{"id":"c536eede-d1a7-4020-9e52-8949fdc106b3","arxiv_id":"2502.02246","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A null test of BNT-transformed cosmic shear cross-spectra can constrain Omega_m and w0 geometrically, but requires photometric redshifts known to about 10^-3 and is subdominant to standard 3x2pt analysis.","lead":"Tomographic cosmic shear maps hide symmetries that depend only on the geometry of the universe, and this paper turns one such symmetry, the BNT nulling property, into a new cosmological probe. The authors forecast that in a Euclid-like survey the test can constrain matter density and dark energy, but its main value may be as a photometric redshift calibration tool and consistency check.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Single-amplitude Q3 bispectrum template (Eqs. 42–45) is the load-bearing approximation; a shape mismatch can bias the Ωm–w0 posteriors despite marginalization.","rationale":"The paper is internally consistent and states its limitations honestly. The BNT construction itself is standard, the nulling-coefficient derivations are explicit, and the degeneracy analysis in Appendix B is a useful independent check. The strongest support for the central claim comes from the Fisher and MCMC forecasts; however, those forecasts are only as credible as the bias model used to build the likelihood. The single-amplitude bispectrum template is the least secure element because the exact cancellation that makes the probe geometrical holds only at first order. The reduced-shear and magnification-bias corrections reintroduce a P(k)- and bispectrum-shape-dependent term, and handling that term with one Q3 amplitude is an unvalidated ansatz. The reader's weakest_assumption correctly identifies this. A shape mismatch could produce a bias of the same order as the claimed 1σ intervals, which would weaken the proof-of-concept. The proposed test—comparing the reduced template against the full Feff2 form or a simulation-based bispectrum—would settle whether the concern lands. I therefore see no reason to alter the CONDITIONAL verdict; the concern is real but, as the reader says, it is addressable in revisions and not grounds for rejection.","tokens_in":22479,"tokens_out":4902,"duration_ms":52703,"concrete_test":"Generate noiseless mock data vectors using the full Feff2 kernel of Eq. (42) at the fiducial cosmology (or, better, using a simulation-calibrated bispectrum), including reduced shear and magnification bias. Then run the MCMC likelihood of Eq. (50) with the reduced Q3 template of Eq. (43). If the resulting 68% posterior for (Ωm, w0) does not contain the fiducial values, the shape mismatch is not absorbed by marginalising Q3. A cheaper version: compute δB_C from Eq. (43) and from Eq. (42) for every (a,i) and ℓ bin; if the relative residuals exceed the statistical error on the nulled data vector, the single-amplitude template is inadequate.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV.D sets the bispectrum correction to δBC_ai = Q3 times a template computed with Eq. (43) from the fiducial Halofit power spectrum. The likelihood in Eq. (50) then treats Q3 as a free amplitude. For this to be unbiased, the scale and redshift shape of the reduced-shear plus magnification-bias bispectrum contribution must be representable by that single template. The paper does not test this. Eq. (43) assumes the Scoccimarro–Couchman kernel reduces to B=Q3(P P+cyc.) on the relevant scales; at ℓ up to 16000 the integrand samples the mildly and fully nonlinear regime, where the full Feff2(k1,k2) is scale- and configuration-dependent. A shape mismatch changes the relative weights of the 28 (a,i) cross-spectra and ℓ bins, so no single Q3 can absorb it. If the true bispectrum shape differs, the model ⟨Ĉ⟩ = Q3 δB_C in Eq. (45) is wrong, and the reported Ωm–w0 intervals in Table I can be biased rather than merely degraded. This is load-bearing because the central claim is a geometric cosmological constraint, and the accuracy of that constraint is controlled precisely by this bispectrum model.