{"id":"d7179ef0-d44d-48e0-943e-b76ed4437ef1","arxiv_id":"1908.01273","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In the affine case, every symmetric graph with a complete quotient that is almost multi-covered and yields a nontrivial linear space is either an affine flag graph or one of a few sporadic graphs.","lead":"This paper classifies the affine case of symmetric graphs whose complete quotient is almost multi-covered, completing a project started in 2002. It introduces a new family of connected flag graphs and proves they are the only ones in this case.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The n=2 classification rests on omitted orbit-counting computations; the self-paired orbit enumeration in §3.3–3.4 and Lemma 7's valency formula need independent verification before the family in Theorem 1(a)(ii) can be accepted as complete.","rationale":"Reading the paper in good faith, the main theorem has a clear and plausible strategy: reduce via Lemma 2 to flag graphs over doubly point-transitive linear spaces, use the classifications [17] and Cameron-Kantor [2] to reduce to affine cases, then construct or eliminate each case. The almost simple case is outsourced to [13], and the affine cases are handled in §3.2–3.5. The structural parts—Lemma 4, the reduction to ASL(n,q), and the divisibility-based eliminations in §3.5—are checkable from the text and appear sound. However, the n=2 affine plane case is the only place where the proof relies on direct orbit computations that are not displayed. Lemma 7 explicitly omits the computational details; §3.4 contains no derivation of the number of self-paired orbits; §3.3 similarly states the form of Ψ without proof. Since Theorem 1(a)(ii) rests on these, the classification is conditionally supported rather than fully verified. The reader's weakest assumption identifies the same issue, so I do not propose a change of verdict: the appropriate action is to supply the omitted computations or an independent computational check. The black-box classifications are standard, named inputs and, in my view, do not constitute a separate load-bearing weakness beyond the usual reliance on the classification of finite simple groups.","tokens_in":12240,"tokens_out":13220,"duration_ms":132790,"concrete_test":"Use GAP/Magma to test the n=2 orbit enumeration exactly as described in §3.3–3.4. For q=4,5,7,9,11, take G=ASL(2,q)⋊H with H as in Definition 1, including the §3.4 cases H=⟨C_ℓ⟩ for each ℓ∈F_q^× of even order. Compute the G-orbits on the set F(2,q) of compatible flag pairs from (2); for each c∈F_q^× satisfying t(A_{c,δ},0,δ)∈G0, check that the orbit of ((e1,⟨e1⟩),(ce2,⟨e2⟩)) is self-paired; count all self-paired compatible orbits and compare with the stated (p−1)/|ℓ|+1; construct the flag graphs Γ_{G,c}(2,q), compute their valencies directly, and compare with Lemma 7. If all counts and valencies match for these small parameters, the omitted computation is corroborated; any mismatch identifies the needed correction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Theorem 1(a)(ii), the genuinely new objects are the graphs Γ_{G,c}(2,q). The proof that these are all the connected graphs in the affine plane case is not fully written out. In §3.3, after ASL(2,q)≤G is established, the assertion that every self-paired compatible G-orbital on Ψ^+(2,q) has the form (3) is made without proof; in §3.4 the analogous claim for q=p is summarized as 'By a similar analysis', and the count (p−1)/|ℓ|+1 is asserted with no derivation. Lemma 7's valency computation contains the explicit sentence 'computational details are omitted' at the point where the formula ℓ_c=|Λ(H)|i/ℓ is needed. These are not merely cosmetic omissions: the completeness and parameter list of the n=2 family in Theorem 1(a)(ii) depend on exactly which c∈F_q^× yield self-paired G-orbits and on the orbit length ℓ_c. A miscount would make the family either too large or too small, and a wrong valency would make Lemma 7 inconsistent with the claimed valency formula iq(q−1)^2/(ts). The quoted classifications [17] and Cameron-Kantor [2] are standard inputs and are not the soft spot; the soft spot is the unstated finite orbit enumeration in the affine plane n=2 case.