{"id":"42c5f02b-ccb2-446a-bc7b-c9fe4359a018","arxiv_id":"2412.20795","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A new projection-based symmetry bootstrap shows that almost commuting operators can be approximated by commuting operators preserving reflection, rotational, and dihedral symmetries, resolving a conjecture for two-matrix topological insulator classes.","lead":"This mathematics paper proves that almost commuting matrices can be replaced by exactly commuting matrices that also preserve symmetries such as realness, transpose symmetry, antisymmetry, or finite-order rotational phase structure. The result resolves a conjecture about the symmetry classes used for topological insulators and introduces a reusable 'symmetry bootstrap' method.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Rotational/dihedral Lin theorem is proved only for invertible elements; the non-invertible extension requires an approximation by invertible phase-symmetric elements that can fail when the phase order does not divide the matrix dimension.","rationale":"The reader's weakest assumption identifies the same load-bearing concern as I do: Theorem 8.2 is conditional on an invertible approximation that is not generally available for non-invertible phase-symmetric matrices. This concern is concrete and demonstrable, but it does not break the main order-2 bootstrap or the resolution of the AZ two-matrix conjecture. Theorems 7.1, 7.3, and 7.4 supply a self-contained argument for order-2 symmetries, and the projection characterization in Sections 5 and 6 appears internally consistent. I also checked the proof of Theorem 7.3: modulo an editorial inconsistency in the displayed statement of Theorem 7.1 (the proof cites 'Theorem 7.1(i)' while the statement shown has no case labels), the argument is sound. The advertised rotational/dihedral extension is therefore less complete than the abstract suggests, but the central claims for reflection-type and order-2 symmetries remain supported. A conditional verdict with a flagged limitation is appropriate; no verdict change is needed from the reader's assessment.","tokens_in":54993,"tokens_out":22561,"duration_ms":228923,"concrete_test":"For W = diag(1, e^{2π i/3}) acting on C^2, solve the linear system W* A W = e^{2π i/3} A and check whether any solution has det A ≠ 0. The solution space is {[[0,a],[0,0]] : a ∈ C}, all of determinant zero, so no invertible 3-phase-symmetric matrix exists. This directly refutes the approximation hypothesis in the final sentence of Theorem 8.2 and shows the non-invertible rotational statement is not proved by the paper in this case. If a different non-invertible argument is intended, it must be supplied separately.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing gap is in Theorem 8.2. The theorem is proved only for invertible A, and the final sentence extends it to a non-invertible A0 only if A0 is approximable by invertible elements satisfying the same phase symmetries with controlled self-commutator. That approximation is not established, and it is false in finite dimensions when the phase order does not divide the matrix dimension. For example, with W = diag(1, e^{2π i/3}) on C^2, the equation W* A W = e^{2π i/3} A forces A to have the form [[0,a],[0,0]]; every such A is nilpotent, so the set of invertible 3-phase-symmetric matrices is empty. Thus no invertible approximant A can exist for nonzero A0, and the conditional extension in Theorem 8.2 does not apply. This is the finite-dimensional analogue of the odd-n antisymmetry obstruction noted in Section 3. Consequently, the abstract's claim that rotational and dihedral Lin theorems are proved for almost normal matrices is not supported for non-invertible contractions; the paper proves an invertible-element theorem plus a conditional statement. The order-2 symmetry bootstrap (Theorems 7.3/7.4) and the AZ two-matrix corollary do not rely on this gap.