{"id":"74998d94-8576-4320-a853-f5bcfc41972e","arxiv_id":"1908.04854","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Tomographic cross-correlation of Planck CMB lensing with SCUSS/SDSS/WISE galaxies yields a linear growth amplitude A_D = 1.16 ± 0.13, consistent with ΛCDM.","lead":"The authors used the bending of ancient light by galaxies to measure how cosmic structures grow over time, combining Planck CMB lensing maps with galaxies from SCUSS, SDSS, and WISE. They find growth consistent with the standard dark-energy model, with no significant deviation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The headline A_D = 1.16 ± 0.13 rests on a single-Gaussian dn/dz whose high-z end is known to be biased; the two highest-z bins also have poor PTE, so the quoted amplitude and error should be re-derived with an empirical photo-z error model.","rationale":"The reader's weakest-assumption analysis identifies the same load-bearing element: the fixed-Gaussian photo-z error model in Eq. (3.2) is unreliable at high redshift, and the paper's own tests show the galaxy bias can shift by up to 30% when σz is varied. I agree that this is the most fragile premise for the central amplitude claim. The concern is strengthened by the poor PTE values of the two highest-z bins in the galaxy-galaxy fits, which indicate the assumed dn/dz is not consistent with the data. The central conclusion, that linear growth is consistent with ΛCDM, is likely correct in a broad sense: the three-lowest-bin fit gives A_D = 1.22 ± 0.19, and the null and foreground tests are reassuring. However, the specific headline value A_D = 1.16 ± 0.13 cannot be fully trusted until the high-z dn/dz is validated with an empirical photo-z error distribution or the high-z bins are excluded from the primary fit. This does not warrant rejection, but it justifies a conditional verdict requiring the additional photo-z robustness check and, ideally, release of the measured D̂_G points and code for reproducibility.","tokens_in":22085,"tokens_out":5923,"duration_ms":68138,"concrete_test":"Use the spectroscopic training sample from [47] to construct an empirical p(z|zph) for each of the six bins, including a z-dependent scatter and catastrophic outliers rather than the single Gaussian in Eq. (3.2); recompute dn/dz, re-derive the theoretical spectra in Eq. (2.5), and re-fit A_D from Eq. (5.2). If the new A_D remains within roughly 0.1 of 1.16 with a similar uncertainty, the photo-z concern does not change the central claim; if it shifts by more than 0.1, the headline precision and the inclusion of the high-z bins in the main A_D fit are not robust.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Equation (3.2) reconstructs the true dn/dz by convolving the photo-z distribution with one Gaussian error of fixed σz = 0.019 in (1+z), as reported by the catalogue paper [47]. Section 5.1 states that for galaxies with zspec ≳ 0.6 the catalogue's photometric redshifts tend to be underestimated because of a lack of high-z galaxies in the training set. The two highest redshift bins are precisely where the galaxy auto-spectrum fit fails (PTE = 0.59% and 0.006% in Table 2), indicating that the assumed dn/dz is not an adequate description of the data there. The D̂_G estimator in Eq. (2.7) is bias-independent, but it is not dn/dz-independent: the theoretical Cκg and Cgg entering the ratio are computed from this assumed dn/dz, so an incorrect or non-Gaussian redshift distribution shifts the inferred D̂_G(z) values. Since A_D = 1.16 ± 0.13 is a single-amplitude fit to all six redshift bins, the headline result absorbs any systematic offset in the two high-z bins. The reported robustness fit using only the three lowest bins, A_D = 1.22 ± 0.19, shows the qualitative conclusion is stable, but it does not quantify how much the high-z bins pull the combined amplitude; a corrected dn/dz could move A_D by more than the quoted 0.13 error.