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper classifies G-symmetric graphs Gamma admitting a nontrivial G-invariant partition B with block size at least 3, such that the quotient graph Gamma_B is complete, Gamma is an almost multicover of Gamma_B, the induced incidence structure D = D(Gamma, B) is a nontrivial linear space, and G contains a regular normal elementary abelian subgroup. The main result, Theorem 1, states that either D is isomorphic to AG(n,q) with n >= 3 and Gamma one of the affine flag graphs Gamma^+(n,q), Gamma^=(n,q), or Gamma^simeq(n,q); or D is affine plane AG(2,q) with n = 2 and Gamma is Gamma^=(2,q) or a new connected graph Gamma_{G,c}(2,q) defined in Definition 1; or D is AG(2,2) or AG(2,4) with the listed sporadic graphs. The proof uses the flag-graph correspondence from [26], the classification of doubly point-transitive linear spaces [17], and the Cameron-Kantor theorem [2]. The n=2 part of the classification relies on several verifications that are omitted from the text, in particular the enumeration of self-paired orbitals and the valency computation in Lemma 7.","tokens_in":12572,"tokens_out":7329,"duration_ms":69994,"significance":"If correct, this result completes the classification of a natural family of imprimitive symmetric graphs with complete quotients, extending earlier work of Gardiner-Praeger, Giulietti et al., and Fang et al. The reduction to flag graphs and the use of the Kantor classification of doubly point-transitive linear spaces are appropriate and give a clean structural framework. The paper explicitly defines the new graphs Gamma_{G,c}(2,q) and states their parameters. However, the completeness and parameter list of the n=2 family in Theorem 1(a)(ii) depend on unproved computational assertions, so the central claim needs additional support before the classification can be accepted as fully established.","major_comments":[{"comment":"The assertion that every self-paired G-orbital Psi on Psi^+(2,q) is of the form (3) for some c in F_q^times is stated without proof. This claim is load-bearing for Theorem 1(a)(ii), since the parameter c controls which graphs occur in the new family. Please provide the orbit enumeration under ASL(2,q) (or a reference that contains it) and prove that all compatible orbitals are accounted for.","section":"Section 3.3, Eq. (3)"},{"comment":"The valency computation for Gamma_{G,c}(2,q) contains the sentence '(computational details are omitted)' immediately after the formula ell_c = |Lambda(H)|i/ell = i(p^ell-1)/(ts). Lemma 7 uses this formula to conclude that the valency is i q (q-1)^2/(ts), and this valency is part of the claimed description of the n=2 family. The omitted computation is therefore not cosmetic; it must be written out or replaced by a precise reference.","section":"Section 3.3, Lemma 7"},{"comment":"The 'similar analysis' for the case G0 <= GL(2,p) with p = 5,7,11,19,23,29,59 asserts three things without proof: that a self-paired G-orbit on Psi^+(2,p) exists iff [[1,0],[0,-1]] is in G0; that this is equivalent to G0 = SL(2,p) ⋊ <C_ell> for some ell of even order; and that the number of self-paired compatible G-orbits is (p-1)/|ell|+1. These assertions determine which graphs appear in this case and how many, so they are essential to the completeness statement in Theorem 1(a)(ii). Please supply the missing derivation.","section":"Section 3.4"},{"comment":"The exclusion of all but two AGL(1,v) cases rests on the inequality d < (s^t - s)/(s - 1) with the exceptions (p,t,d) = (2,2,2) or (2,2,4), stated with 'It can be verified'. Since this is a necessary step in ruling out the AΓL(1,v) family, the verification should be included or made explicit, for example as a short number-theoretic argument.","section":"Section 3.2"},{"comment":"In cases (ix), (x) and (xi), the sentence 'Similarly, there is no feasible G-orbit on the flag set of D' is given without the details of the divisibility check. If these checks are routine, they should still be summarized so that the reader can confirm that the sporadic Hering designs are excluded in the same way as cases (vi) and (vii).","section":"Section 3.5"}],"minor_comments":[{"comment":"The phrase 'belongs to a family of connected graphs' is vague; the theorem should state explicitly that the graphs are exactly the Gamma_{G,c}(2,q) defined in Definition 1 (for those c satisfying the stated condition), together with their order and valency.","section":"Theorem 1(a)(ii)"},{"comment":"The paper does not discuss isomorphisms among the graphs Gamma_{G,c}(2,q) for different c or different groups G. A remark on when two such graphs are isomorphic would help the reader interpret the classification.","section":"Definition 1 / Lemma 7"},{"comment":"In the p=2 case, the claim that 'we can choose a in F_q^times such that a+1 != 0 and h_{a,c}^2 g in J_0 setminus G_{0,<e2>}' is left to the reader. The element should be written out and its non-membership in G_{0,<e2>} justified.","section":"Lemma 6, Case 2"},{"comment":"The statement that 'there are exactly three self-paired G-orbits