{"id":"9d018801-12e3-4129-962d-f1a96fd0a72b","arxiv_id":"2411.16288","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":0.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A literature survey that frames Cameron-Liebler sets and low-degree Boolean functions through the common lens of association schemes and Delsarte theory.","lead":"This paper surveys the mathematics of Cameron-Liebler sets and low-degree Boolean functions, unifying results across Hamming, Johnson, and Grassmann graphs using association schemes. It is a reference guide for researchers in finite geometry, coding theory, and Boolean function analysis, though it contains several mathematical errors in key examples.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reader's line-count objection to §5.1.1 misfires; the real load-bearing defect is the O^+(4,q)/O^-(4,q) contradiction in the Bruen–Drudge entry of the central inventory.","rationale":"The paper's central aim is to give a usable survey from an association-scheme perspective, and the Jq(4,2) inventory in Section 5.1/Table 1 is the part a reader would most rely on. The reader's headline objection about the Bruen–Drudge line census is based on a misreading: the displayed binomial coefficients in (i) and (ii) do sum, together with (iii) and (iv), to the full line count. I therefore do not regard that as a real defect. The real soft spot is the stabilizer contradiction: the prose says O^+(4,q) while the table and the mathematics of the non-square quadratic form point to O^-(4,q). Because the survey's value is precisely that it organizes these scattered examples into one reliable table, an unresolved contradiction in one of the oldest and best-known families is enough to justify the reader's CONDITIONAL verdict. The Hamming eigenvalue formula in Example 2.1 should also be re-checked against [12, Th. 9.2.1]; as rendered it is wrong, but it is less central to the survey's main claim than the stabilizer conflict.","tokens_in":67,"tokens_out":12351,"duration_ms":234390,"concrete_test":"Check the Bruen–Drudge example against the original source [15], or independently compute the line orbits of the orthogonal group of Q = ξx1^2 + x2^2 + x3^2 + x4^2 with ξ a non-square. If the stabilizer is O^-(4,q), as Table 1 claims, replace the prose in §5.1.1; if it is truly O^+(4,q), then the stated point count q^2+1 is wrong. This single check settles which side of the contradiction is the typo.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's numerical objection to §5.1.1 does not hold as stated: items (i) and (ii) each count \\binom{q^2+1}{2} lines, not (q^2+1)/2 lines, and summing all four items gives exactly (q^2+1)(q^2+q+1), the total number of lines. The genuine load-bearing issue is the stabilizer-group contradiction. The prose in §5.1.1 says the Bruen–Drudge example has stabilizer O^+(4,q), while Table 1 and the construction itself require O^-(4,q): the displayed form Q has non-square discriminant, its zero set has q^2+1 points, and the group fixing such a quadric is the elliptic orthogonal group. Moreover O^+(4,q) is not a simple group, so the phrase 'finite simple group of type O+(4,q)' is internally impossible. Since Section 5.1/Table 1 is the promised inventory of exceptional degree-1 examples in Jq(4,2), a reader cannot determine the correct stabilizer without going to the primary source. This is likely a one-line typo, but it is exactly the type of inconsistency that prevents the survey from being usable as a reference. The Hamming eigenvalue formula in Example 2.1 also appears corrupted as rendered (θ_j = q(n−j)−d with dimension \\binom{d}{j}(q−1)^j); if that is not an OCR artifact, it is a second reference-level error.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This survey aims to present known results on Cameron-Liebler sets and low-degree Boolean functions for Hamming graphs, Johnson graphs, and Grassmann graphs from a unified association-scheme and Delsarte-theory perspective. It covers spectral preliminaries, design-orthogonality, equitable partitions, the FKN theorem and its analogues, Nisan-Szegedy-type junta bounds, the Khot-Minzer-Safra theorem, the classification of degree-1 functions on Grassmann graphs, and an inventory of exceptional degree-1 examples in J_q(4,2). The paper is primarily an exposition with a broad bibliography, and its main value would be as a reference connecting finite geometry, Boolean function analysis, coding theory, and cryptography.","tokens_in":1261,"tokens_out":1444,"duration_ms":273916,"significance":"If the reference-level details are correct, this survey fills a genuine need: it collects recent results from several communities that rarely cite each other and phrases them in a common association-scheme language. The table of exceptional degree-1 examples in J_q(4,2), the summary of small-q classification results, and the discussion of the Khot-Minzer-Safra theorem are potentially useful entry points for both finite geometers and Boolean-function analysts. I verified that the line census in §5.1.1 is arithmetically consistent when items (i) and (ii) are read as binomial