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a tomographic measurement of the linear growth of structure using the cross-correlation between Planck CMB lensing and a galaxy overdensity map built from the SCUSS/SDSS/WISE photometric catalogue. The analysis splits the sample into six redshift bins over 0.1 < z < 0.7, measures galaxy auto and galaxy-CMB lensing cross power spectra with a pseudo-C_l (MASTER) estimator, and fits galaxy bias and the lensing amplitude in each bin. It then applies the bias-independent D_G estimator of Giannantonio et al. (2016) to constrain the linear growth factor, obtaining A_D = 1.16 ± 0.13 relative to the fiducial Planck cosmology, consistent with ΛCDM. The paper includes extensive systematics tests: tSZ deprojection, extinction masking, null tests with simulated maps, and a comparison of jackknife, Monte Carlo, and analytic covariances.","tokens_in":22391,"tokens_out":6703,"duration_ms":65082,"significance":"If the result holds, this is a useful new measurement of the growth of linear density fluctuations at z < 0.7 from a photometric survey covering a different sky region than previous analyses, and it is consistent with Planck ΛCDM. The paper's strengths are its careful pseudo-C_l implementation, the explicit validation of the covariance matrix against analytic and Monte Carlo estimates, and the battery of null and foreground tests. The main weakness is the reliance on a single-Gaussian photo-z error model and the poor goodness-of-fit in the two highest redshift bins, which may bias the headline A_D in a way not captured by the quoted statistical error.","major_comments":[{"comment":"The headline amplitude A_D = 1.16 ± 0.13 is obtained from a fit to all six redshift bins, including the two highest-z bins whose galaxy auto-spectra are poorly fitted (Table 2: PTE = 0.59% and 0.006%). Section 5.1 shows that the galaxy bias in those bins shifts by up to ~30% when σz is changed to 0.04 or 0.0, but the analysis never propagates these variations to D̂G or A_D. Because D̂G in Eq. (2.7) is not dn/dz-independent—the slashed theoretical spectra are computed from the assumed dn/dz—an incorrect or non-Gaussian photo-z error can bias A_D by an amount not reflected in the quoted 0.13 error. The authors should recompute D̂G and A_D with an empirical photo-z error distribution, or at least with the σz = 0.04 and σz = 0.0 variants already used in §5.1, and report the resulting shift in A_D. They should also quantify how much the two high-z bins pull the combined A_D, for example by comparing the six-bin fit to a fit that excludes those bins with a proper account of the covariance.","section":"§5.3, Eq. (5.2)"},{"comment":"Table 2 reports PTE = 0.006% for the galaxy-galaxy fit in the 0.6 < z < 0.7 bin, yet the Abstract and Conclusions state that no significant evidence for systematic effects is found. Even if the poor fit is attributed to the known photo-z underestimation at z ≳ 0.6 (as argued in §5.1), a 0.006% PTE is itself a detected model-data discrepancy in a bin that enters the headline A_D fit. The language of the paper should be qualified, and the A_D result should be presented with the caveat that two of the six bins are not well described by the assumed dn/dz model.","section":"Abstract and §5.2"},{"comment":"The D̂G error bars and the A_D uncertainty are computed from 500 Gaussian Monte Carlo realizations that are generated using the same assumed dn/dz and the fiducial cosmology (Section 4.3). These errors therefore account only for statistical fluctuations given the model, not for the uncertainty in the redshift distribution. Since §5.1 demonstrates that the inferred bias depends strongly on σz in the high-z bins, the D̂G errors are likely understated; at minimum, the photo-z contribution to the A_D error should be estimated by repeating the D̂G pipeline for the σz = 0.04 and σz = 0.0 cases or with a more realistic photo-z error model derived from the spectroscopic training sample.","section":"§4.2 and §5.3"}],"minor_comments":[{"comment":"The word \"Tomographic\" appears as \"T omographic\" in the title and abstract; in addition, the sentence in §5.3 beginning \"Wecanassesstheamplitudeofthelineargrowthfunction...