on F(n,q) compatible with Omega when n >= 3' is cited to [26, Lemma 3.9], but the notation in [26] differs from the present paper; a short explanation of how the three orbitals correspond to intersecting, parallel and skew lines would improve readability.","section":"Section 3.3, n >= 3"},{"comment":"The displayed line for Eq. (3) contains two equal expressions on the same line; the notation for the paired orbital would be clearer if the direction of the orbital were indicated explicitly.","section":"Equation (3)"}],"recommendation":"major_revision","confidential_remarks":"This is a competent classification paper that appears to be on the right track, but the n=2 part is under-supported by omitted computational details. I recommend asking the authors to add the missing verifications, at least in an appendix. Also, reference [3] is cited as an 'incomplete version' downloaded from ResearchGate; the authors should check whether a published or more reliable version exists."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper completes a classification project that has been open since [26], by handling the affine case. The genuinely new objects are the graphs Γ_{G,c}(2,q) in Theorem 1(a)(ii) for n=2, and the paper rightly notes that this completes the classification of symmetric graphs with complete quotients satisfying (i) and (ii). The overall architecture is sound: it reduces the problem to flag graphs over doubly transitive linear spaces, invokes the known classification of those spaces, and uses the Cameron–Kantor theorem to pin down the possible point stabilizers. The case analysis in §3.5 ruling out the sporadic possibilities is straightforward and convincing.\n\nThe soft spot is concentrated in the n=2 affine plane case. The proof that the self-paired compatible orbitals have the form (3) is asserted without derivation; §3.4's \"By a similar analysis\" hides the count (p−1)/|ℓ|+1; and Lemma 7's valency formula rests on the phrase \"computational details are omitted\" for the orbit length ℓ_c. These are not cosmetic. The completeness and parameter list of the n=2 family depend on exactly which c ∈ F_q^× yield self-paired orbitals, and on the orbit length. A miscount would make the family too large or too small, and a wrong valency would make Lemma 7 inconsistent with the claimed valency formula. The external inputs—the classification of doubly transitive linear spaces and Cameron–Kantor—are standard and not the issue.\n\nMy assessment is that the paper is very likely correct, but the written proof is not fully self-contained at exactly the point where the new family is defined. The authors should be asked to supply the omitted computations, either as an appendix or a supplementary file, before the classification is taken as fully verified. The referee should check those computations specifically.\n\nThis paper is for algebraic graph theorists and design theorists working on symmetric graphs and imprimitive quotients. It deserves serious peer review, not desk rejection, but with a request for the missing details. I would not cite the n=2 family as a black box until the computations are available, but I would bring the paper to a reading group to go through the affine case.","headline":"Genuine completion of a classification, but the n=2 case hides orbit-counting computations that need to be supplied before the theorem is fully verified.","tokens_in":13025,"tokens_out":3103,"would_cite":false,"duration_ms":28354,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05C25","05B05","51E15"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper classifies affine flag graphs and completes the symmetric-graph classification in the linear-space case.","keywords":["symmetric graph","arc-transitive graph","flag graph","linear space","affine space","imprimitive symmetric graph","complete quotient","classification"],"falsifier":"Recompute the orbit length $\\ell_c$ of $c$ under $\\Lambda(H)$ for a concrete plane case, say $q=9$ and a subgroup of $\\Gamma\\mathrm{L}(1,9)$ in standard form, and compare it with the claimed $i(p^\\ell-1)/(ts)$; a mismatch would invalidate Lemma 7 and the classification of the new family $\\Gamma_{G,c}(2,q)$.","tokens_in":12081,"feed_emoji":"📐","tokens_out":11505,"duration_ms":104113,"temperature":0.7,"pith_summary":"This paper classifies a family of symmetric graphs, graphs whose automorphism group is transitive on ordered adjacent pairs. The setting is a nontrivial block partition with a complete quotient graph, where each block is an almost multicover: for any two adjacent blocks, exactly one vertex of the first block has no neighbour in the