{"id":"a9fb859a-5afd-4167-956b-11add6de40bf","arxiv_id":"2509.01753","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every regular dihedral equiangular tight frame of redundancy 2 is genuinely projective and corresponds exactly to a 2-negacirculant skew Hadamard matrix.","lead":"It proves that equiangular tight frames with cyclic or abelian symmetry can only come from projective group actions, and gives a complete description of those with dihedral symmetry in terms of structured skew Hadamard matrices. It also shows that Paley and double Paley ETFs belong to this dihedral class and classifies all regular dihedral ETFs for sizes up to 22.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The structural theorem appears sound, but the n≤22 classification rests on an undisclosed exhaustive search; without a reproducible algorithm or code, Theorems 6.1–6.2 are not verifiable.","rationale":"The reader's weakest_assumption was that the non-regular dihedral ETF case is entirely open and thus the classification is incomplete. That is an accurate scope limitation, but it is explicitly acknowledged in the paper and does not threaten the central regular-case theorem. The reader's rationale also flagged the missing code/algorithm for the computational classification; this is the more load-bearing concern for the paper's actual classification claims. Theorems 6.1 and 6.2 depend on an unverifiable exhaustive search, and the manuscript's §6.3 only proves switching equivalence/inequivalence of the listed candidates, not exhaustiveness. The structural proof of Theorem 4.24 appears sound after checking the key steps: the Gram-block characterization, the passage from G1 to G2 via the monomial switch, the use of Corollary 4.18 to force P real, and the application of Lemma 2.2 to obtain the skew Hadamard matrix. Therefore, my concern does not move the reader's CONDITIONAL verdict; it reinforces it, and the appropriate recommendation remains conditional pending release of the search code or an independent verification.","tokens_in":29388,"tokens_out":32117,"duration_ms":342232,"concrete_test":"Independently re-run the exhaustive search: for each even n≤22, enumerate all first rows p∈{±1}^n with neg(p)+neg(p)^T=2I and all q∈{±1}^n up to the stated nega-rotation equivalence, checking neg(p)neg(p)^T+neg(q)neg(q)^T=2nI. Use a backtracking or SAT solver with a proof log to certify that n=18 has no solutions and to enumerate all solutions for n=16,20,22; compare counts and representatives with Table 1. If any additional solution appears, Theorem 6.2 is false; if an n=18 solution appears, Theorem 6.1 is false. Alternatively, the authors should release the search code and a machine-readable certificate of exhaustiveness.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The main reduction in Theorem 4.24 is internally coherent: regular dihedral ETFs are shown to be equivalent to 2-negacirculant skew Hadamard matrices, and Lemma 4.22 rules out the circulant case. The weak point is the computational classification. Section 6.2 states that an exhaustive search up to n=22 was conducted over pairs (P,Q) of negacirculant ±1 matrices satisfying P+P^T=2I and P P^T + QQ^T = 2nI, but no algorithm, pseudocode, search bounds, or code is supplied. The reduction to this pair condition is justified by Theorem 4.21 and commutativity of negacirculant matrices, and the nega-rotation quotient is plausible; however, the search space for n=22 is substantial and the text gives no certificate of exhaustiveness. The §6.3 procedure verifies switching (in)equivalence of the found candidates, so it supports pairwise inequivalence and non-Paley type for the listed examples, but it cannot prove that no further solutions were missed. Consequently Theorem 6.1 (no n=18 solution) and Theorem 6.2 (complete classification for n≤22) are not checkable from the manuscript. This does not undermine Theorem 4.24, whose proof appears sound, but it does affect a headline classification claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies group frames indexed by a group, allowing projective representations. For abelian groups, Theorem 3.2 shows that no strictly abelian ETF(N,n) exists when n(N-n)/(N-1) is not an integer, and Corollary 3.3 extends this to cyclic group frames. The main body