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a general framework of symmetry maps on unital C*-algebras and uses a projection characterization of pairs (U,B) that are near commuting pairs to \"bootstrap\" symmetries from an initial nearby commuting pair to a final commuting pair. The main applications are a self-adjoint symmetry bootstrap with explicit constants (Theorem 7.1), Lin's theorem with linear symmetries and with an order-2 conjugate-linear symmetry or linear antisymmetry (Theorems 4.5, 7.3, 7.4), corollaries for two and three almost commuting self-adjoint matrices, a resolution of Conjecture 2.2 of [21] for the Altland-Zirnbauer classes, and a rotational/dihedral version of Lin's theorem in Section 8.","tokens_in":55169,"tokens_out":12485,"duration_ms":131129,"significance":"If the central results stand, the paper gives a genuinely new and useful method: the projection characterization converts an initial nearby commuting pair into symmetry-preserving commuting approximants, and the self-adjoint case is worked out with explicit constants. The proof is not circular: it uses external Lin-type theorems to obtain the initial commuting pair and then constructs the symmetric pair from projections. The resolution of the two-matrix conjecture from [21] is a substantial achievement. However, the Section 8 rotational/dihedral Lin theorem is proved only for invertible elements, and the advertised non-invertible extension rests on an unproved and sometimes false approximation assumption, so that part of the abstract is not supported as stated.","major_comments":[{"comment":"The advertised rotational/dihedral Lin theorem is proved only for invertible A. The last sentence of Theorem 8.2 extends it to a non-invertible A0 only under the extra assumption that A0 can be approximated by invertible elements A satisfying the same hypotheses. This approximation is not established, and the paper itself notes the obstruction: Section 3, in the paragraph after Proposition 2.14, explains that for a phase symmetry of order n an invertible approximant may fail to exist in finite dimensions when the order of the phase does not divide the matrix dimension. Concretely, for W = diag(1, e^{2πi/3}) on C^2, the condition W*AW = e^{2πi/3}A forces A to have the form [[0,a],[0,0]], so every such A is nilpotent and no nonzero A0 has an invertible approximant with this phase symmetry. Consequently the abstract's claim of a rotational/dihedral Lin theorem for almost normal matrices is not supported for non-invertible contractions; the paper proves an invertible-element statement plus a conditional assertion. This is load-bearing because Theorem 8.2 is one of the paper's advertised main results.","section":"Theorem 8.2 and final paragraph; cf. Section 3"},{"comment":"Theorem 7.6 is stated as \"for ||[U,B]|| small enough\" and with an unspecified universal constant \"Const.\", while cases (ii) and (iii) say that the constant and the smallness condition depend on n without quantifying either. The proof chooses L = sqrt(Const * eps) and requires L to lie below an L0, but L0 is never computed or bounded, so the theorem does not provide a threshold or a rate. Since Theorem 8.2 invokes Theorem 7.6 to obtain its \"Const.\" bound, the quantitative claim in the rotational/dihedral theorem inherits this unquantified status. The statement also calls Const. universal while later making it depend on n, which should be reconciled. If the intended statement is merely an existence result with some rate, it should be reformulated as such; as written, it gives the appearance of a quantitative theorem without the required data.","section":"Theorem 7.6 and its use in Theorem 8.2"}],"minor_comments":[{"comment":"\"Atland-Zirnbauer\" should be \"Altland-Zirnbauer\" throughout.","section":"Abstract, Introduction, Section 8"},{"comment":"The displayed condition \"F = F- + E_{Omega_0}(U), + F+\" contains a stray comma; it should read \"F = F- + E_{Omega_0}(U) + F+\".","section":"Lemma 6.4, condition 2"},{"comment":"The phrase describing Const. as universal should be reconciled with the later statement that the constant depends on n in cases (ii) and (iii), since both appear in the same theorem statement.","section":"Theorem 7.6"}],"recommendation":"major_revision","confidential_remarks":"The main obstacle is Section 8: the non-invertible rotational/dihedral Lin theorem is conditional on an approximation hypothesis that can fail in finite dimensions. The order-2 self-adjoint bootstrap, Theorem 7.3/7.4, and the AZ two-matrix corollary do not rely on this gap and appear sound. If the approximation statement cannot be proved, the authors should either add explicit hypotheses under which it holds or restrict the abstract and Section 8 to invertible elements. The paper would still be substantial, but the advertised claims need to match what is actually proved."