{"id":"f2808051-8522-4a9a-8f59-00db6feff72e","arxiv_id":"2509.05419","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A first non-Gaussian, 1-loop analytical template for the 4PCF covariance is derived using second-order densities and isotropic basis functions, reducing the problem to low-dimensional radial integrals.","lead":"This paper derives an analytical template for the 1-loop, non-Gaussian covariance of the galaxy four-point correlation function, using second-order density perturbations and galaxy bias. It matters because accurate covariance matrices are needed to assess claimed detections of parity violation in galaxy clustering.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (A.7)'s W(2) kernel normalization is inconsistent with Eq. (A.2): the stated c-coefficients make F2 and S(2) too large by factors (2j+1), so every final covariance template carries wrong j=1,2 weights.","rationale":"The reader's weakest assumption, the angular-momentum reduction in Appendices B through D, remains a reasonable thing to check. However, the single most load-bearing and immediately falsifiable problem sits upstream of all that, in Appendix A: the generalized kernel W(2) does not, as printed, equal the kernels it is supposed to unify. Direct substitution for F2 at equal momenta and theta=0 gives 2 from the first line of Eq. (A.2) but 100/21 from the second line with the stated c^(F) and D_j. The S(2) coefficient is similarly off by a factor of 5. Since every final template, Eqs. (3.14), (3.21), (3.27), (3.36), and (3.43), is linear in these c-coefficients, the numerical content of the paper's non-Gaussian covariance is wrong unless the coefficient values (or D_j) are corrected. This is an internal inconsistency rather than a disagreement with external consensus, and it is fixable: with c_{1,+/-1}=-1/(2sqrt(3)), c_{2,0}=4/(21sqrt(5)), and c^(0)_{2,0}=2sqrt(5)/15, the kernel normalization restores the standard F2 and S(2). The angular template structure might then survive, but the paper as printed cannot be used for survey covariances until this is corrected and the angular reduction is checked against a known limit such as the Gaussian covariance of Hou et al. Thus the reader's CONDITIONAL verdict is unchanged, but the condition is sharpened: not merely 'needs validation,' but 'contains a specific normalization error that must be corrected before use.'","tokens_in":30717,"tokens_out":45871,"duration_ms":416121,"concrete_test":"Evaluate Eq. (A.7) for F2 with q1=q2=(1,0,0) and k2=k3=1: the first line of Eq. (A.2) gives 17/21 + 1 + 4/21 = 2, whereas the second line with the stated c-coefficients and D_j gives 17/21 + 3 + 20/21 = 100/21. If the printed value is reproduced, the normalization is confirmed wrong; recompute the five templates with corrected coefficients c_{1,+/-1}=-1/(2sqrt(3)), c_{2,0}=4/(21sqrt(5)), and c^(0)_{2,0}=2sqrt(5)/15, and verify that the kernel recovers the standard F2 and S(2).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing issue is an internal inconsistency in the defining object of the calculation. Eq. (A.7) sets W(2)_mu = 4pi sum_{j,m,n} c^(mu)_{j,n} D_j Y_{jm}(qhat1)Y*_{jm}(qhat2) q1^n q2^{-n} with D_j = (-1)^j/sqrt(2j+1). Using sum_m Y_{jm}Y*_{jm} = (2j+1)/(4pi) L_j, the F2 kernel from the stated c^(F) values has j=2 coefficient 4pi c_{2,0} D_2 (5/4pi) = sqrt(5) c = 20/21, whereas Eq. (A.2)'s first line (standard SPT) requires 4/21; the j=1,n=+/-1 terms give 3/2 instead of 1/2. For S(2), c^(0)_{2,0}=2sqrt(5)/3 gives (10/3)L2 instead of (2/3)L2. The j=0 case (delta^2_lin) is the only correct one. Because W(2) enters every final template through the c sums, the printed coefficients would mis-weight all non-Gaussian j=1 and j=2 contributions by factors of 3 and 5. This is not a missing-validation issue; the kernel expansion as written does not reproduce the second-order SPT, tidal, or bias kernels it claims to encode.