groups, the starting case for the minimal-counterexample reduction."},{"cited_title":"Morita equivalence classes of blocks with elementary abelian defect groups of order 16","cited_arxiv_id":"1612.03485","evidence_quote":"Provides the classification for defect group $(C_2)^4$ used to rule out normal subgroups of index 2 and to identify several crossed-product representatives."},{"cited_title":"Koshitani and B","cited_arxiv_id":null,"evidence_quote":"Gives the splitting theorem over $k$ that the paper extends to $O$ using local systems."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the $O$-splitting criterion that turns a perfect isometry invariant under the star construction into a Morita equivalence."},{"cited_title":"Puig and Y","cited_arxiv_id":null,"evidence_quote":"Develops $(G,B)$-local systems and gives the inductive extension criterion used in Proposition 3.4."},{"cited_title":"Watanabe, On perfect isometries for blocks with abelian defect groups and cyclic hyper- focal subgroups, Kumamoto Journal of Mathematics 18 (2005), 85-92","cited_arxiv_id":null,"evidence_quote":"Handles the cyclic-inertial-quotient cases by producing the required perfect isometries."},{"cited_title":"Boltje, R","cited_arxiv_id":null,"evidence_quote":"Yields the exact sequences for Picard groups that bound the possible crossed-product actions in Method 4.5."},{"cited_title":"K¨ ulshammer, Crossed products and blocks with normal defect groups , Communications in Algebra 13 (1985), 147-168","cited_arxiv_id":null,"evidence_quote":"Describes blocks with normal defect group as twisted group algebras, used to classify the normal-defect-group case."},{"cited_title":"Linckelmann, On automorphisms and focal subgroups of blocks , PIMS Summer School and Workshop, Springer, Cham (2016)","cited_arxiv_id":null,"evidence_quote":"Gives the corollary that the defect group is invariant under Morita equivalence for the final statement."}],"review_version":1}