coefficients, so the line-count objection does not land. However, the manuscript contains internal inconsistencies in exactly the places a reader would use as reference, most importantly the stabilizer-group contradiction in the central inventory and the corrupted eigenvalue formula in the spectral preliminaries.","major_comments":[{"comment":"The stabilizer of the Bruen-Drudge family is stated in §5.1.1 as \"the finite simple group of type O+(4,q)\", while Table 1 lists O-(4,q) and §5.1.2 refers to \"a point in O-(4,q)\". The displayed quadratic form Q has non-square discriminant, its zero set has q^2+1 points, and the group fixing such a quadric is the elliptic orthogonal group O-(4,q); moreover O+(4,q) is not a simple group for any q, so the phrase \"finite simple group of type O+(4,q)\" is internally impossible. Since Table 1 is the promised complete inventory of exceptional degree-1 examples in J_q(4,2), this contradiction prevents a reader from trusting the inventory without returning to the primary sources. Please correct the prose or the table and verify the remaining group entries in the table.","section":"§5.1.1 / Table 1"},{"comment":"The Hamming graph eigenvalue formula as rendered is theta_j = q(n-j)-d with dimension d choose j times (q-1)^j, where d is not defined and the formula is not the spectrum of H(n,q). The correct value is theta_j = q(n-j)-j, equivalently n(q-1)-qj, with multiplicity n choose j times (q-1)^j. This is part of the common spectral dictionary on which the survey's unified framework relies, so the formula should be corrected; if this is an artifact of the rendering, the displayed version in the manuscript should be fixed.","section":"§2.1, Example 2.1"}],"minor_comments":[{"comment":"In the inversion formula for the idempotents, the sum is written as sum_{i=1}^m Q_{ij}A_i; it should run from i=0 to m, since E_0 = v^{-1}J is part of the basis of minimal idempotents.","section":"§2.2"},{"comment":"The eigenvalue index ranges are stated as 0 ≤ j ≤ n for both the Johnson and Grassmann graphs, but these graphs have m+1 distinct eigenvalues; the range should be 0 ≤ j ≤ m.","section":"§2.1, Examples 2.2 and 2.3"},{"comment":"The sentence \"Then the set consists of the following is an Boolean degree 1 function with x = 7\" should be rephrased, for example as \"Then the set consisting of the following lines is a Boolean degree 1 function with x = 7.\"","section":"§5.1.4"},{"comment":"The phrase \"show the that the non-trivial examples\" contains a typo and should read \"show that the non-trivial examples\".","section":"§5.2"},{"comment":"The abbreviation \"pt.-stab.\" in the group column should be expanded or defined in the table caption, since it is not explained elsewhere in the text.","section":"Table 1"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: the survey earns its keep as a map, not as a source of new results. It fills a real gap by putting Cameron-Liebler sets, Boolean degree functions, and their relatives under one association-scheme framework, and its inventory of exceptional degree-1 examples in Jq(4,2) is useful. But two internal errors need correcting. The reader's line-count objection to the Bruen-Drudge census in §5.1.1 is a misfire. Items (i) and (ii) are binomial coefficients—each is (q^2+1 choose 2)—and the four items sum exactly to the total (q^2+1)(q^2+q+1) lines. That part is fine. The real problems are elsewhere. The prose says the Bruen-Drudge example has stabilizer 'the finite simple group of type O+(4,q)', while Table 1 correctly says O^-(4,q). The quadratic form in question has non-square discriminant, so the elliptic group is the one that acts; O^+(4,q) is not simple anyway. This contradiction sits in the central inventory, so it's load-bearing. It looks like a one-line typo, but a reference survey cannot carry it. Second, the Hamming graph eigenvalues in Example 2.1 are corrupted as printed: θ_j = q(n−j)−d with dimension binom(d,j)(q−1)^j. The correct formulas are θ_j = (q−1)n − qj with multiplicity binom(n,j)(q−1)^j. As written, 'd' appears where only 'n' makes sense. Also fixable, but it will mislead a newcomer. The rest reads as careful and honest. The author credits the right people, the cross-community translations are accurate, and the discussion of FKN, Nisan-Szegedy, and the Khot-Minzer-Safra theorem is reliable. The paper leans on the author's own results, but he actually proved a good chunk of what is surveyed, so that is not a problem. Who benefits: researchers or students who want to see why finite geometry and Boolean function analysis are converging, and where the open problems sit. The central structure is sound. I would send it to a serious referee, ask for verification of the Table 1 group entries and the Section 2 formulas, and accept after those fixes. I'd cite it once it is cleaned up. Recommendation: give it a proper referee; the errors are minor but real, and the survey deserves to be a standard entry point.","headline":"A useful cross-disciplinary survey with two fixable but real internal errors that should be cleaned up before it becomes the go-to