{"id":"80e95e1e-ddd8-42ab-b473-d8b2254ca206","arxiv_id":"1908.10309","paper_version":3,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"LoTSS-DR1 radio source statistics are non-Poissonian at the counts-in-cell level but isotropic at the percent level and consistent with Planck 2018 cosmology above 2 mJy.","lead":"This paper measures how radio sources detected by the LOFAR Two-metre Sky Survey first data release are distributed on the sky, using counts per cell, source counts as a function of flux, and the angular two-point correlation function. The result is a validation that the radio sky is statistically isotropic at the percent level and consistent with the standard cosmological model, while also identifying calibration systematics that limit the faintest samples.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Photo-z error propagation is the weak link: the parameter-free agreement in Sec. 6.3 can be fortuitous if z_best is biased, and only ~48% of sources have redshifts.","rationale":"The reader's weakest assumption (photo-z accuracy and representativeness) is the same load-bearing concern I identify. The paper's own disclaimers in Secs. 6.2 and 7 confirm the missing error propagation and the uncertainty about the other half of the sources. The rest of the analysis—completeness, source counts, and the 2 mJy mask-1 w(theta) measurement—is well supported and honestly caveated, so the verdict remains CONDITIONAL rather than escalating. My concrete test would settle whether the agreement is robust to photo-z errors; until that is done, the conditional verdict is appropriate.","tokens_in":44206,"tokens_out":7855,"duration_ms":81926,"concrete_test":"Recompute the CAMB prediction for the 2 mJy mask-z1 'Anyz' sample after convolving the observed N(z) with a realistic photometric redshift error model (e.g., sigma_z = 0.05(1+z) plus a 5% catastrophic outlier fraction, as in Duncan et al. 2019), and compare the predicted w(theta) with the measured values over 0.1-6 deg. If the prediction shifts by more than the quoted measurement errors, or if the chi-squared worsens by more than ~1 per degree of freedom, the agreement in Fig. 25 is not robust to photo-z uncertainties.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The Sec. 6.3 comparison uses the z_best histogram as the source window function in CAMB without propagating photometric redshift errors; the authors state in Sec. 6.2 that 'we did not estimate and propagate errors on the redshift estimation.' A biased or smeared N(z) directly shifts the predicted w(theta), so the parameter-free agreement could be fortuitous. Additionally, only 47.9% of the 2 mJy mask-z sources have z_best, and the paper acknowledges (Sec. 7) that it is unclear how representative these are of the full sample. If the half without redshifts has a different redshift distribution and bias, the 'Anyz' subsample consistency does not establish that LoTSS-DR1 as a whole is consistent with the standard model. The disagreement with the NVSS bias model at low z (Fig. 26, needing b1=1.2 instead of the model's 1.6 at z=0) further shows that the agreement in the full 'Anyz' sample may be an average that masks population-dependent failures. The central measurement is solid, but the cosmological consistency claim is conditional on the unverified photo-z distribution.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses the LOFAR Two-metre Sky Survey first data release (LoTSS-DR1) to measure one- and two-point statistics of 144 MHz radio sources over 424 square degrees. The authors estimate point-source completeness by injecting sources into residual maps, define a set of masks and flux-density thresholds, and find that the counts-in-cell distribution is not Poissonian and is well described by a compound Poisson distribution. The differential source counts agree with earlier low-frequency surveys and with the SKADS and T-RECS simulations. The angular two-point correlation function, estimated with the Landy-Szalay estimator and TreeCorr, is fitted by a power law with best-fit values A = (5.1 +/- 0.6) x 10^-3 and gamma = 0.74 +/- 0.16 for the 2 mJy, low-noise ('mask 1') sample. The paper then compares w(theta) in the redshift-selected 'Anyz' sample with CAMB predictions based on Planck 2018 parameters, the z_best photometric-redshift histogram, and the Nusser and Tiwari (2016) bias model, reporting agreement between 0.1 and 6 degrees and concluding that the radio sky is statistically isotropic at the percent level.","tokens_in":44397,"tokens_out":8769,"duration_ms":83576,"significance":"If the conclusions