\" is missing spaces between words.","section":"Title and Abstract"},{"comment":"The phrase \"we set the lowest value of ell on ell_min = 20\" should read \"we set the lowest value of ell to ell_min = 20\".","section":"§4.1"},{"comment":"The galaxy counts in Table 1 (e.g., \"2,208869\", \"3,178981\") lack commas in the thousands separators, which makes the numbers difficult to read.","section":"Table 1"},{"comment":"The comparison between jackknife and Monte Carlo covariances is valuable, but the sentence explaining that the JK off-diagonal terms \"may incorporate also the non-Gaussian variance produced on small scales by the nonlinear evolution\" is awkward; please rephrase for clarity. Additionally, the D̂G weighting in Eq. (4.8) appears to use the analytic errors of Eq. (4.6) while the parameter fits in §5.1 use the JK covariance; the paper should state explicitly which errors are used in the D̂G weighting and justify that choice.","section":"§4.3 and Appendix A"}],"recommendation":"major_revision","confidential_remarks":"The paper is generally careful: the pseudo-C_l treatment, covariance validation, and null tests are solid, and the result is of interest to the JCAP readership. The main concern is that the headline A_D = 1.16 ± 0.13 is derived from bins where the assumed dn/dz is recognized to be inadequate, and the photo-z uncertainty is not propagated into A_D. This is fixable within the manuscript's scope by re-deriving D̂G and A_D under alternative photo-z error models and reporting the sensitivity. If the shift in A_D is small relative to 0.13, the paper can proceed; if it is comparable to or larger than the statistical error, the conclusions need to be substantially qualified."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a careful, competent application of Giannantonio's D_G estimator to a new photometric catalogue. The headline result, A_D = 1.16 ± 0.13, is a plausible consistency check with ΛCDM, but it is only as good as the assumed photo-z distribution, and the paper's own tests show that distribution is shaky at the high-z end.\n\nWhat's new: the first tomographic D_G measurement using the SCUSS+SDSS+WISE photo-z catalogue, with per-bin galaxy bias and lensing amplitude fits. The estimator itself is established, so novelty is modest. The analysis is solid: MASTER pseudo-C_l, JK covariance validated against MC and analytic, null tests with 300 simulated lensing maps, and tSZ/extinction checks. They also honestly report a ~3.5σ A < b discrepancy in the full sample and show it is driven by the high-z bins. That is good practice.\n\nSoft spots, in proportion. The reconstructed dn/dz uses a single Gaussian photo-z error, σz = 0.019, and Section 5.1 admits the catalogue photo-zs are underestimated at z ≳ 0.6 because of the training set. The two highest bins have poor PTE in the auto-spectrum (0.59% and 0.006%). When the authors vary σz, the inferred bias changes by up to 30%. D_G is bias-independent but not dn/dz-independent; the theoretical spectra used to normalize it are computed from the assumed dn/dz. The headline A_D is a single amplitude fit to all six bins, so it absorbs any systematic offset in those two bins. The three-lowest-bin fit gives 1.22 ± 0.19, so the qualitative conclusion is stable, but the quoted 0.13 error likely underestimates the systematic. Also, no code or numerical D_G values are released, which makes the specific numbers hard to reuse.\n\nThe circularity worry is not real: normalizing by the fiducial model makes this a null test, and the data can in principle disagree. The paper does not.