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a new cosmological probe based on the BNT nulling transform of tomographic cosmic shear maps. The nulling coefficients p_ai are constructed from geometric quantities (source-redshift moments n^(0)_i and n^(1)_i) so that the nulled lensing kernels vanish below a chosen redshift. The authors propose to test the geometry by cross-correlating nulled convergence maps with low-redshift galaxy density maps and requiring the cross-spectra C^{κg}_{ai}(ℓ) to vanish for non-overlapping index pairs. Any residual signal then constrains the background geometry, specifically Ω_m and w0. Using a Euclid-like survey configuration, they compute Fisher and MCMC forecasts with a Gaussian likelihood (Eq. 50), marginalizing over a Q3 amplitude intended to absorb reduced-shear and magnification-bias bispectrum corrections. They report constraints such as Ω_m = 0.37^{+0.19}_{-0.11}, w0 = -1.17^{+0.38}_{-0.075} at ℓ_max = 16000 (Table I), and find that a photo-z mean accuracy of order 10^-3 is required to preserve the cosmological information. They conclude that the method is better suited as a photo-z calibration probe or consistency check than as a primary cosmological probe.","tokens_in":22640,"tokens_out":11500,"duration_ms":112095,"significance":"If the robustness issues are satisfactorily resolved, the probe is attractive because the nulling condition itself depends only on background geometry and is independent of the matter power spectrum and growth physics; this is a genuine complement to standard 3x2pt analyses, and the degeneracy structure (Eq. 14) is explicitly derived. The paper is also honest about the test's subdominant constraining power and identifies the photo-z precision requirement. The main limitation is the treatment of the bispectrum bias as a single universal amplitude, which currently controls the accuracy of the headline constraints. With additional validation of the bispectrum template shape, the methodology would be publishable as a proof of concept.","major_comments":[{"comment":"The bispectrum bias is modeled as a single amplitude Q3 times a template δB_C computed from Eq. (43) at fiducial parameters, and Eq. (50) subtracts Q3 δB_C from the data vector. For this to be unbiased, the scale and redshift shape of the reduced-shear plus magnification-bias bispectrum contribution must be representable by that one template. Eq. (43) is justified only in the deeply nonlinear regime where the Scoccimarro–Couchman kernel reduces to a constant; at ℓ up to 16000 the line-of-sight integrand also receives contributions from mildly nonlinear and quasi-linear scales, where the full kernel in Eq. (42) is configuration- and scale-dependent. A shape mismatch cannot be absorbed by a single amplitude: it changes the relative weights of the 28 (a,i) cross-spectra and ℓ bins, so inference from the likelihood (50) would be biased rather than merely degraded. The manuscript should test this approximation, for example by repeating the forecasts with the full fitting formula (42), with a tree-level bispectrum, or with a redshift-dependent shape parameter, and by showing that the reported Ω_m–w0 intervals in Table I are stable.","section":"§IV.D, Eq. (45)"},{"comment":"The covariance for the data vector is approximated as diagonal in the low-redshift bin indices via C^{gg}_{ij} ∝ δ_ij. With nonzero photo-z scatter, the optimized low-z galaxy distributions in Table III have overlapping true-redshift kernels, so C^{gg}_{ij}(ℓ) is generally nonzero for i ≠ j. The Gaussian covariance of the cross-spectra then receives an additional contribution of the form C^{κκ}_{ab}(ℓ) C^{gg}_{ij}(ℓ) that is not included in Eq. (27). If this contribution is non-negligible, the quoted signal-to-noise levels and the Fisher/MCMC error bars in Figs. 5–10 and Tables I–II are underestimated. Please quantify the overlap and either include the full C^{gg}_{ij} matrix or demonstrate that the off-diagonal terms are subdominant.","section":"§III.B, Eq. (27)"},{"comment":"The paper repeatedly states that the nulling test is independent of the matter power spectrum P(k). This is true for the expectation value ⟨C^{κg}_{ai}⟩ = 0, but it is not true for the forecasted constraints: the covariance in Eq. (27), the Fisher matrix in Eq. (49), and the signal-to-noise estimates all use C^{κκ}_{ab}(ℓ) and