reference.","tokens_in":23633,"tokens_out":9808,"would_cite":true,"duration_ms":70399,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05E30","05B05","05D05","05C50"],"pacs":[],"model":"deepseek-v4-flash","headline":"One association-scheme language describes low-degree Boolean functions and Cameron-Liebler sets in Hamming, Johnson, and Grassmann graphs.","keywords":["Cameron-Liebler sets","low-degree Boolean functions","Grassmann graphs","association schemes","Delsarte theory","Hamming scheme","Johnson scheme","junta theorems"],"falsifier":"Count the lines in the Section 5.1.1 construction: the four listed types total $(q^2+1)(q+2)$ lines, but $\\mathrm{PG}(3,q)$ has $(q^2+1)(q^2+q+1)$ lines, so the types do not partition the line set; checking whether an omitted line type is forced into the example would settle whether the construction stands. The table and text also disagree on whether the stabilizer is $\\mathrm{O}^+(4,q)$ or $\\mathrm{O}^-(4,q)$, and resolving that against the definition of the quadratic form would settle the inconsistency.","tokens_in":22636,"feed_emoji":"🧮","tokens_out":13222,"duration_ms":114841,"temperature":0.7,"pith_summary":"This survey's thesis is that association schemes and Delsarte theory give a single language for two topics usually studied separately: low-degree Boolean functions on the hypercube and its slices, and Cameron-Liebler sets in finite projective geometry. In that language, degree $d$ means living in the first $d+1$ eigenspaces of the scheme, and the known results line up: degree-1 functions on the Hamming and Johnson graphs are only dictators and their complements, while the Grassmann graph $J_q(4,2)$ admits many exceptional degree-1 examples, which the survey tabulates. The paper also records size bounds, divisibility conditions, and the structure theorem behind the 2-to-2 Games proof as facets of the same spectral picture. A reader would care because this is the closest existing statement that Boolean function analysis and finite geometry are studying the same objects.","feed_headline":"Only Grassmann graphs hide exotic low-degree Boolean functions","feed_subtitle":"A unified survey shows Hamming and Johnson degree-1 sets are trivial, while Jq(4,2) is full of exceptions.","key_machinery":"The central object is the association scheme of the relevant graph: a decomposition of the ambient vector space $\\mathbb{C}^X$ into common eigenspaces $V_0, V_1, \\ldots, V_m$ of the adjacency matrices, with $V_0$ spanned by the all-ones vector. A Boolean function has degree $d$ exactly when its characteristic vector lies in $V_0 + \\cdots + V_d$, and Delsarte's linear programming bound checks orthogonality to eigenspaces using only the small $Q$-matrix rather than the full projection. For the Grassmann case, the workhorse is a weighted function $g_{P,H}$ on lines of $\\mathrm{PG}(3,q)$ whose design-orthogonality with any degree-1 set yields an equation relating the size parameter $x$ to counts of lines through a point, in a hyperplane, and in both, with the possible line patterns strongly restricted; tactical decompositions and Block's lemma explain why group actions produce such sets.","core_discovery":"On the paper's own terms, the discovery is that a common formalism---the eigenspace decomposition of a cometric association scheme, together with Delsarte's linear programming bound and design-orthogonality---correctly organizes the classification results on Boolean degree-1 functions and Cameron-Liebler sets. In the Hamming scheme and the Johnson scheme, degree-1 Boolean functions are trivial; the Friedgut-Kalai-Naor theorem and its slice analogue say that near-degree-1 functions are close to unions of few stars. In the Grassmann scheme, the paper presents the known landscape: for $q \\in \\{2,3,4,5\\}$ and for $|n-2m|$ sufficiently large all degree-1 functions are trivial, but in $J_q(4,2)$ there are many exceptional families, listed with their parameters, sizes, and stabilizers, including the quadric, derived, projective, affine, and sporadic examples. The survey thereby presents all known non-trivial degree-1 behavior as concentrated in this one Grassmann setting.","pith_inferences":["Beyond the paper: the Section 5.1.1 construction's census of lines should be checked before the inventory is treated as closed, since the listed line types appear not to cover all lines of $\\mathrm{PG}(3,q)$.","The survey suggests that any q-analogue of the FKN theorem for Grassmann graphs must explicitly exclude small $(n,m)$, because $J_q(4,2)$ already contains many counterexamples to a naive stability statement.","A testable extension is to look for hypercontractivity on Grassmann graphs through their induced bilinear-forms subgraphs, since the paper notes hypercontractivity is available there but not on the Grassmann scheme itself."],"forward_implications":["If the survey's organization is right, degree-1 in the Hamming and Johnson schemes is completely understood, with FKN-type stability showing approximate degree-1 functions are close to dictators or small unions of stars.","For the Grassmann scheme, all currently known non-trivial degree-1 sets