{"id":"567001e5-3521-4298-beb1-2f7920d295e6","arxiv_id":"2502.00505","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors derive closed-form moment formulas and limiting spectral densities for anticommutators of GOE, palindromic Toeplitz, block circulant, and checkerboard ensembles, including multi-scale blip behavior.","lead":"The paper derives exact eigenvalue statistics for anticommutators AB+BA of structured random matrix families such as GOE, palindromic Toeplitz, block circulant, and checkerboard matrices. It also finds that checkerboard anticommutators split their spectra into several well-separated scales, with explicit moment formulas for the largest blips.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The blip regime theorems rest on an explicitly empirical, unproved input: the exact locations and multiplicities of the non-bulk eigenvalues asserted in Appendix B; Theorems 1.18 and 1.19 are not supported without a derivation or citation for that input.","rationale":"The paper's strongest contributions are the Section 2 combinatorial moment calculations for {GOE, GOE}, {PTE, PTE}, and {GOE, PTE}; these appear carefully derived and are supported by generating functions, OEIS identification, and consistency with known free-probability results. The central problem is the checkerboard blip analysis. The abstract states that the spectrum of {GOE, k-checkerboard} consists of a bulk plus a blip at \\pm N^{3/2}/k, and that {k-checkerboard, j-checkerboard} has two intermediary blips and one largest blip at 2N^2/(kj). The proofs of Theorems 1.18 and 1.19, which give the limiting blip moments, depend on this regime structure through the weight functions and normalizations. Yet the regime structure is not proved: Appendix B explicitly bases Lemma B.5 on an observed spectral-location assumption, and Lemma B.7 supplements it with additional observed spectra and with the unproved statement that borderline eigenvalues fall into the regimes indicated by their Weyl bounds. This is not merely a missing citation; without the exact locations and multiplicities, the cancellations used in Section 3 could be computing moments supported on the wrong set of eigenvalues. The reader's weakest-assumption diagnosis is exactly this point, and I agree with it. The Appendix E mismatch is corroborating rather than the primary concern: it suggests that at least one part of the computational results is unreliable, but the decisive logical gap remains the unproved blip locations. Since the central claim as stated is not fully supported, the REJECT verdict is appropriate; the paper would need either a proof of the spectral-location input or a clear separation of the blip results as conjectural rather than established.","tokens_in":47059,"tokens_out":5263,"duration_ms":58957,"concrete_test":"Replace the empirical input with a derivation. Conjugate the k-checkerboard mean matrix M_N by the permutation that groups indices by residue modulo k; then M_N becomes diag(J_{N/k}, ..., J_{N/k}). By orthogonal invariance, B_N remains GOE, so the nonzero part of {M_N, B_N} is a k-by-k block matrix with (r,s)-block J B_{rs} + B_{rs} J. Compute the eigenvalues of this low-rank block Gaussian matrix exactly; they are determined by row and column sums and should be k eigenvalues at +N^{3/2}/k + O(N) and k at -N^{3/2}/k + O(N). If this calculation gives different centers, multiplicities, or fluctuation scales, then the normalization in Theorem 1.18 is wrong. Run the analogous block calculation for the mean matrices in {k-checkerboard, j-checkerboard} and for the mixed mean/perturbation terms in Lemma B.7 to verify the claimed counts in Theorem 1.19.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Appendix B is titled 'Proof of Multiple Regimes', but its key spectral input is not proved. Before Lemma B.5 the paper states: 'Empirically, we observe that AN has k blip eigenvalues at N^{3/2}/k + O(N) and k blip eigenvalues at -N^{3/2}/k + O(N). By assuming this, we are able to prove the existence of multiple regimes...'