C^{gg}_i(ℓ) computed with a Halofit model. The authors note that the covariance could be measured from the survey, which is a valid route, but as presented the forecasts are not P(k)-free. The text should clearly distinguish the nulling test's immunity to growth and small-scale physics from the modeling dependence of its noise, and should report the sensitivity of the constraints to the assumed P(k) model.","section":"§II.C and §III.B"}],"minor_comments":[{"comment":"The section heading 'Biqspectrum Bias' contains a typo; it should read 'Bispectrum Bias'.","section":"§IV heading"},{"comment":"'working whith the 3x2pt method' should read 'working with the 3x2pt method'; similar typos such as 'teh' in Appendix A should be corrected throughout.","section":"§II.C"},{"comment":"The text says the noise is divided by 100 ≈ √10,000 because roughly 10,000 ℓ modes are available. Since the analysis uses 32 logarithmically spaced ℓ bins, the effective mode count is an integral of (2ℓ+1) f_sky over each bin, not a single factor of 10,000. Please clarify how the scaled noise amplitude is defined.","section":"Fig. 3 and §V.A"},{"comment":"The diagonalization of the Fisher matrix is used to identify exact degeneracy directions, e.g., a 'kernel of dimension 3'. Fisher eigenvalues identify approximate null directions at a given fiducial model; the language should be softened unless an exact algebraic proof is intended.","section":"§VI.C"},{"comment":"The posteriors in Figs. 6–9 are visibly non-Gaussian; reporting the maximum-a-posteriori point and credible intervals in addition to the marginal one-sigma ranges would make the quoted intervals easier to interpret.","section":"Tables I and II"}],"recommendation":"major_revision","confidential_remarks":"The paper's central idea is sound and the non-circularity of the null test is clear: the nulling coefficients are computed for an assumed cosmology, and a wrong cosmology produces non-vanishing residuals rather than being absorbed into the fit. My main concern is that the headline parameter constraints in Table I rest on the single-amplitude Q3 bispectrum template, which is not validated for the scale and redshift range used. I would ask for an explicit test of the template shape before publication. The covariance diagonalization issue in Eq. (27) is secondary but should be quantified. I do not see a need to reject the manuscript; the requested checks are within the scope of a revision, and the authors could alternatively strengthen the photo-z calibration conclusion, which is already their stated practical recommendation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this is a genuine conceptual step, not just another forecast. BNT nulling existed; using the nulling condition itself as a geometric observable—requiring the nulled shear–galaxy cross-spectra to vanish—is new, and the paper does the work to make it usable. The derivation of the ξ''/ξ' degeneracy (Eq. 14, Appendix B) is clean and useful, the counting argument (Nz−3 constraints) is clear, and the photo-z requirement of ~10^-3 is a concrete, falsifiable statement. The covariance is independent of P(k) and super-sample covariance by construction, which is a real advantage. The Fisher and MCMC results are consistent with each other, and the authors are honest that nulling is subdominant and better used as a photo-z calibration or consistency check. The null test is non-circular: coefficients are fixed by an assumed geometry and checked against data. The comparison with DES shear ratios is also handled properly; Appendix E is a useful clarification, not just a box-ticking exercise.\n\nThe soft spot is exactly where the stress-test note points. Section IV.D models the reduced-shear and magnification-bias bispectrum correction as a single amplitude Q3 times a template computed from Eq. (43), the Q3 (P P + cyc.) form of Scoccimarro–Couchman. That form is valid at small scales, but the ℓ integrals run to 16000 and sample the mildly and fully nonlinear regime, where the effective kernel is scale- and configuration-dependent. If the true shape differs from the template, marginalizing over Q3 will not absorb it; the reported Ωm and w0 intervals in Table I could be biased, not just wider. The paper does not test this. That is the main load-bearing approximation, and I would want it stress-tested before quoting the numbers.