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the self-adjoint symmetry bootstrap is a real new tool, and the AZ two-matrix conjecture is settled. But the advertised rotational/dihedral Lin theorem is only proved for invertible elements; the extension to non-invertibles rests on an approximation condition that fails, for example, for order-3 phase on C^2. The abstract overstates this.\n\nWhat's actually new: the projection characterization (Lemmas 6.1/6.4, 6.8/6.9) turns near commutation with a unitary/self-adjoint into existence of almost reducing, symmetry-respecting projections; then the bootstrap uses a symmetry-blind commuting pair from external Lin-type theorems to construct a symmetric pair. The self-adjoint version has explicit constants; Theorem 7.3 gives a 90 sqrt(eps) bound, and Theorem 1.9 resolves Conjecture 2.2 of [21] for all ten AZ classes for two self-adjoint matrices. That's a genuine advance, and the proofs are worked out in enough detail that the argument appears sound. The citation pattern looks fair; the reductions in Section 3 are useful.\n\nThe soft spots are where the reader said they are. Theorem 7.6 leaves 'Const.' and the smallness threshold on ||[U,B]|| completely unspecified, so the unitary bootstrap is a qualitative statement until someone fills in the arithmetic. More serious, Theorem 8.2 is proved for invertible A, and the final sentence extends to non-invertible A0 only under an approximation hypothesis. The stress-test example is correct: for W = diag(1, e^{2πi/3}) on C^2, the equation W* A W = e^{2πi/3} A forces A = [[0,a],[0,0]], so all such matrices are nilpotent; there are no invertible approximants with that phase. So the non-invertible rotational/dihedral case is not proven, and the abstract's claim needs scaling back. The paper itself states the condition, so I don't call it dishonest — but the advertised theorem is narrower than the abstract suggests.\n\nThe order-2 symmetry results and the AZ corollary don't touch this gap; they stand. If I were refereeing, I would ask for the non-invertible gap to be either closed or explicitly stated as open, and for the constants in Theorem 7.6 to be made explicit or the theorem flagged as a qualitative statement. But the core bootstrap and the AZ resolution are solid enough to deserve full referee time.\n\nRecommendation: send to review, likely after minor-to-moderate revision. The paper is for people working on Lin-type theorems, almost commuting matrices with symmetries, and topological insulator classification; they'll want this.","headline":"Solid new bootstrap for order-2 symmetries, but the rotational/dihedral Lin theorem is only proved for invertible elements and the abstract oversells the non-invertible extension.","tokens_in":55772,"tokens_out":2915,"would_cite":true,"duration_ms":28338,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["46L10","46L55","15A27"],"pacs":[],"model":"deepseek-v4-flash","headline":"Symmetry maps can be required of nearby commuting approximants, proving Lin's theorem with reflection, rotational, and dihedral symmetries.","keywords":["Lin's theorem","symmetry bootstrap","almost commuting matrices","von Neumann algebras","projection characterization","phase symmetries","Altland-Zirnbauer classes"],"falsifier":"Take a finite-dimensional matrix $A_0$ with a phase symmetry of order $q>2$ for which $q$ does not divide the matrix dimension $n$; any phase-symmetric matrix in that class has zero as a forced eigenvalue, so $A_0$ cannot be approximated by invertible phase-symmetric matrices with controlled self-commutator. Exhibiting such an $A_0$ with arbitrarily small $\\|[A_0^*,A_0]\\|$ but no nearby invertible symmetric approximant would disprove the non-invertible version of Theorem 8.2.","tokens_in":54704,"feed_emoji":"🔄","tokens_out":7218,"duration_ms":62076,"temperature":0.7,"pith_summary":"This paper establishes that symmetries of an almost normal operator—transpose, conjugation, reflection, rotation, and dihedral phase symmetries—can be forced onto the nearby normal operator whose existence is asserted by Lin's theorem. The core tool is a projection characterization: from a pair of operators that are already close to commuting, the paper extracts spectral projections of the nearby commuting pair, perturbs them into projections subordinate to the original operator, and uses those projections to rebuild a commuting pair that carries the original symmetries. For finite matrices this yields symmetry-preserving