. Lemma B.5 then turns this assumption into the containment claims used for {GOE, k-checkerboard}. For {k-checkerboard, j-checkerboard}, Lemma B.7 similarly relies on 'empirically' observed spectra of {A, \\tilde{B}} and {\\tilde{A}, B}, and on the unproved assertion that the four borderline eigenvalues fall in the regimes of their lower or upper bounds; the note after Lemma B.7 admits that Weyl's inequality only gives intervals. The weight functions in Definitions 3.1 and 3.2, and the trace expansions (3.3)-(3.4), are centered at exactly these empirical locations (\\pm N^{3/2}/k and 2N^2/(kj)). If the actual centers or counts differ, the cancellation in the proofs of Theorems 3.17/1.18 and 3.20/1.19 isolates the wrong eigenvalue population, so the claimed limiting blip moments are not those of the blip. A separate warning comes from Appendix E: Table 4 lists the theoretical fourth moment of {2-BCE, 2-BCE} as 5.52 against an empirical value of 10.19, with no explanation, indicating that the computational pipeline around the genus-expansion claims is not internally consistent as presented.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the limiting spectral distribution of anticommutators {A,B}=AB+BA of real symmetric random matrix ensembles. For {GOE,GOE}, {PTE,PTE}, and {GOE,PTE}, Section 2 gives closed-form moment formulas or recurrences: Theorem 1.9/2.10 gives M_{2m}=(1/m)\\sum_{k=1}^m 2^k \\binom{2m}{k-1}\\binom{m}{k}; Theorem 1.11/2.12 gives M_{2m}=4^m((2m-1)!!)^2; Theorem 1.13/2.23 gives a recurrence for {GOE,PTE}. For {GOE,k-BCE} and {k-BCE,k-BCE}, Theorem 1.15/2.28 gives genus-expansion formulas. Section 3 introduces weighted empirical blip spectral measures for the anticommutators involving checkerboard ensembles and claims limiting blip moments in Theorems 1.18/3.17 and 1.19/3.20. Appendix B aims to prove the multiple-regime structure, Appendix C gives convergence statements, Appendix D gives explicit weight functions, and Appendix E reports numerical lower moments.","tokens_in":47393,"tokens_out":5393,"duration_ms":54864,"significance":"If the main results hold, the closed-form moment formulas and densities for {GOE,GOE} and {PTE,PTE} are useful additions to the random-matrix literature, and the blip moment formulas for checkerboard anticommutators would provide new benchmarks for non-bulk spectral regimes. The combinatorial machinery in Section 2 is coherent and self-contained: the Wick expansion, the non-crossing matching arguments, the recurrence for {GOE,PTE}, and the identification of the {GOE,GOE} generating function with the 3-Schr\\\"oder numbers are all presented in a verifiable way. The convolution representation of the {PTE,PTE} density is elegant. However, the blip theorems rest on an explicitly empirical spectral-location input in Appendix B, and Appendix E contains an unexplained numerical discrepancy; these issues are load-bearing for the paper's central new claims.","major_comments":[{"comment":"","section":"Appendix B, Lemmas B.5 and B.7; Theorems 1.18/3.17 and 1.19/3.20"},{"comment":"","section":"Lemma B.7(1); Definition 3.2 and equation (3.4)"},{"comment":"","section":"Appendix E, Table 4"}],"minor_comments":[{"comment":"","section":"Section 1.3, the paragraph after Figure 8"},{"comment":"","section":"Section 3.1, displayed line after equation (3.18)"},{"comment":"","section":"Bibliography reference [NR]"},{"comment":"","section":"Appendix D, equation (D.1)"},{"comment":"","section":"Section 2, theorem numbering"}],"recommendation":"major_revision","confidential_remarks":"The Section 2 results may be publishable on their own, and the combinatorial derivations there appear sound. My concern is confined to the Section 3 blip results, whose proof relies on an unverified empirical spectral-location input, and to the unexplained numerical discrepancy in Appendix E. These are fixable within the manuscript's scope, so I am not recommending rejection, but the current version does not meet the standard for acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know before you read it. The moment computations for anticommutators of GOE, PTE, and the mixed ensemble are real work: the recurrences in Section 2 are carefully built on Wick expansion and known matching rules, and I believe they are correct. Second, the headline checkerboard blip theorems are conditional: Appendix B explicitly assumes the locations and multiplicities of the blip eigenvalues that the weight functions are centered on. That is not a proof of multiple regimes, and Section 1.3 overstates what is proved.