\n\nTwo smaller things. The claim that nulling is subdominant to 3x2pt is asserted rather than demonstrated; it is probably true, but a quantitative comparison would strengthen the conclusion. And there is no code or data release, so independent replication is harder than it should be for a methodology paper. Neither is fatal.\n\nBottom line: this is a serious paper by people who know the subject. The conceptual contribution—nulling as a geometric probe, with the degeneracy structure worked out—will stand regardless of the bispectrum template question. I would send it out for peer review and ask for a robustness test of the Q3 template against a different bispectrum model or a simulation, plus a quantitative subdominance check.","headline":"A clean, honest forecast paper that makes a real conceptual step—using BNT nulling itself as a geometric probe—but the single-amplitude Q3 bispectrum template is the place to push before trusting the quoted intervals.","tokens_in":23267,"tokens_out":2798,"would_cite":true,"duration_ms":27973,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The BNT nulling transform turns the geometry of cosmic shear into a cosmological test that constrains $\\Omega_m$ and $w_0$ without modelling the matter power spectrum.","keywords":["cosmic shear","BNT nulling transform","tomographic weak lensing","dark energy equation of state","geometrical cosmology","photometric redshift calibration","bispectrum corrections","null test"],"falsifier":"Compute the nulled cross-spectra $C^{\\hat\\kappa g}_{ai}$ from a large hydrodynamical simulation with realistic baryonic feedback and a spectroscopically calibrated source redshift distribution, or from real survey data with photo-z calibrated to $10^{-4}$; if the residuals after subtracting the best-fit $Q_3$ template exceed the Knox-approximation noise and grow with multipole or vary across bin pairs in a way the template cannot match, the universal-bispectrum assumption fails and the claimed constraints are biased.","tokens_in":22203,"feed_emoji":"🔭","tokens_out":13062,"duration_ms":116569,"temperature":0.7,"pith_summary":"The paper tries to establish that the BNT nulling transform of tomographic cosmic shear maps is a purely geometrical cosmological probe: the cross-spectra between nulled shear maps and low-redshift galaxy maps should vanish, and requiring them to vanish constrains the background expansion, in particular the matter density $\\Omega_m$ and the dark-energy equation-of-state parameter $w_0$. Because the nulling coefficients are built from angular-diameter-distance ratios alone, the test is independent of the matter power spectrum and can therefore use small angular scales without modelling non-linear growth, baryonic physics, galaxy bias, or intrinsic alignments. For a large 15,000 square-degree survey with 10 tomographic bins and 30 galaxies per square arcminute, the forecast reaches $\\Omega_m = 0.37^{+0.19}_{-0.11}$ and $w_0 = -1.17^{+0.38}_{-0.075}$ at $1\\sigma$ after marginalising over a nuisance amplitude that absorbs reduced-shear and magnification-bias corrections. The paper identifies a precision requirement of about $10^{-3}$ on photometric mean redshifts and concludes that nulling is best used as a photometric-redshift calibrator and consistency check alongside the usual galaxy clustering and weak-lensing two-point analyses.","feed_headline":"Nulling test reads dark energy from cosmic shear geometry","feed_subtitle":"Vanishing nulled shear-galaxy spectra pin down matter density and dark energy without power-spectrum models.","key_machinery":"The central object is the BNT nulling transform: a linear map with coefficients $p_{ai}$ applied to redshift-binned convergence (or shear) maps so that each nulled lensing kernel $\\hat w_a$ is supported only on bins $a-2$ through $a$. The coefficients are fixed by the source-distance moments $n_i^{(0)}=\\int d\\chi\\, n_i(\\chi)$ and $n_i^{(1)}=\\int d\\chi\\, n_i(\\chi)/F_K(\\chi)$ through the constraints $\\sum_i p_{ai}n_i^{(0)}=0$ and $\\sum_i