\n\nWho is this for? People working on CMB lensing tomography and growth-rate measurements. It is a useful consistency check, not a breakthrough. I would cite it, and I think a serious editor should send it to a referee: the analysis is careful, the limitations are openly discussed, and the main issue is fixable—re-derive A_D with an empirical photo-z error model or at least propagate the dn/dz uncertainty into the headline error. Recommend that, plus a data release of the D_G values.","headline":"A careful, competent tomographic D_G measurement on a new photometric catalogue; the headline A_D = 1.16 ± 0.13 is plausible but rests on a single-Gaussian photo-z error model that the paper itself shows is shaky at z > 0.5.","tokens_in":22960,"tokens_out":2714,"would_cite":true,"duration_ms":24571,"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":"The linear growth of structure between redshifts 0.1 and 0.7 is consistent with the ΛCDM prediction, at measured amplitude $A_D = 1.16 \\pm 0.13$.","keywords":["cosmic microwave background lensing","galaxy clustering","tomographic cross-correlation","linear structure growth","galaxy bias","photometric redshifts","Lambda CDM","angular power spectrum"],"falsifier":"Take a large spectroscopic subsample of the same galaxy catalogue in the $0.5<z<0.7$ bins, use the empirical photo-z error distribution in place of the assumed Gaussian with $\\sigma_z=0.019$, rebuild $dn/dz$, and refit $A_D$; a shift larger than about $1\\sigma$ away from $A_D=1.16$ would falsify the claimed consistency.","tokens_in":21861,"feed_emoji":"🌌","tokens_out":11194,"duration_ms":105051,"temperature":0.7,"pith_summary":"This paper combines the auto-correlation of a 20-million-galaxy photometric sample with its cross-correlation against CMB lensing, in six redshift slices from $0.1<z<0.7$, to measure how fast cosmic structure grows. The key quantity is a bias-independent estimator $\\hat{D}_G(z)$ built from the ratio of observed to theoretical galaxy and galaxy-lensing power spectra. The paper finds the growth amplitude relative to the standard cosmological model is $A_D=1.16\\pm0.13$, consistent with the $\\Lambda$CDM expectation $A_D=1$, and reports no significant systematic contamination. If this holds, the growth of density fluctuations in the recent universe follows the standard model, so tests of dark energy or modified gravity in this redshift range have no anomaly to explain.","feed_headline":"Structure growth matches ΛCDM at amplitude 1.16 ± 0.13","feed_subtitle":"A six-bin galaxy–CMB lensing cross-correlation finds no departure from standard growth.","key_machinery":"The load-bearing object is the $\\hat{D}_G$ estimator, a weighted ratio $$\\hat{D}_G = \\left\\langle \\frac{(C_\\$ell^{{\\kappa g}}$)_{\\rm obs}/(C_\\$ell^{{\\kappa g}}$)_{\\rm th}}{\\sqrt{(C_\\$ell^{{gg}}$)_{\\rm th}/(C_\\$ell^{{gg}}$)_{\\rm obs}}} \\right\\rangle_\\ell,$$ where the theoretical spectra are evaluated at $z=0$ with the growth factor removed. Because the galaxy auto-spectrum scales as $b^2 D^2$ and the cross-spectrum as $b D^2$, the combination cancels the linear galaxy bias $b$ and isolates the growth factor $D(z)$, normalized to $D(0)=1$. The machinery also includes the Limber-approximated power spectra, the linear and scale-independent bias model, jackknife and Monte-Carlo covariance estimates, and a conservative multipole cut where nonlinear corrections stay below 5%.","core_discovery":"On the paper's own terms, the central discovery is a tomographic measurement of the linear growth factor $D(z)$ that does not depend on knowing the galaxy bias. Using the auto-power spectrum of galaxy density and the cross-power spectrum with the Planck CMB lensing convergence, the authors derive $\\hat{D}_G$ in six bins and fit its overall amplitude against the fiducial $\\Lambda$CDM template, obtaining $A_D=1.16\\pm0.13$, in agreement with $A_D^{\\Lambda CDM}=1$. The galaxy bias fitted in each bin agrees with the cross-correlation amplitude within $1\\sigma$, indicating a lensing amplitude consistent with unity; a lower amplitude seen in the full non-tomographic sample is traced to the high-redshift bins, where the photometric redshift uncertainties are largest. Null tests and foreground checks show no significant contamination of the cross-correlation.","pith_inferences":["A direct spectroscopic calibration of the photo-z errors in the $0.5<z<0.7$ bins is the natural extension; given the paper's own finding that inferred bias shifts by up to 30% when $\\sigma_z$ is changed, such a calibration could move $A_D$ toward or away from 1.","Applying the same estimator to lensing maps from upcoming CMB experiments combined