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives analytical templates for the 1-loop, non-Gaussian contribution to the covariance of the connected 4-point correlation function (4PCF), using second-order density contrasts and a second-order galaxy bias expansion. The calculation is organized into five Wick-contraction configurations (primary–secondary, half–half, primary–primary, PC–primary–secondary, PC–half–half), and each is reduced using the isotropic basis functions of Cahn & Slepian to products of angular-momentum coefficients and low-dimensional radial integrals g and f. The final expressions are Eqs. (3.14), (3.21), (3.27), (3.36), and (3.43). The paper claims to provide the first non-Gaussian 4PCF covariance template, with numerical validation deferred to a companion paper.","tokens_in":30986,"tokens_out":9071,"duration_ms":81978,"significance":"If correct, the result would be a valuable new tool for 4PCF analyses, including parity-odd statistics, and would extend the Gaussian covariance framework of Hou et al. to include nonlinear mode coupling and biasing. The paper is systematic and clearly organized: it introduces a unified second-order kernel W^(2), gives explicit angular-momentum machinery in Appendices B–D, and presents illustrative plots of the radial integrals. These are genuine strengths. However, the central derivation currently contains a normalization inconsistency in the defining kernel W^(2), and the manuscript provides no numerical or limit-based verification of the final templates. Because the error affects every final equation, the central claim as stated is not presently supported.","major_comments":[{"comment":"Equation (A.7) is internally inconsistent with Eq. (A.2). Expanding the second equality of (A.7) using the addition theorem sum_m Y_jm(qhat1)Y*_jm(qhat2) = (2j+1)/(4pi) L_j(qhat1·qhat2) gives W^(2)_mu = sum_{j,n} c^(mu)_{j,n} (-1)^j sqrt(2j+1) L_j(qhat1·qhat2) q1^n q2^{-n}. Inserting the stated c^(F) values yields a j=2 coefficient of (4sqrt(5)/21)*sqrt(5) = 20/21 and a j=1 coefficient of 3/2 (q1/q2+q2/q1), whereas Eq. (A.2) requires 4/21 and 1/2, respectively. For S^(2), c^(0)_{2,0}=2sqrt(5)/3 gives (10/3)L_2 instead of (2/3)L_2. Thus the kernel expansion does not reproduce the second-order SPT, tidal, or bias kernels it claims to encode. Since W^(2) enters every term of Eqs. (3.14), (3.21), (3.27), (3.36), and (3.43), all non-Gaussian j=1 and j=2 contributions are mis-weighted by factors of 3 and 5. This is a load-bearing error in the definition of the central object of the calculation, and it must be corrected (by fixing either the c coefficients or the D_j normalization) before the claimed templates can be accepted.","section":"§A.4, Eq. (A.7)"},{"comment":"The paper contains no numerical or symbolic verification that the final five templates are correct. In particular, there is no check that in the limit W^(2)=0 (equivalently b2=bs=0 with no second-order matter density) the expressions vanish, nor that specializing W^(2) to F^(2) reproduces known perturbation-theory results or reduces to the Gaussian covariance of Hou et al. in an appropriate limit. Given the complexity of the Gaunt/3j identities in Appendices B–D and the rotation-averaged Q coefficients in Appendix C, even one targeted validation—for example, numerically evaluating a single S integral against direct quadrature, or evaluating the W^(2) expansion for the F^(2) case—would substantially increase confidence. The absence of any such check is especially consequential here because the normalization inconsistency in Eq. (A.7) is exactly the kind of error such a check would have caught.","section":"§3 and Appendices B–D"}],"minor_comments":[{"comment":"In Eq. (3.16), the text says '+89 perms.' and the footnote correctly counts 90 total permutations, but the footnote text says 'other 35 combinations'; this should read 'other 89 combinations.'","section":"Footnote 3 and Eq. (3.16)"},{"comment":"In the definition of S for the half–half configuration, the argument of the second g-factor is written as g^{[0]}_{L2,L'2,L2s}(r2,r'_1,s); from the structure of the other terms and from Eq. (3.17), this should be (r2,r'_2,s).","section":"Eq. (3.22)"},{"comment":"In the primary–primary radial integral S, two arguments appear to be inconsistent: g^{[-n]}_{L3,L'3,L3s}(r3,r3,s) should likely be (r3,r'_3,s), and g^{[-n-n']}_{L02,L'02,L2qs}(r2,r'_2,s) should likely be (r0,r'_0,s) to match Eqs. (3.25)–(3.26).","section":"Eq. (3.28)"},{"comment":"In the PC–primary–secondary result, the fourth factor on the second line is written as wLk1,L10 bGL10,L10,j; based on the pattern of the preceding terms, this should involve the primed indices, e.g., wL'k'_2,L'20 bGL'20,L'20,j'.","section":"Eq. (3.36)"}],"recommendation":"major_revision","confidential_remarks":"The normalization error in Eq. (A.7) is fixable by rescaling the c coefficients or D_j, but because it propagates through all final templates, the correction must be made and checked before publication. Given the paper's heavy reliance on the authors' own prior work on isotropic basis functions, I suggest asking for an independent numerical check of the W^(2) expansion and at least one final S integral. The lack of numerical validation is acceptable for a methods paper only if the internal algebra is demonstrably consistent; currently it is not."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi [Name],\n\nShort version: this is the first non-Gaussian 4PCF covariance template, and the five-diagram decomposition is a real structural contribution. But there is a load-bearing normalization problem in the W(2) kernel. Appendix A.4 defines W(2) = 4π Σ c_j,n D_j Σ_m Y_jm Y*_jm q1^n q2^{-n}, with D_j = (-1)^j / sqrt(2j+1). Since Σ_m Y_jm Y*_jm = (2j+1)/(4π) L_j, the j=2 coefficient in F2 becomes 5 c_2,0 D_2 = 20/21, not the 4/21 required by the standard kernel in Eq. (A.2). The j=1 terms are off by factor 3, and S(2) is off by factor 5. Only the j=0 piece (δ^2_lin) is consistent. Every final template in §3 is built from these c sums, so the printed non-Gaussian results are wrong as they stand. This is not a missing validation; it is an internal inconsistency in the paper's own equations.\n\nCredit where due: the decomposition into three fully-connected and two partially-connected configurations is a genuine advance, and reducing the angular integrals to radial g and f integrals using the isotropic basis is a large amount of careful algebra. Appendix E's handling of the connected covariance subtraction is thoughtful and likely correct. The paper also avoids overclaiming numerical readiness, instead flagging validation for the companion paper.\n\nThe other soft spots are minor by comparison: the permutation count in the half-half section is garbled (Eq. (3.16) says ‘89 perms’ but footnote 3 gives 90 total, and the text says ‘other 35’), and the ‘All’ sums are left implicit, which makes independent verification painful. There is no numerical check, no recovery of a known limit, and no mock comparison.\n\nBottom line: I would send this to a serious referee. The flaw is concrete and fixable, and the framework is worth engaging with. But I would not use the templates as printed, and I would not cite this version. Ask the authors to correct the W(2) normalization, clean up the permutation counts, and provide at least one sanity check—e.g., a low-multipole comparison to the Gaussian covariance of Hou et al. or an N-body covariance.\n\nBest,\n[Name]","headline":"First non-Gaussian 4PCF covariance template, but the W(2) kernel in Appendix A carries a factor-(2j+1) normalization error that invalidates the printed non-Gaussian coefficients.","tokens_in":31565,"tokens_out":9341,"would_cite":false,"duration_ms":68637,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper derives the first analytic non-Gaussian template for the covariance of the 4-point correlation function.","keywords":["4-point correlation function","covariance matrix","standard perturbation theory","second-order density contrast","galaxy bias","isotropic basis functions","loop corrections","large-scale structure"],"falsifier":"Pick a low angular momentum sector (say $L\\le 4$), evaluate the left-hand side