\n\nWhat is genuinely new: the GOE/PTE recurrence and the blip moment formulas themselves. The GOE/GOE density overlaps with the free-probability results cited as [NR] and [Vas], but the paper gives an independent combinatorial derivation, which has value. The PTE/PTE result is a short consequence of free matching, but the derivation is clean and the chi-square convolution characterization is a nice touch.\n\nThe decisive soft spot is the empirical input in Appendix B. Lemma B.5 states: \"Empirically, we observe that A_N has k blip eigenvalues at N^{3/2}/k + O(N)\" and then proceeds by assuming this. The weight functions in Definitions 3.1 and 3.2 are centered at exactly those locations, and the cancellation arguments in Theorems 3.17 and 3.20 extract moments only if the eigenvalues are where the weight function puts mass. If the true centers or multiplicities differ by more than the stated errors, the claimed limiting blip moments are not the moments of the blip. Lemma B.7 similarly relies on observed spectra for mixed mean/perturbation anticommutators. This is fixable if a proof or citation exists, but without it Theorems 1.18 and 1.19 are conditional, not established.\n\nA second issue, smaller but real: Appendix E reports a 5.52 theoretical fourth moment for {2-BCE, 2-BCE} against an empirical 10.19, with no explanation. That kind of discrepancy suggests either an error in the genus expansion formulas or a misnormalized simulation; the paper should address it before publication.\n\nWho benefits: researchers working on structured random matrices and moment methods. The combinatorial machinery here is worth having, and the checkerboard blip problem is a good open challenge. But as it stands, I would not cite the blip theorems as proven, and I would not claim existence of the regimes based on this paper.\n\nMy recommendation: send it to a serious referee, asking them to focus on whether the blip-location input can be proved or replaced by a citation, and to check the BCE numerics. It is not ready for acceptance as is, but the core Section 2 material is solid enough that a major revision could make this a useful contribution.","headline":"A genuinely useful combinatorial paper whose headline checkerboard blip theorems are conditional on an explicit but unproved spectral-location assumption, plus an unexplained numerical mismatch in Appendix E.","tokens_in":47971,"tokens_out":2678,"would_cite":false,"duration_ms":27879,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60B20","15B52"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper derives exact limiting spectral laws for anticommutators of structured random matrix ensembles, including split-spectrum blip moments for checkerboard pairs.","keywords":["random matrix theory","anticommutator","limiting spectral distribution","Gaussian orthogonal ensemble","palindromic Toeplitz ensemble","checkerboard ensemble","block circulant ensemble","blip regime"],"falsifier":"For moderately large $N$ with $k \\mid N$ and $\\gcd(k,j)=1$, $jk \\mid N$, diagonalize $\\{A_N,B_N\\}$ for GOE/$k$-checkerboard and $k$-checkerboard/$j$-checkerboard and count eigenvalues in windows $[N^{3/2}/k - CN, N^{3/2}/k + CN]$ and $[2N^2/(jk)-CN^{3/2}, 2N^2/(jk)+CN^{3/2}]$; the central claim fails if the counts differ from $k$ per sign and $1$, respectively, or if the empirically weighted blip moments do not approach the formulas in Theorems 1.18 and 1.19.","tokens_in":46845,"feed_emoji":"📊","tokens_out":8205,"duration_ms":73997,"temperature":0.7,"pith_summary":"The paper studies the anticommutator {A,B}=AB+BA of two independent real symmetric random matrices and asks how the eigenvalue distribution changes when the inputs carry extra symmetry. For Gaussian orthogonal ensemble pairs it obtains exact even moments, M_{2m} = (1/m)\\sum_{k=1}^m 2^k \\binom{2m}{k-1}\\binom{m}{k}, plus an explicit algebraic limiting density; for palindromic Toeplitz pairs it gets M_{2m}=4^m((2m-1)!!)