hold, this paper establishes LoTSS-DR1 as a usable cosmological data set and provides one of the first robust measurements of the angular two-point correlation function at 144 MHz over a few hundred square degrees. The data-quality analysis is a clear strength: completeness is assessed by injecting sources and re-running PyBDSF, the random catalogues are built from local rms noise maps, five estimators are compared in Appendix B, and TreeCorr is validated against a brute-force code in Appendix C. The authors are also commendably explicit about their limitations. However, the headline cosmological consistency claim is conditional: the theoretical prediction depends on an external bias model and on the photometric-redshift distribution of only about half of the sources, with no propagation of photo-z errors. The measurement itself is solid, but the cosmological interpretation needs to be either strengthened by sensitivity tests or clearly downgraded to a consistency check under stated assumptions.","major_comments":[{"comment":"The central claim that the measured w(theta) 'agrees well with the expectation of the cosmological standard model' is not yet established at the claimed precision, because the theoretical curve is computed from the z_best histogram as the source window function without propagating photometric-redshift errors, and with the external bias model b(z) = 1.6 + 0.85z + 0.33z^2 from Nusser and Tiwari (2016). The authors state in Section 6.2 that 'we did not estimate and propagate errors on the redshift estimation'; a biased or smeared N(z) shifts the predicted w(theta) directly, so the agreement shown in Fig. 25 could be fortuitous. I request a sensitivity analysis, such as convolving N(z) with the photo-z error distribution or repeating the prediction with an alternative redshift or bias model, and a correspondingly softened statement in the abstract and conclusions.","section":"Section 6.3, Eqs. (14) and (32), Fig. 25"},{"comment":"Only 24,420 out of 50,977 sources above 2 mJy in mask z (47.9%) have a z_best value, and Section 7 explicitly states that it is currently unclear how representative these sources are of the full sample. The agreement of the clustering parameter nc between sources with and without redshifts (Fig. 18) is a one-point statistic and does not guarantee that the two-point clustering of the 'Anyz' subsample equals that of the full LoTSS-DR1 population. To support the conclusion that LoTSS-DR1 sources as a whole are consistent with the standard model, the paper should either measure w(theta) for the No-z sample or model the selection function; otherwise the consistency statement should be restricted to the redshift-selected subsample.","section":"Section 5.3, Table 5, and Section 6.2"},{"comment":"The binned-redshift comparison shows that the adopted bias model fails in the lowest-redshift bin: the authors must reduce b1 from 1.6 to 1.2 to bring the model into agreement with the data. This demonstrates that the 'no adjusted parameters' agreement in the full Anyz sample is not a clean test of the standard model, since it can be an average over populations with different bias and redshift selection. The full-sample comparison should be presented as a consistency check under an assumed bias model, with the dependence of the conclusion on the bias assumption quantified.","section":"Section 6.3, Fig. 26"}],"minor_comments":[{"comment":"The phrase 'The deviation from a nmhg distribution' contains a typographical error; it should presumably read 'Poisson' or 'compound Poisson distribution'.","section":"Abstract and Section 7"},{"comment":"The text says the CAMB window function uses the observed redshift distribution 'for sources with z <= 2', but Fig. 16 shows counts up to z ~ 4; please clarify how the z > 2 tail is treated and whether omitting it could bias the prediction at small angular scales.","section":"Section 6.3"},{"comment":"The column labelled 'z' is used for both the redshift-binned rows and the rows with 'n.a.' for full samples; the caption should state explicitly that the reported A and gamma parameters are power-law fits to the angular correlation function, not constraints on cosmological parameters.","section":"Table 6"},{"comment":"The caption refers to 'see also text in Sec. 5.3' for mask z, but mask z is first described in Section 6.2; the cross-reference should be corrected.","section":"Fig. 5 caption"}],"recommendation":"major_revision","confidential_remarks":"This is a strong observational paper with a careful treatment of survey systematics, and the measurement part