of Eq. (B.5) or (B.6) by direct numerical angular quadrature, and compare with the closed-form $G$/$H$/$J$ expression; any mismatch invalidates the corresponding template. Alternatively, set all second-order kernels to zero and check that the five-template covariance reduces to the Gaussian covariance of [8], a limit the paper does not demonstrate.","tokens_in":30469,"feed_emoji":"📐","tokens_out":9226,"duration_ms":73050,"temperature":0.7,"pith_summary":"The paper aims to replace the Gaussian-random-field covariance of the 4-point correlation function (4PCF) with a non-Gaussian, 1-loop template built from second-order density perturbations and second-order galaxy bias. It claims this is the first such calculation. The payoff would be error bars on 4PCF measurements---both parity-even and parity-odd---that reflect nonlinear mode coupling, which becomes non-negligible near $k \\gtrsim 0.1\\,h\\,\\mathrm{Mpc}^{-1}$. The central reduction turns high-dimensional integrals into angular integrals over isotropic basis functions plus low-dimensional radial integrals, so the final templates are meant to be usable in practice. A companion paper extends the treatment to third-order densities.","feed_headline":"First non-Gaussian covariance template for the 4-point function","feed_subtitle":"Includes 1-loop second-order density and galaxy bias, cutting the covariance to low-dimensional radial integrals.","key_machinery":"The load-bearing object is the generalized second-order kernel $W^{(2)}$, which unifies the standard $F^{(2)}$ kernel of SPT, the unit kernel of $\\delta_{\\mathrm{lin}}^2$, and the tidal kernel $\\tfrac{2}{3}L_2$ into one expression with coefficients $c^{(\\mu)}_{j,n}$. Alongside it, the calculation relies on the isotropic basis functions, the Gaunt-type integration coefficients $G$, $H$, and $J$ of Appendix B, the rotation-averaged coefficients $Q$ of Appendix C, and the mixed-space splitting $\\omega$ of Appendix D. These identities convert the high-dimensional angular integrals into products of one-dimensional radial integrals $g$ and $f$, coupled only through a single line-of-sight separation $s$; the final expressions are sums over angular momenta of these radial pieces times Legendre polynomials in the two tetrahedron orientations.","core_discovery":"The central claim is that, within standard perturbation theory at second order in the linear density field and with a second-order Eulerian bias expansion, every 1-loop contribution to the connected 4PCF covariance falls into one of five Wick-contraction configurations---three fully connected and two partially connected---and each can be written in closed angular form as Eqs. (3.14), (3.21), (3.27), (3.36), and (3.43). The full covariance is assembled by mapping any contraction to one of these universal templates, with all second-order density types ($\\delta^{(2)}$, $\\delta_{\\mathrm{lin}}^2$, and the tidal term $S^{(2)}$) absorbed into a single generalized kernel $W^{(2)}$; substituting the appropriate kernel coefficients then gives the covariance for matter or biased galaxies. The angular parts of every integral are performed analytically, leaving only products of radial integrals $g$ and $f$ over a single loop separation $s$.","pith_inferences":["A testable extension follows from the paper's logic: the same five-template reduction should hold for any second-order kernel expressible in the $W^{(2)}$ form, so redshift-space or effective-field-theory kernels could be inserted without re-deriving the angular identities.","If the Appendix B--D identities are correct, the same Gaunt-based technology may also simplify covariances of higher-order $N$-point functions, since the same angular bottleneck appears there.","The practical numerical cost of the final expressions sits in the radial integrals $g$ and $f$; the paper's plots suggest these have sharp rectangular boundaries that may admit analytic evaluation for power-law spectra, as prior work on such integrals indicates.","The paper states that the non-Gaussian corrections