^2, whose density is the convolution of a chi-squared and a negated chi-squared variable. When a checkerboard ensemble is involved, the spectrum splits into regimes of different scales, and the paper isolates the largest outlier regime with a polynomial weight function to get closed-form moment formulas. The payoff is a rare set of explicit benchmarks for structured random matrices, where the usual free-probability anticommutator recipe becomes intractable.","feed_headline":"Random-matrix anticommutators get exact spectral formulas","feed_subtitle":"New combinatorial proofs give closed-form moments and isolate outlier blip eigenvalues at N^{3/2} and N^2 scales.","key_machinery":"The load-bearing machinery is the moment method coupled to matching combinatorics. Wick's formula turns expected traces into sums over pairings of matrix entries; for GOE only non-crossing pairings survive in the limit via the genus bound $\\#(\\gamma_{2m}\\pi)\\le m-1$ unless $\\pi$ is non-crossing, for PTE essentially all pairings survive because of its palindromic structure, and for mixed GOE/PTE a layer decomposition restricts the PTE terms to stay inside layers created by non-crossing GOE matchings. The blip results use the polynomial weight function $f^{(2n)}(x)=x^{2n}(2-x)^{2n}$ with $n=\\log\\log N$, which is close to 1 at a blip location and decays rapidly elsewhere; expanding the weight and applying binomial identities such as $\\sum_i (-1)^i\\binom{m}{i}i^p=0$ for $p<m$ cancels all lower-order contributions and leaves the closed-form moments.","core_discovery":"The central discovery is that the anticommutator operation preserves enough of the input ensembles' combinatorial structure to yield exact limiting spectral information. For {GOE,GOE} the limiting even moments are the 3-Schr\\\"oder numbers, and the density has the closed algebraic form given in Corollary 1.10; for {PTE,PTE} the moments factor as 4^m((2m-1)!!)^2 and the density is the convolution of the densities of $\\chi_1^2$ and $-\\chi_1^2$; for {GOE,PTE} a two-variable recurrence $\\sigma_{n,s}=\\sum_{k=1}^n (\\sigma_{k-1,1}\\sigma_{n-k,s}+\\sigma_{k-1,0}\\sigma_{n-k,s+1})$ governs the moments. When a checkerboard ensemble is involved, the spectrum splits: {GOE,$k$-checkerboard} has a bulk of size $\\Theta(N)$ plus a blip of $2k$ eigenvalues near $\\pm N^{3/2}/k$, and the weighted blip moments converge to $(1/k)(2/k^2)^m \\mathbb{E}_k[\\operatorname{Tr} C_k^m]$, where $C_k$ is a $k\\times k$ hollow GOE. For {$k$-checkerboard,$j$-checkerboard} there is a largest blip near $2N^2/(jk)$ whose moments have the explicit multinomial formula of Theorem 1.19.","pith_inferences":["If the Appendix B spectral-location input were proved rather than observed, the weight-function method would upgrade from conditional to unconditional; a promising route is a deterministic equivalent for the mean-matrix anticommutator.","The same weighted-moment cancellation should apply to other split-limiting ensembles with several outlier regimes, provided the outlier locations are known to the correct order; the paper's block taxonomy of 1-blocks and 2-blocks may generalize.","The appearance of 3-Schr\\\"oder numbers in {GOE,GOE} hints at a lattice-path or walk interpretation of anticommutator moments that could connect to enumerative combinatorics beyond the paper.","The $k$-dependence of the blip moment $(1/k)(2/k^2)^m \\mathbb{E}_k[\\operatorname{Tr} C_k^m]$ suggests a universality: only the dimension of the checkerboard's mean-space matters, not the detailed distribution of the non-weight entries, given finite higher moments."],"forward_implications":["If the {GOE,GOE} moment