with wide photometric surveys should reduce the statistical error on $A_D$ well below 10%, turning the consistency test into a sharp growth-index measurement.","The estimator could also be cross-correlated with galaxy shear instead of density, providing an independent bias-free growth measurement that shares no photo-z-dependent selection with the galaxy density map."],"forward_implications":["If $A_D=1.16\\pm0.13$ is correct, no significant deviation from $\\Lambda$CDM growth exists in $0.1<z<0.7$; modified-gravity or dark-energy models that change $D(z)$ in this interval are constrained.","The bias-independent $\\hat{D}_G$ method yields usable growth points from photometric surveys alone, which otherwise cannot measure $f\\sigma_8$.","The per-bin agreement between galaxy bias and lensing amplitude supports treating the galaxy bias as linear and deterministic on the scales used.","The high-redshift-driven $A<b$ tension implies that tomographic treatment is preferable to a single-bin analysis when photometric redshift quality varies with redshift."],"supporting_citations":[{"why":"Introduces the bias-independent $\\hat{D}_G$ estimator that turns the auto- and cross-power spectra into a growth-factor measurement.","marker":"[1]"},{"why":"Provides the photometric redshift galaxy catalogue and the $\\sigma_z=0.019$ photo-z error model used to build $dn/dz$ in each bin.","marker":"[47]"},{"why":"Supplies the CMB lensing convergence map used for the cross-correlation and for the simulated null tests.","marker":"[22]"},{"why":"Defines the fiducial $\\Lambda$CDM parameters used for theoretical spectra and for the $A_D=1$ prediction.","marker":"[53]"},{"why":"Computes the matter power spectrum used to generate the theoretical auto- and cross-spectra.","marker":"[51]"},{"why":"Supplies the nonlinear correction prescription used to set the conservative multipole cuts.","marker":"[52]"}],"fun_headline_variants":["A_D=1.16±0.13: growth matches ΛCDM","Tomographic lensing-galaxy analysis sees standard growth","Lensing-clustering cross-correlation pins growth to ΛCDM","No deviation in linear growth: A_D=1.16±0.13","Galaxy-CMB lensing measures growth, A_D=1.16±0.13"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result rests on the assumption that the reconstructed redshift distribution in each bin is correct, with photometric redshift error $\\sigma_z=0.019$; if photo-z's are systematically underestimated at $z\\gtrsim0.6$, the theoretical spectra shift and the fitted growth amplitude changes.","fun_headline_variants_meta":{"raw":{"variants":["A_D=1.16±0.13: growth matches ΛCDM","Tomographic lensing-galaxy analysis sees standard growth","Lensing-clustering cross-correlation pins growth to ΛCDM","No deviation in linear growth: A_D=1.16±0.13","Galaxy-CMB lensing measures growth, A_D=1.16±0.13"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00158,"raw_usage":{"total_tokens":6302,"prompt_tokens":942,"completion_tokens":5360,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":558,"completion_tokens_details":{"reasoning_tokens":5260}},"tokens_in":558,"tokens_out":5360,"duration_ms":39642,"temperature":1.0,"reasoning_tokens":5260,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:30:39.397799+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a large spectroscopic subsample of the same galaxy catalogue in the $0.5<z<0.7$ bins, use the empirical photo-z error distribution in place of the assumed Gaussian with $\\sigma_z=0.019$, rebuild $dn/dz$, and refit $A_D$; a shift larger than about $1\\sigma$ away from $A_D=1.16$ would falsify the claimed consistency.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the photometric redshift galaxy catalogue and the $\\sigma_z=0.019$ photo-z error model used to build $dn/dz$ in each bin."},{"cited_title":"Lewis and A","cited_arxiv_id":null,"evidence_quote":"Computes the matter power spectrum used to generate the theoretical auto- and cross-spectra."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the nonlinear correction prescription used to set the conservative multipole cuts."}],"review_version":1}