p_{ai}n_i^{(1)}=0$, so they depend on the expansion history through the comoving distance and curvature factor, with $H_0$ cancelling. The observable is the cross-spectrum $C^{\\hat\\kappa g}_{ai}$ between nulled maps and optimised low-redshift galaxy bins, with covariance $\\Sigma_{ai,bj}(\\ell)$ in the Knox approximation; the construction ensures the covariance requires no trispectrum or super-sample covariance terms. Any wrong background geometry shifts the coefficients and makes the supposedly vanishing cross-spectra grow, and at the order considered the only astrophysical contamination enters through a single bispectrum template amplitude $Q_3$.","core_discovery":"The paper's central claim is that nulling is not merely a way to reorganise shear data, but a fundamental probe of space-time geometry: the same BNT coefficients that make nulled lensing kernels vanish at low redshift carry information about the redshift dependence of the angular diameter distance and hence about $\\Omega_m$ and $w_0$. The observable is the cross-spectrum $C^{\\hat\\kappa g}_{ai}$ between a nulled convergence map $\\hat\\kappa_a$ and a low-redshift galaxy density bin $\\delta^g_i$ with $i \\leq a-3$; the nulling condition $\\langle C^{\\hat\\kappa g}_{ai}\\rangle = 0$ is the likelihood constraint. In a 10-bin survey configuration the Fisher and MCMC analyses deliver constraints like $\\Omega_m = 0.37^{+0.19}_{-0.11}$ and $w_0 = -1.17^{+0.38}_{-0.075}$ at $\\ell_{\\max}=16{,}000$ after marginalising over the bispectrum amplitude $Q_3$, with the covariance computed in the Knox approximation and needing no trispectrum or super-sample covariance model. The constraints are limited by a geometrical degeneracy between models sharing the same ratio $\\xi_K''/\\xi_K'$ (or equivalently the same $H^{-1}dH/dz + 2/\\chi$), which the paper analyses in Appendix B, and by the need for source mean redshifts known to $10^{-3}$. The authors themselves frame the method's practical payoff as a photometric-redshift calibration tool and a complement to two-point clustering and lensing analyses rather than as a standalone probe.","pith_inferences":["The paper does not explore this, but the same geometric argument should apply to convergence maps reconstructed from other tracers, such as CMB lensing or line-intensity maps, yielding a nulling-based dark-energy check at redshifts where galaxy photo-z are not involved.","A direct extension would be to replace the single $Q_3$ amplitude with a shape-flexible bispectrum model (two or three amplitudes for different triangle configurations) and rerun the Fisher analysis; if the contours widen substantially, the quoted errors are sensitive to the template choice.","The local degeneracy relation suggests a diagnostic use: binning the nulling constraints in redshift and comparing them with constraints from probes that break the degeneracy could isolate which redshift range is responsible for a $\\Omega_m$–$w_0$ shift.","A nulling residual that varies smoothly under artificial bin shifts could be inverted into per-bin mean-redshift corrections, effectively turning the $10^{-3}$ requirement into a calibration measurement rather than a limitation."],"forward_implications":["The nulling condition yields constraints on $\\{\\Omega_m, w_0\\}$ that need no model for $P(k)$ or for non-linear growth, baryonic feedback, galaxy bias, or intrinsic alignments, so small angular scales can be included safely.","The covariance is built from spectra the survey already measures, and it is insensitive to trispectrum and super-sample covariance terms, making the probe nearly cost-free to append to an existing two-point analysis pipeline.","With 10 tomographic bins the BNT matrix has $N_z-3 = 7$ independent coefficients, so nulling supplies at most seven constraints and cannot simultaneously fit cosmology and all ten photometric mean-redshift errors.","Because the $\\Omega_m$–$w_0$ correlation follows the geometrical degeneracy relation at redshift around 1, nulling contours are nearly perpendicular to standard weak-lensing contours and add complementary information when combined.","The requirement that cosmological constraints survive forces photometric mean redshifts to be known