formula is correct, the limiting density is a new explicit algebraic benchmark for anticommutators of structured random matrices.","The {PTE,PTE} result implies the limiting spectrum of the anticommutator of two palindromic Toeplitz matrices is the same as the difference of two independent chi-squared variables.","The {GOE,$k$-checkerboard} blip formula shows that large-$N$ extreme eigenvalues are governed by a fixed $k\\times k$ hollow GOE, making small-matrix computations a proxy for extreme spectral statistics.","The {$k$-checkerboard,$j$-checkerboard} largest-blip formula provides an exact moment sequence for a single outlier eigenvalue that can be compared with numerical spectra.","The genus expansions for the block circulant anticommutators give a route to compute higher moments numerically even where no closed form is known."],"supporting_citations":[{"why":"provides the free-probability anticommutator moment series and the generating-function/density comparison used to identify the GOE-GOE limiting density.","marker":"[NR]"},{"why":"supplies Wick's formula, the genus expansion, and the non-crossing bound $\\#(\\gamma_{2m}\\pi)\\le m-1$ used throughout the moment calculations.","marker":"[MS]"},{"why":"introduces the palindromic Toeplitz ensemble and its free matching property, which underlies the PTE-PTE moment formula and the layer argument.","marker":"[MMS]"},{"why":"furnishes the block/weight vocabulary, the split-limiting analysis, and the polynomial-weight cancellation technique used for the checkerboard blips.","marker":"[BCDHMSTPY]"},{"why":"defines the $k$-block circulant ensemble and its matching rules, the basis for the genus expansion formulas for BCE anticommutators.","marker":"[KKMSX]"},{"why":"identifies the $m$-Schr\\\"oder numbers, giving the closed form used for the GOE-GOE moments.","marker":"[YJ]"},{"why":"provides the moment-generating-function uniqueness criterion that identifies the PTE-PTE density as a chi-squared difference.","marker":"[Bil]"}],"fun_headline_variants":["Exact spectral laws for anticommutator random matrices","Anticommutator ensembles: closed-form moments, blip spectra","Random anticommutators: from exact densities to N^2 blips","Anticommutator random matrices: combinatorial exact spectra","Blip eigenvalues pinned down in anticommutator ensembles"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The regime classification for checkerboard anticommutators rests on an unproved empirical input from Appendix B: that the mean-matrix part has exactly $k$ eigenvalues at each of $\\pm N^{3/2}/k$, and the analogous counts for the $k,j$ case, so if those positions or multiplicities were wrong the weight functions would isolate the wrong eigenvalues.","fun_headline_variants_meta":{"raw":{"variants":["Exact spectral laws for anticommutator random matrices","Anticommutator ensembles: closed-form moments, blip spectra","Random anticommutators: from exact densities to N^2 blips","Anticommutator random matrices: combinatorial exact spectra","Blip eigenvalues pinned down in anticommutator ensembles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000778,"raw_usage":{"total_tokens":3607,"prompt_tokens":1280,"completion_tokens":2327,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":896,"completion_tokens_details":{"reasoning_tokens":2241}},"tokens_in":896,"tokens_out":2327,"duration_ms":17950,"temperature":1.0,"reasoning_tokens":2241,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T18:44:05.142959+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For moderately large $N$ with $k \\mid N$ and $\\gcd(k,j)=1$, $jk \\mid N$, diagonalize $\\{A_N,B_N\\}$ for GOE/$k$-checkerboard and $k$-checkerboard/$j$-checkerboard and count eigenvalues in windows $[N^{3/2}/k - CN, N^{3/2}/k + CN]$ and $[2N^2/(jk)-CN^{3/2}, 2N^2/(jk)+CN^{3/2}]$; the central claim fails if the counts differ from $k$ per sign and $1$, respectively, or if the empirically weighted blip moments do not approach the formulas in Theorems 1.18 and 1.19.","supporting_citations":[],"review_version":1}