versions of Lin's theorem, including reflection and finite rotational/dihedral phase symmetries for invertible elements, and resolves a conjecture about two almost commuting self-adjoint matrices in the Altland–Zirnbauer symmetry classes. The paper also proves bootstrap symmetry results for two and three almost commuting self-adjoint operators.","feed_headline":"Almost normal matrices keep their symmetries","feed_subtitle":"Projection proof secures Lin's theorem with reflection and rotation symmetries, settling a topological-insulator conjecture.","key_machinery":"The projection characterization: if $U,B$ are close to commuting $U',B'$, then for arcs $\\Omega_0 \\subset \\Omega$ in the spectrum circle there is a projection $F$ satisfying $E_{\\Omega_0}(U)\\le F\\le E_\\Omega(U)$, decomposable as $F_- + E_{\\Omega_0}(U) + F_+$ with $F_\\pm$ subordinate to the complementary subarcs, and with $[F,B]$ small. These projections are assembled into a spectral resolution defining $U''$ and a localization/pinching operation $B\\mapsto B''$ that commutes with the symmetries. Symmetry maps—$\\mathbb{R}$-linear hermitian maps that are (anti-)multiplicative and linear or conjugate-linear—are the formal language in which 'transpose', 'conjugation', 'reflection', and 'phase rotation' symmetries are expressed.","core_discovery":"The central claim is that the existence of nearby commuting approximants $U'$, $B'$ (with $U'$ unitary or self-adjoint) is characterized by the existence of a family of projections $F$ that sit between spectral projections of $U$ and almost commute with $B$; this characterization allows the approximants to be rebuilt as $U''$, $B''$ that inherit any symmetries of $U$ and $B$, provided the symmetries are generated by weakly continuous linear, conjugate-linear, multiplicative or anti-multiplicative maps satisfying mild commuting conditions. As a corollary, the paper proves Lin's theorem with symmetries: an $S$-symmetric almost normal contraction is within distance $O(\\sqrt{\\epsilon_\\ell(\\|[A^*,A]\\|)})$ of a normal $S$-symmetric element, and with a linear antisymmetry or conjugate-linear symmetry of order two. For a unitary operator with reflection, rotational, or dihedral phase symmetries, the same bootstrap yields nearby commuting unitary and self-adjoint operators preserving those phase symmetries. For invertible elements, the rotational and dihedral versions extend to Lin's theorem directly through a symmetry-respecting polar decomposition lemma.","pith_inferences":["The projection/rebuilding method is likely to transfer to other approximation problems where the spectrum lies on a one-dimensional set, yielding symmetry-preserving approximants for almost commuting pairs of unitaries or for normal operators with more general spectral curves.","For non-invertible elements the rotational/dihedral Lin theorem is not proven; if the required invertible approximation fails (e.g., phase order not dividing matrix dimension), the non-invertible statement would be false as stated.","The bootstrap may give a quantitative route to index-theoretic obstructions, since the constructed projections encode the same almost-invariant subspace data used in counterexamples like Voiculescu's unitaries.","The automatic symmetry results for three almost commuting matrices suggest analogous automatic-symmetry phenomena for larger families when the only obstacle is the lack of commutants."],"forward_implications":["For matrices, any almost normal matrix with a transpose, conjugation, or reflection-type symmetry has a nearby normal matrix with the same symmetry (Lin's theorem with symmetries).","For two almost commuting self-adjoint matrices, the bootstrap preserves arbitrary collections of commuting unitary/anti-unitary symmetries, including all ten Altland–Zirnbauer classes in 1D, resolving Conjecture 2.2 of [21].","For three almost commuting self-adjoint matrices that are already near commuting ones, certain symmetry patterns (one or two antisymmetries) are automatic for the nearby commuting triple, without needing an extra index to vanish.","For unitaries with rotational phase symmetry of finite order $n$ or dihedral symmetry, the error bound must depend on $n$; the paper exhibits a scaling-invariant example showing this dependence is necessary."],"supporting_citations":[{"why":"Lin's theorem that an almost normal contraction is near a normal matrix; the base result being