characterizes dihedral configurations: Theorem 4.3 describes the Gram matrix form, Theorems 4.8 and 4.9 parameterize all dihedral tight frames via Artin-Wedderburn decompositions of the algebras SD_n and PD_n, and Proposition 4.15 characterizes regularity. The central structural results are Theorem 4.21, which reduces regular dihedral ETFs to skew Hadamard matrices with 2-circulant or 2-negacirculant block structure, and Theorem 4.24, which shows that every regular dihedral ETF(2n,n) is genuine projective and corresponds to a 2-negacirculant skew Hadamard matrix. The paper also proves finiteness of dihedral ETFs for fixed n, shows Paley and double Paley ETFs are genuine projective dihedral, and reports a classification for n ≤ 22, including the nonexistence of regular dihedral ETF(36,18).","tokens_in":29632,"tokens_out":9942,"duration_ms":105282,"significance":"If the computational classification is verified, the paper gives a clean, parameter-free reduction of regular dihedral ETFs of redundancy 2 to a finite combinatorial object: 2-negacirculant skew Hadamard matrices. The algebraic derivations, especially the Artin-Wedderburn decomposition of the relevant matrix algebras and the equivalence in Theorem 4.21, are elegant and appear sound. The connection to Paley and double Paley ETFs is a substantive new observation. However, the headline classification claims in Theorems 6.1 and 6.2 rest on an undisclosed exhaustive search and an informal switching-inequivalence procedure; these parts are not currently reproducible or machine-checkable. The gap is localized to Section 6 but it is load-bearing for the classification contribution.","major_comments":[{"comment":"The asserted exhaustive search up to n=22 over pairs (P,Q) of negacirculant ±1 matrices with P+P^T=2I and PP^T+QQ^T=2nI is not reproducible. No algorithm, pseudocode, pruning strategy, search-space size, or code is supplied. The reduction to this pair condition is justified, but the claim that the search was exhaustive is a computational certificate that the manuscript does not provide. Without an independent way to verify the enumeration, Theorems 6.1 and 6.2 are not checkable. Please provide a full description of the search, preferably with code or a certificate (e.g., a SAT/CP proof, or a verifiable archive).","section":"§6.2, Theorems 6.1 and 6.2"},{"comment":"The proof of pairwise switching inequivalence and the classification claims depends on claims that are not substantiated. In particular, the statement that 'the set of all pairs of indices (i,j) with i≠j has at most two orbits under the automorphism group of the Paley type matrix' is used to limit the search, but no proof of this orbit-count claim is given, and the Sage code that implements the verification is not included. The graph-isomorphism reduction is reasonable, but the final inequivalence and classification results need either a complete proof of the orbit structure or a fully reproducible computational script.","section":"§6.3, Theorem 6.2 and Table 1"},{"comment":"The abstract and introduction state that the paper gives 'a characterization of all dihedral tight frames and dihedral ETF(2n,n)'. In fact, Theorem 4.21 applies only to regular dihedral ETFs (Definition 4.12); the paper explicitly states that no example of a non-regular dihedral ETF is known and that this case remains open. The non-regular case is not characterized. Please qualify the abstract and introduction so that the claims match the theorem statements.","section":"Abstract and §1.2"}],"minor_comments":[{"comment":"The introduction says the numerical/exact classification covers 'n ≤ 20, n≠16', while Theorem 6.2 and Table 1 state n ≤ 22. Please make the range consistent.","section":"§1.2 vs §6.2"},{"comment":"The text says 'We do know that at size n=16, 20 there are dihedral ETFs, for instance those coming from the double Paley construction.' For n=16, the double Paley construction would require q=15, which is not a prime power; the Paley construction at q=31 is the relevant one. Please correct this attribution.","section":"§6.1"},{"comment":"'The proof will be given below' is followed by a computational procedure, not a mathematical proof. Please state explicitly which parts are computer-assisted and which are rigorous.","section":"§6.2 after Theorem 6.2"},{"comment":"The definition of \\hat M is typographically unclear; it should state explicitly that the exponents are the odd integers 1,3,...,2n-1.","section":"Equation (4.3)"},{"comment":"There are several typos: 'cosider' in Definition 2.3, 'the the' in Definition 2.13, 'SDnv we can generate' after Lemma 4.1, and 'sub quotient group' in Definition 2.5. Also, references [21] and [22] appear to be the same arXiv paper and should be merged or distinguished.