second. Such a graph yields a 2-design, and the paper treats the case where that design is a nontrivial linear space whose automorphism group contains an elementary abelian regular normal subgroup, the affine case. The conclusion is that, apart from a few sporadic small-plane exceptions, the only possible graphs are affine flag graphs built from the Desarguesian affine space $\\mathrm{AG}(n,q)$: the three families $\\Gamma^+(n,q)$, $\\Gamma^=(n,q)$, $\\Gamma^{\\simeq}(n,q)$ for $n\\ge 3$, and for $n=2$ the graph $\\Gamma^=(2,q)$ or a newly constructed connected family $\\Gamma_{G,c}(2,q)$. Together with earlier results, this completes the classification of symmetric graphs satisfying the block-size and complete-quotient conditions.","feed_headline":"Affine flag graphs finish a symmetric-graph classification","feed_subtitle":"Under an affine symmetry assumption, all such graphs are flag graphs or two sporadic families.","key_machinery":"The carrying construction is the flag graph $\\Gamma(D,\\Omega,\\Psi)$: vertices are flags $(\\sigma,L)$ of a design $D$ lying in a feasible $G$-orbit $\\Omega$, and two flags are adjacent exactly when their ordered pair lies in a self-paired compatible orbital $\\Psi$. The feasibility axioms (A1)--(A4) and the compatibility axiom (A5) encode the block partition and the almost-multicover property, so the graph problem becomes a design problem. Lemma 2 reduces the given $\\Gamma$ to a flag graph of a $(G,2)$-point-transitive and block-transitive 2-design, and the classification of doubly point-transitive linear spaces supplies all possible affine pairs $(G,D)$. Within the affine-space cases, a theorem on 2-transitive subgroups of $\\Gamma\\mathrm{L}(n,q)$ forces $\\mathrm{SL}(n,q)\\le G_0$, leaving exactly the three self-paired orbitals for $n\\ge 3$; for $n=2$ the new graphs $\\Gamma_{G,c}(2,q)$ appear when a matrix $A_{c,\\delta}$ with a field automorphism $\\delta$ lies in $G_0$.","core_discovery":"Theorem 1 states: if $\\Gamma$ is $G$-symmetric with a nontrivial $G$-invariant partition $\\mathcal{B}$ of block size at least 3, the quotient $\\Gamma_{\\mathcal{B}}$ is complete, $\\Gamma$ almost multicovers $\\Gamma_{\\mathcal{B}}$, the induced incidence structure $D = D(\\Gamma,\\mathcal{B})$ is a nontrivial linear space, and $G$ contains a regular normal elementary abelian subgroup of order $q^n$, then only the following occur. Either $D \\cong \\mathrm{AG}(n,q)$ with $|B|=(q^n-1)/(q-1)$ and multiplicity $m=q-1$; in that case for $n\\ge 3$ the graph is one of $\\Gamma^+(n,q)$, $\\Gamma^=(n,q)$, $\\Gamma^{\\simeq}(n,q)$, and for $n=2$ it is $\\Gamma^=(2,q)$ or one of the connected graphs $\\Gamma_{G,c}(2,q)$ of order $q^2(q+1)$. Or $D \\cong \\mathrm{AG}(2,2)$, yielding $3\\cdot K_{2,2}$ or $4\\cdot K_3$, or $D \\cong \\mathrm{AG}(2,4)$, yielding $\\Gamma^+(2,4)$ or $\\Gamma^=(2,4)$. The proof also shows that the exceptional nearfield plane, the Hering plane, and the Hering designs admit no feasible flag orbit, so they produce no graphs.","pith_inferences":["A direct consequence not spelled out in the paper is an arithmetic parametrization: once Lemma 7 is confirmed, the new $n=2$ graphs are indexed by the standard parameters $(t,e,s)$ of $\\Lambda(H)$ and by $c\\in\\mathbb{F}_q^\\times$, so an enumeration of pairwise non-isomorphic $\\Gamma_{G,c}(2,q)$ is a natural next step.","The same flag-graph reduction could be tested on doubly transitive 2-designs that are not linear spaces; the paper shows the affine-space case is exhausted, so any further examples would have to come from designs in the $\\lambda=m+1$ regime or from groups that are neither almost simple nor affine.","The argument suggests even characteristic is structurally different for planes: because $A_{c,\\mathrm{id}}\\in\\mathrm{SL}(2,q)$ when $q$ is even, every orbital on intersecting line pairs is self-paired, whereas odd characteristic imposes a field-automorphism condition; thus the family $\\Gamma_{G,c}(2,q)$ is likely richer for even $q$."],"forward_implications":["Together with the earlier classifications in [6], [11], [13] and [26], Theorem 1 completes the classification of all $G$-symmetric triples $(\\Gamma,G,\\mathcal{B})$ with $|B|\\ge 3$, complete quotient, and almost-multicover property.","For $n\\ge 3$ the only symmetric graphs in the affine linear-space case are the three affine flag graphs $\\Gamma^+(n,q)$, $\\Gamma^=(n,q)$, $\\Gamma^{\\simeq}(n,q)$; in particular, the families require no sporadic group.","For $n=2$, the new connected graphs $\\Gamma_{G,c}(2,q)$ have order $q^2(q+1)$ and valency $i q(q-1)^2/(ts)$, with parameters read from the standard form of a subgroup