become non-negligible near $k\\approx0.1\\,h\\,\\mathrm{Mpc}^{-1}$ and lists numerical validation against $N$-body simulations as future work, so the template's domain of validity is not yet established by the paper itself."],"forward_implications":["All 1-loop contractions of second-order densities are exhausted by five templates, so any future calculation can match contractions to a template and substitute the appropriate $W^{(2)}$ coefficients rather than redo the angular algebra.","The covariance's angular dependence separates into two Legendre factors $P_\\Lambda$ and $P_{\\Lambda'}$ in the two tetrahedron orientations, with all scale dependence carried by radial integrals that can be precomputed.","The Gaussian (disconnected) covariance is recovered as the leading piece, with the new terms supplying the next-to-leading-order correction relevant for 4PCF analyses around $k \\sim 0.1\\,h\\,\\mathrm{Mpc}^{-1}$.","By choosing the coefficients $c^{(\\mu)}_{j,n}$, the same templates give the covariance for the matter density, the squared density, or the tidal field, and for galaxies with second-order bias.","The companion paper on third-order densities completes the 1-loop covariance, making the full analytic template available for survey analyses."],"supporting_citations":[{"why":"Supplies the isotropic basis functions and rotation-averaging identities used throughout the angular reduction.","marker":"[3]"},{"why":"Gives the Gaussian 4PCF covariance that this work extends and to which the non-Gaussian result must reduce in the linear limit.","marker":"[8]"},{"why":"Provides the plane-wave expansion constants, the $F^{(2)}$ kernel expansion in the isotropic basis, and the radial integral $g$.","marker":"[13]"},{"why":"Supplies the standard perturbation theory framework and the second-order density kernel $F^{(2)}$.","marker":"[1]"},{"why":"Gives the Eulerian bias expansion used to model galaxy density fluctuations.","marker":"[17]"},{"why":"Supplies the Gaunt integral and 3j-symbol identities used in Appendix B.","marker":"[7]"},{"why":"Provides analytic results for the triple-spherical Bessel radial integrals that the final templates rely on for evaluation.","marker":"[5]"}],"fun_headline_variants":["First non-Gaussian 4PCF covariance template","1-loop 4PCF covariance beyond Gaussian field","Non-Gaussian 4PCF covariance with second-order bias","4PCF covariance reduced to radial integrals","Five templates close all 1-loop 4PCF covariance"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole reduction depends on the angular-momentum identities in Appendices B--D---the Gaunt-based coefficients $G$, $H$, $J$, the mixed-space splitting $\\omega$, and the rotation-averaged $Q$---being algebraically correct; a sign or 3j-phase error anywhere propagates into every final template.","fun_headline_variants_meta":{"raw":{"variants":["First non-Gaussian 4PCF covariance template","1-loop 4PCF covariance beyond Gaussian field","Non-Gaussian 4PCF covariance with second-order bias","4PCF covariance reduced to radial integrals","Five templates close all 1-loop 4PCF covariance"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000223,"raw_usage":{"total_tokens":1445,"prompt_tokens":924,"completion_tokens":521,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":540,"completion_tokens_details":{"reasoning_tokens":443}},"tokens_in":540,"tokens_out":521,"duration_ms":4623,"temperature":1.0,"reasoning_tokens":443,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:23:45.148727+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Pick a low angular momentum sector (say $L\\le 4$), evaluate the left-hand side of Eq. (B.5) or (B.6) by direct numerical angular quadrature, and compare with the closed-form $G$/$H$/$J$ expression; any mismatch invalidates the corresponding template. Alternatively, set all second-order kernels to zero and check that the five-template covariance reduces to the Gaussian covariance of [8], a limit the paper does not demonstrate.","supporting_citations":[],"review_version":1}