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The structural theorems (Theorems 3.2, 4.3, 4.21, 4.24, and the Paley sections) are valuable and, to my reading, sound. The main obstacle is the computational classification: the paper advertises an exhaustive search and a complete classification for n≤22, but supplies neither an algorithm specification nor code/certificates. If the authors can provide a verifiable computational appendix or script archive, the paper would be suitable for publication. The current form is not, because the classification claims are central to the abstract and conclusion."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Solid paper, and I think the main results hold. The two things to know: Theorem 3.2 gives a clean algebraic-integer argument that strict abelian ETFs basically do not exist for redundancy 2, and Theorems 4.21/4.24 reduce regular dihedral ETFs to 2-negacirculant skew Hadamard matrices. The algebra throughout is well done: the Artin-Wedderburn decomposition gives a complete description of dihedral tight frames, Lemma 4.22 ruling out the circulant case is elegant, and the identification of Paley and double Paley ETFs as projective dihedral frames is a nice payoff.\n\nThe weaknesses are real but minor, except for one. The abstract says 'characterization of all dihedral ETF(2n,n)', but what is proved is for regular dihedral ETFs; the non-regular case remains open and the authors say so. That overstatement should be fixed. Lemma 2.2 has a typo: the display is missing the division by sqrt(2n-1). Both are easy to correct.\n\nThe substantive issue is the computational classification. Theorems 6.1 and 6.2 claim an exhaustive search up to n=22 and a complete table, but no algorithm, pseudocode, or code is given, and no certificate of exhaustiveness. The verification procedure in Section 6.3 checks pairwise switching inequivalence of the found examples, but cannot prove nothing was missed. So I would not rely on the 'no regular dihedral ETF(36,18)' claim or the 'all n≤22 are Paley type except n=16' claim without more detail. This does not damage Theorem 4.24, which is the real contribution.\n\nOne more small thing: Theorem 5.6 is stated for q prime, while the surrounding section works with prime powers; likely just a conservative statement, but inconsistent.\n\nThis paper is for frame theorists and Hadamard matrix people. It deserves a serious referee. I would ask the authors to either supply the search code or a precise algorithm, or downgrade the classification to a computational observation. Even with that caveat, the structural theorems are new, credible, and worth citing.","headline":"Clean structural results: strict abelian ETFs are ruled out, and regular dihedral ETFs reduce to negacirculant skew Hadamards; the n≤22 classification is not reproducible.","tokens_in":30204,"tokens_out":9752,"would_cite":true,"duration_ms":96465,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["42C15","52C30"],"pacs":[],"model":"deepseek-v4-flash","headline":"Regular dihedral equiangular tight frames are exactly 2-negacirculant skew Hadamard matrices, and they are always genuinely projective.","keywords":["equiangular tight frames","group frames","dihedral group","projective unitary representations","skew Hadamard matrices","Paley ETFs","Welch bound","tight frames"],"falsifier":"A single regular dihedral ETF(2n,n) that is strict, or with odd n greater than 1, would falsify Theorem 4.24; equivalently, a 2-circulant skew Hadamard matrix H=[P,Q;-Q^T,P^T] with circulant P,Q would contradict Lemma 4.22. Since Theorem 4.21 reduces the regular case to 2-negacirculant skew Hadamard matrices, an independent computer search for such matrices at n=18 already finds none; exhibiting a regular dihedral ETF(36,18), or checking any candidate against the normal form (4.12), would settle the claim.","tokens_in":1889,"feed_emoji":"📐","tokens_out":2245,"duration_ms":93027,"temperature":0.7,"pith_summary":"This paper asks which groups can produce equiangular tight frames—optimal sets of N lines in C^n that meet the Welch bound—and shows that symmetry forces genuine projectivity. For abelian groups, it proves that a strictly abelian ETF(N,n) can exist only when n(N-n)/(N-1) is an integer; in particular, cyclic frames cannot realize the redundancy-2 or maximal cases. For the dihedral group, it characterizes all dihedral tight frames and proves that regular dihedral ETF(2n,n) are precisely the frames coming from skew Hadamard matrices whose four n-by-n blocks are negacirculant, so they are always genuinely projective. Paley ETFs and their doubles are shown to be dihedral frames of exactly this type, and an exhaustive classification for n up to 22 shows almost all small examples are of Paley type. The paper leaves the existence of non-regular dihedral ETFs open, while proving only finitely many can exist for each n.","feed_headline":"Dihedral ETFs reduce to skew Hadamard matrices","feed_subtitle":"Abelian symmetry almost never reaches the Welch bound; dihedral symmetry does, via projective frames.","key_machinery":"The workhorse is the Gram matrix of a dihedral configuration, which after switching equivalence has the block form [A,B;B^T,A^T] with A,B circulant for strict representations and negacirculant for projective ones. The paper computes the Artin-Wedderburn decomposition of the two matrix algebras, converts tightness into self-adjoint rank-n idempotents in that decomposition, and then reads off equiangularity as the condition that the off-diagonal blocks come from a skew Hadamard matrix. The central identity is Theorem 4.21/4.24: a configuration is a regular dihedral ETF(2n,n) exactly when, up to switching equivalence, its Gram matrix is I_{2n} + (1/sqrt(2n-1)) [i(P-I), Q; Q^T, i(-P+I)] with P,Q","core_discovery":"The paper's central discovery is a structural dichotomy. For abelian groups, if n(N-n)/(N-1) is not an integer—which covers ETF(2n,n) and ETF(n^2,n)—no strictly abelian group frame can be equiangular tight, so any such ETF must come from a genuinely projective representation. In the dihedral case, the paper characterizes all dihedral tight frames by their Gram matrices' block structure, and proves that a regular dihedral ETF(2n,n) is equivalent to a skew Hadamard matrix H=[P,Q;-Q^T,P^T] whose four n-by-n blocks are negacirculant. In particular, regular dihedral ETFs are always genuinely projective, can occur only for even n, and the search is exactly the search for 2-negacirculant skew Hadam","pith_inferences":["If the classification is right, the bottleneck for building regular dihedral ETFs becomes the existence of 2-negacirculant skew Hadamard matrices; the Paley family supplies infinitely many, but whether all such matrices are accounted for is a separate combinatorial question.","The non-regular case is completely open: a non-regular dihedral ETF, if it exists, would lie outside the Hadamard-matrix correspondence, and the finiteness proof suggests a finite search space but no constructive route into it.","The algebraic-integer obstruction for abelian groups may extend to other nonabelian groups, suggesting that projectivity is a generic requirement for symmetry-constrained ETFs at redundancy 2.","The double Paley ETF is dihedral because a larger dihedral copy lives inside the projective line's symmetry group; this hints at a criterion for when Hadamard doubling preserves dihedral structure."],"forward_implications":["For abelian group frames, ETF(2n,n) and ETF(n^2,n) require genuinely projective representations; cyclic-group frames for such parameters do not exist.","There are no strict regular dihedral ETFs, and no regular dihedral ETFs for odd n greater than 1; regular dihedral ETFs correspond exactly to 2-negacirculant skew Hadamard matrices.","Paley ETFs and their doubles are genuine projective dihedral frames for primes q congruent to 3 modulo 4.","An exhaustive search finds no regular dihedral ETF(36,18); all regular dihedral ETFs for n up to 22 except n=16 are of Paley type, and non-Paley examples exist for n=16, 24, and 26.","For each fixed n, only finitely many dihedral