of $\\Gamma\\mathrm{L}(1,q)$.","The sporadic small affine planes produce exactly four graphs: $3\\cdot K_{2,2}$ and $4\\cdot K_3$ from $\\mathrm{AG}(2,2)$, and $\\Gamma^+(2,4)$ and $\\Gamma^=(2,4)$ from $\\mathrm{AG}(2,4)$.","Exceptional affine linear spaces, namely the nearfield plane, the Hering plane, and the Hering designs, are ruled out by divisibility and $\\mathrm{SL}(2,9)$ arguments, so they contribute no flag graphs."],"supporting_citations":[{"why":"Introduces the flag graph construction and the feasibility/compatibility axioms, and supplies Lemma 2 reducing the graph problem to a design problem.","marker":"[26]"},{"why":"Classifies doubly point-transitive linear spaces and supplies the list of possible affine pairs (G,D) used to launch the proof.","marker":"[17]"},{"why":"Gives the 2-transitive subgroup theorem for ΓL(n,q) that forces SL(n,q) ≤ G0 in the affine-space cases and limits the n=2 possibilities.","marker":"[2]"},{"why":"Provides the earlier complete-quotient classification in the λ=m+1 case and the feasibility criterion used to rule out exceptional spaces.","marker":"[6]"},{"why":"Classifies the trivial linear-space case, one of the ingredients Theorem 1 completes.","marker":"[11]"},{"why":"Classifies the almost simple case, the counterpart Theorem 1 handles for affine groups.","marker":"[13]"},{"why":"Gives the standard form of subgroups of ΓL(1,p^ℓ), used to compute the orbit length and valency of the new n=2 graphs.","marker":"[8]"}],"fun_headline_variants":["Affine flag graphs complete symmetric graph classification","Symmetric graph classification final piece: affine flag graphs","Affine flag graphs solve symmetric graph classification","Classification completed: symmetric graphs with complete quotients","Affine flag graphs: the missing piece in symmetric graph classification"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof inherits the correctness of two classification theorems used as black boxes, and for the plane case it depends on an orbit-length calculation in Lemma 7 whose computational details are omitted; if that valency formula is wrong, the completeness of the new $n=2$ family fails.","fun_headline_variants_meta":{"raw":{"variants":["Affine flag graphs complete symmetric graph classification","Symmetric graph classification final piece: affine flag graphs","Affine flag graphs solve symmetric graph classification","Classification completed: symmetric graphs with complete quotients","Affine flag graphs: the missing piece in symmetric graph classification"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000614,"raw_usage":{"total_tokens":2986,"prompt_tokens":1213,"completion_tokens":1773,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":829,"completion_tokens_details":{"reasoning_tokens":1700}},"tokens_in":829,"tokens_out":1773,"duration_ms":14119,"temperature":1.0,"reasoning_tokens":1700,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:17:19.561056+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the orbit length $\\ell_c$ of $c$ under $\\Lambda(H)$ for a concrete plane case, say $q=9$ and a subgroup of $\\Gamma\\mathrm{L}(1,9)$ in standard form, and compare it with the claimed $i(p^\\ell-1)/(ts)$; a mismatch would invalidate Lemma 7 and the classification of the new family $\\Gamma_{G,c}(2,q)$.","supporting_citations":[{"cited_title":"Zhou, Constructing a class of symmetric graphs, European J","cited_arxiv_id":null,"evidence_quote":"Introduces the flag graph construction and the feasibility/compatibility axioms, and supplies Lemma 2 reducing the graph problem to a design problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Classifies doubly point-transitive linear spaces and supplies the list of possible affine pairs (G,D) used to launch the proof."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the 2-transitive subgroup theorem for ΓL(n,q) that forces SL(n,q) ≤ G0 in the affine-space cases and limits the n=2 possibilities."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the earlier complete-quotient classification in the λ=m+1 case and the feasibility criterion used to rule out exceptional spaces."},{"cited_title":"Gardiner and C","cited_arxiv_id":null,"evidence_quote":"Classifies the trivial linear-space case, one of the ingredients Theorem 1 completes."},{"cited_title":"Giulietti, S","cited_arxiv_id":null,"evidence_quote":"Classifies the almost simple case, the counterpart Theorem 1 handles for affine groups."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the standard form of subgroups of ΓL(1,p^ℓ), used to compute the orbit length and valency of the new n=2 graphs."}],"review_version":1}