at the $10^{-3}$ level; at $10^{-2}$ precision the constraints degrade, which is the paper's stated reason for recommending nulling as a redshift calibrator and consistency check."],"supporting_citations":[{"why":"Introduces the BNT transform and derives the nulling coefficients from source-distance moments; the probe is built directly on this construction.","marker":"[8]"},{"why":"Describes the earlier nulling and boosting techniques for weak lensing statistics that the paper contrasts with its global nulling approach.","marker":"[13]"},{"why":"Applies lensing shear ratios to survey data and serves as the comparison observable against which BNT nulling is positioned.","marker":"[14]"},{"why":"Exploits lensing shear ratios with small-scale information and provides the other main baseline for nulling.","marker":"[15]"},{"why":"Supplies the reduced-shear and magnification-bias formalism used to compute the bispectrum-level corrections.","marker":"[23]"},{"why":"Provides the fitting formula $B(k_1,k_2,k_3)=Q_3[P(k_1)P(k_2)+\\mathrm{cyc.}]$ that defines the bispectrum template and the nuisance parameter $Q_3$.","marker":"[28]"},{"why":"Defines the survey configuration (sky fraction, redshift distribution, shape noise) used for the forecasts.","marker":"[22]"},{"why":"Provides the Markov Chain Monte Carlo sampler used to obtain the posterior constraints.","marker":"[29]"}],"fun_headline_variants":["Shear nulling exposes geometric backbone of dark energy","Nulled shear spectra pin down Omega_m and w0","Cosmic shear nulling: a geometric shortcut to cosmology","Nulling test reads cosmic geometry without power-spectrum models","Shear nulling: pure geometry, no model dependence"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise, introduced in the bispectrum-correction section (Eqs. 42–45) and used in the likelihood (Eq. 50), is that all reduced-shear and magnification-bias corrections share one fixed scale and redshift pattern, so that a single free amplitude $Q_3$ multiplying that template absorbs them; if the true pattern differs, the inferred $\\Omega_m$ and $w_0$ will be biased.","fun_headline_variants_meta":{"raw":{"variants":["Shear nulling exposes geometric backbone of dark energy","Nulled shear spectra pin down Omega_m and w0","Cosmic shear nulling: a geometric shortcut to cosmology","Nulling test reads cosmic geometry without power-spectrum models","Shear nulling: pure geometry, no model dependence"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000232,"raw_usage":{"total_tokens":1587,"prompt_tokens":1138,"completion_tokens":449,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":754,"completion_tokens_details":{"reasoning_tokens":369}},"tokens_in":754,"tokens_out":449,"duration_ms":4763,"temperature":1.0,"reasoning_tokens":369,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T12:49:15.871831+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the nulled cross-spectra $C^{\\hat\\kappa g}_{ai}$ from a large hydrodynamical simulation with realistic baryonic feedback and a spectroscopically calibrated source redshift distribution, or from real survey data with photo-z calibrated to $10^{-4}$; if the residuals after subtracting the best-fit $Q_3$ template exceed the Knox-approximation noise and grow with multipole or vary across bin pairs in a way the template cannot match, the universal-bispectrum assumption fails and the claimed constraints are biased.","supporting_citations":[{"cited_title":"This procedure is purely geometri- cal and thus scale-independent and independent of the power spectrum or matter density contrast","cited_arxiv_id":null,"evidence_quote":"Introduces the BNT transform and derives the nulling coefficients from source-distance moments; the probe is built directly on this construction."},{"cited_title":"pre-factor unity approximation","cited_arxiv_id":null,"evidence_quote":"Exploits lensing shear ratios with small-scale information and provides the other main baseline for nulling."},{"cited_title":"Kaiser, Weak gravitational lensing of distant galaxies, As- trophys","cited_arxiv_id":null,"evidence_quote":"Provides the Markov Chain Monte Carlo sampler used to obtain the posterior constraints."}],"review_version":1}