generalized.","marker":"[19]"},{"why":"Loring and Sørensen's Lin-theorem-with-symmetry results for the real and self-dual cases; the single-order-2 anti-multiplicative case the paper reduces to.","marker":"[24]"},{"why":"Kachkovskiy–Safarov's explicit square-root bound for distance to normal elements; used for constants and the self-adjoint bootstrap.","marker":"[18]"},{"why":"Friis–Rørdam's short proof of Lin's theorem for stable-rank-one C*-algebras; supplies the no-symmetry base case.","marker":"[11]"},{"why":"Loring's conjecture on K-theory and pseudospectra for topological insulators; the conjecture the paper resolves.","marker":"[21]"},{"why":"Davidson's projection reformulation of almost commuting Hermitian matrices; the source of the projection method.","marker":"[7]"},{"why":"Voiculescu's construction of asymptotically commuting unitaries without commuting approximants; supplies the filtration technique and the necessity of condition (ii).","marker":"[31]"}],"fun_headline_variants":["Almost normal matrices keep their symmetries","Projection method bootstraps symmetries in commuting approximants","Lin's theorem with symmetries settled by projection characterization","Symmetries survive approximate commutation via projections","New proof: symmetric almost normal matrices have symmetric normals"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The advertised rotational and dihedral Lin theorem is proved only for invertible elements; extending it to arbitrary non-invertible elements requires that those elements be approximable by invertible elements with the same phase symmetries and controlled self-commutator, a condition that can fail in finite dimensions (for example, when the order of the phase does not divide the matrix dimension).","fun_headline_variants_meta":{"raw":{"variants":["Almost normal matrices keep their symmetries","Projection method bootstraps symmetries in commuting approximants","Lin's theorem with symmetries settled by projection characterization","Symmetries survive approximate commutation via projections","New proof: symmetric almost normal matrices have symmetric normals"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000362,"raw_usage":{"total_tokens":2082,"prompt_tokens":1202,"completion_tokens":880,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":818,"completion_tokens_details":{"reasoning_tokens":805}},"tokens_in":818,"tokens_out":880,"duration_ms":7389,"temperature":1.0,"reasoning_tokens":805,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T23:14:23.398359+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a finite-dimensional matrix $A_0$ with a phase symmetry of order $q>2$ for which $q$ does not divide the matrix dimension $n$; any phase-symmetric matrix in that class has zero as a forced eigenvalue, so $A_0$ cannot be approximated by invertible phase-symmetric matrices with controlled self-commutator. Exhibiting such an $A_0$ with arbitrarily small $\\|[A_0^*,A_0]\\|$ but no nearby invertible symmetric approximant would disprove the non-invertible version of Theorem 8.2.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Lin's theorem that an almost normal contraction is near a normal matrix; the base result being generalized."},{"cited_title":"07, 1650017","cited_arxiv_id":null,"evidence_quote":"Loring and Sørensen's Lin-theorem-with-symmetry results for the real and self-dual cases; the single-order-2 anti-multiplicative case the paper reduces to."},{"cited_title":"1, 61–80","cited_arxiv_id":null,"evidence_quote":"Kachkovskiy–Safarov's explicit square-root bound for distance to normal elements; used for constants and the self-adjoint bootstrap."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Friis–Rørdam's short proof of Lin's theorem for stable-rank-one C*-algebras; supplies the no-symmetry base case."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Loring's conjecture on K-theory and pseudospectra for topological insulators; the conjecture the paper resolves."},{"cited_title":"2, 222–240","cited_arxiv_id":null,"evidence_quote":"Davidson's projection reformulation of almost commuting Hermitian matrices; the source of the projection method."},{"cited_title":"Math.(Szeged) 45 (1983), no","cited_arxiv_id":null,"evidence_quote":"Voiculescu's construction of asymptotically commuting unitaries without commuting approximants; supplies the filtration technique and the necessity of condition (ii)."}],"review_version":1}