ETF(2n,n) exist up to switching equivalence, but no non-regular example is known."],"supporting_citations":[{"why":"introduces 2-circulant ETFs from a fiducial-pair construction, the family this paper seeks to enrich with dihedral symmetry.","marker":"[13]"},{"why":"constructs block-circulant ETF(2n,n)s and conjectures abundance, motivating the search for more-symmetric examples.","marker":"[21]"},{"why":"supplies the known correspondence between skew-symmetric conference matrices and ETFs used to seed Lemma 2.2.","marker":"[10]"},{"why":"provides Paley's skew Hadamard matrices, which the paper shows are genuine projective dihedral ETFs.","marker":"[30]"},{"why":"gives the Hadamard doubling construction behind the double Paley ETF.","marker":"[41]"},{"why":"supplies the character-extension fact used to reduce a strictly abelian frame to a quotient group in Lemma 3.1.","marker":"[42]"},{"why":"provides the graph-isomorphism algorithm used to decide switching equivalence in the classification.","marker":"[28]"},{"why":"implements the graph-isomorphism engine used for the exhaustive n up to 22 classification.","marker":"[36]"}],"fun_headline_variants":["Abelian groups rarely admit equiangular tight frames","Dihedral equiangular tight frames require projective reps","Dihedral ETFs equivalent to skew Hadamard matrices","Abelian symmetry almost never yields equiangular tight frames","Regular dihedral ETFs are exactly skew Hadamard"],"cache_read_input_tokens":31872,"weakest_assumption_plain":"The classification of dihedral ETFs assumes regularity—that the first n orbit vectors form a basis of C^n—and the paper knows no non-regular dihedral ETF, so a non-regular example would lie outside the theorem's reach.","fun_headline_variants_meta":{"raw":{"variants":["Abelian groups rarely admit equiangular tight frames","Dihedral equiangular tight frames require projective reps","Dihedral ETFs equivalent to skew Hadamard matrices","Abelian symmetry almost never yields equiangular tight frames","Regular dihedral ETFs are exactly skew Hadamard"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000194,"raw_usage":{"total_tokens":1175,"prompt_tokens":714,"completion_tokens":461,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":458,"completion_tokens_details":{"reasoning_tokens":384}},"tokens_in":458,"tokens_out":461,"duration_ms":5026,"temperature":1.0,"reasoning_tokens":384,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T12:14:21.623591+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A single regular dihedral ETF(2n,n) that is strict, or with odd n greater than 1, would falsify Theorem 4.24; equivalently, a 2-circulant skew Hadamard matrix H=[P,Q;-Q^T,P^T] with circulant P,Q would contradict Lemma 4.22. Since Theorem 4.21 reduces the regular case to 2-negacirculant skew Hadamard matrices, an independent computer search for such matrices at n=18 already finds none; exhibiting a regular dihedral ETF(36,18), or checking any candidate against the normal form (4.12), would settle the claim.","supporting_citations":[{"cited_title":"On the optimal arrangement of $2d$ lines in $\\mathbb{C}^d$","cited_arxiv_id":"2312.09975","evidence_quote":"introduces 2-circulant ETFs from a fiducial-pair construction, the family this paper seeks to enrich with dihedral symmetry."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the known correspondence between skew-symmetric conference matrices and ETFs used to seed Lemma 2.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides Paley's skew Hadamard matrices, which the paper shows are genuine projective dihedral ETFs."},{"cited_title":"2, 203–204","cited_arxiv_id":null,"evidence_quote":"gives the Hadamard doubling construction behind the double Paley ETF."},{"cited_title":"Weibel, An introduction to homological algebra , Cambridge University Press, April 1994","cited_arxiv_id":null,"evidence_quote":"supplies the character-extension fact used to reduce a strictly abelian frame to a quotient group in Lemma 3.1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the graph-isomorphism algorithm used to decide switching equivalence in the classification."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"implements the graph-isomorphism engine used for the exhaustive n up to 22 classification."}],"review_version":1}