lie in $J_q(4,2)$, and for $q \\in \\{2,3,4,5\\}$ or for $|n-2m|$ sufficiently large every degree-1 function is one of the trivial examples.","The divisibility condition that a certain gcd of Gaussian coefficients divides $|Y|$ for degree-$d$ functions supplies the correct size restriction across the Johnson and $q$-Johnson schemes.","The Khot-Minzer-Safra theorem becomes a low-degree structure theorem for $J_2(n,m)$: a set with small expansion has significant weight on low-degree eigenspaces and contains a dense interval of subspaces between some $R$ and $S$."],"supporting_citations":[{"why":"Supplies the distance-regular graph background: spectra and eigenspace dimensions for Hamming, Johnson, and Grassmann graphs that the survey's degree language relies on.","marker":"[12]"},{"why":"The 1982 paper that introduced Cameron-Liebler sets as a combinatorial relaxation of orbit counts for projective groups, the core problem of the Grassmann part.","marker":"[17]"},{"why":"Delsarte's thesis provides the association-scheme formalism, the linear programming bound, and the polynomial computations used throughout.","marker":"[32]"},{"why":"The FKN theorem is the model stability result for the hypercube that the survey rephrases in eigenspace terms and generalizes to other schemes.","marker":"[60]"},{"why":"The Nisan-Szegedy junta theorem gives the benchmark degree-d classification for the hypercube that later results are measured against.","marker":"[92]"},{"why":"The Khot-Minzer-Safra theorem is the beyond-degree-1 centerpiece: small expansion in $J_2(n,m)$ forces low-degree weight and a dense subspace interval.","marker":"[79]"},{"why":"Provides the weighted design and pattern analysis that constrains degree-1 sets in $J_q(4,2)$, including the equation on which the exceptional-family analysis depends.","marker":"[65]"},{"why":"Classifies Boolean degree-1 functions on classical association schemes, including the small-$q$ Grassmann cases behind the classification theorem.","marker":"[54]"}],"fun_headline_variants":["Grassmann graphs break the triviality rule for low-degree Boolean functions","All exotic degree-1 Boolean functions live in Jq(4,2)","Degree-1 Boolean functions: trivial except in Grassmann graphs","Survey shows Hamming and Johnson trivial; Jq(4,2) is exceptional","Low-degree Boolean functions: trivial in Hamming/Johnson, exotic in Grassmann"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the survey's table of exceptional examples in the Grassmann graph $J_q(4,2)$ is complete and correct as written; if the Section 5.1.1 construction's list of line types does not actually cover all lines, the claimed inventory fails.","fun_headline_variants_meta":{"raw":{"variants":["Grassmann graphs break the triviality rule for low-degree Boolean functions","All exotic degree-1 Boolean functions live in Jq(4,2)","Degree-1 Boolean functions: trivial except in Grassmann graphs","Survey shows Hamming and Johnson trivial; Jq(4,2) is exceptional","Low-degree Boolean functions: trivial in Hamming/Johnson, exotic in Grassmann"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000531,"raw_usage":{"total_tokens":2484,"prompt_tokens":796,"completion_tokens":1688,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":412,"completion_tokens_details":{"reasoning_tokens":1588}},"tokens_in":412,"tokens_out":1688,"duration_ms":12692,"temperature":1.0,"reasoning_tokens":1588,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:17:18.813921+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Count the lines in the Section 5.1.1 construction: the four listed types total $(q^2+1)(q+2)$ lines, but $\\mathrm{PG}(3,q)$ has $(q^2+1)(q^2+q+1)$ lines, so the types do not partition the line set; checking whether an omitted line type is forced into the example would settle whether the construction stands. The table and text also disagree on whether the stabilizer is $\\mathrm{O}^+(4,q)$ or $\\mathrm{O}^-(4,q)$, and resolving that against the definition of the quadratic form would settle the inconsistency.","supporting_citations":[{"cited_title":"Friedgut, G","cited_arxiv_id":null,"evidence_quote":"The FKN theorem is the model stability result for the hypercube that the survey rephrases in eigenspace terms and generalizes to other schemes."},{"cited_title":"Nisan & M","cited_arxiv_id":null,"evidence_quote":"The Nisan-Szegedy junta theorem gives the benchmark degree-d classification for the hypercube that later results are measured against."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The Khot-Minzer-Safra theorem is the beyond-degree-1 centerpiece: small expansion in $J_2(n,m)$ forces low-degree weight and a dense subspace interval."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the weighted design and pattern analysis that constrains degree-1 sets in $J_q(4,2)$, including the equation on which the exceptional-family analysis depends."},{"cited_title":"Filmus & F","cited_arxiv_id":null,"evidence_quote":"Classifies Boolean degree-1 functions on classical association schemes, including the small-$q$ Grassmann cases behind the classification theorem."}],"review_version":1}