is publishable. My main concern is that the headline cosmological consistency claim is conditional on unverified photometric redshifts and an external bias model, as the authors themselves acknowledge. Adding the requested sensitivity tests and tightening the wording of the abstract and conclusions should be sufficient; I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this one if you care about radio continuum surveys as LSS tracers. It gives the first one- and two-point statistics from LoTSS DR1, and the central 2 mJy low-noise measurement is solid. The paper is unusually careful about systematics: completeness injections, mocks built from the local rms noise, a comparison of five estimators, a TreeCorr accuracy check, and jack-knife regional splits. They identify flux-calibration problems at 1 mJy and don't paper over them; instead they restrict their headline result to the 2 mJy mask-1 sample, where the clustering amplitude is A=(5.1±0.6)e-3 and slope gamma=0.74±0.16. That's a real milestone for LOFAR/SKA pathfinder work.\n\nNow the soft spot, and it's the one you'd expect. The 'agreement with Planck 2018' in Sec 6.3 is parameter-free in the sense that they fix cosmology, use the measured z_best histogram as the window function, and use an off-the-shelf bias model from Nusser & Tiwari. But only about 48% of the 2 mJy sources have z_best, and the authors explicitly say they did not estimate or propagate photo-z errors. A biased or smeared N(z) changes the predicted w(theta) directly. And in Fig. 26 the NVSS-derived bias model overshoots the low-z bin (b1=1.2 fits better than 1.6), which suggests the agreement in the full 'Anyz' sample is an average over populations, not a clean validation of the standard model. The authors themselves flag all of this in Sec 6.2 and the conclusions. So the conditional reading is right: the clustering measurement is solid, the cosmological consistency claim is promising but not yet load-bearing.\n\nWho gets value? Cosmologists who want to know what LoTSS DR1 can already say about clustering; people using radio sources for ISW, dipole, or fNL forecasts; and anyone planning to use future LoTSS data releases—the masking and quality-control recipe here is directly reusable. The paper deserves a serious referee and, with minor revision to make the photo-z caveat more prominent in the abstract and conclusions, it should be published. I would bring it to the reading group, and I'd cite the 2 mJy clustering result.\n\nRecommended decision: accept with revisions, not a desk reject.","headline":"A careful first clustering measurement from LoTSS-DR1; trust the 2 mJy result, but the Planck agreement remains conditional until photo-zs are validated.","tokens_in":45050,"tokens_out":4547,"would_cite":true,"duration_ms":43513,"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":"A clean 2 mJy sample of 19,719 LoTSS radio sources shows angular clustering consistent with the standard cosmological model and with percent-level statistical isotropy of the radio sky.","keywords":["angular correlation function","counts-in-cells","radio source counts","LOFAR survey","photometric redshifts","large-scale structure","statistical isotropy","cosmological principle"],"falsifier":"Recompute the predicted $w(\\theta)$ for the 2 mJy low-noise sample using a spectroscopic redshift distribution rather than the photometric one, and check whether the prediction still lies inside the measured 1-$\\sigma$ band between 0.2 and 2 degrees; if it drifts outside, the claimed agreement depended on the assumed $N(z)$.","tokens_in":43959,"feed_emoji":"📡","tokens_out":11998,"duration_ms":118088,"temperature":0.7,"pith_summary":"The paper uses the first data release of the LOFAR Two-metre Sky Survey, covering 424 square degrees and 318,520 sources, to ask whether the large-scale distribution of faint radio sources matches the standard cosmological model. It establishes that the angular two-point correlation function of a clean, low-noise subsample above 2 mJy is consistent with the prediction built from the best-fit Planck 2018 cosmology, the photometric redshift distribution of the sources, and a literature bias model, with no fitted parameters. It also shows that counts-in-cells reject a Poisson distribution and are well described by a compound Poisson distribution, and that differential source counts agree with earlier low-frequency surveys and with two independent simulations of the radio sky. The paper's conclusion is that the radio sky is statistically isotropic at the percent level and that current-generation radio surveys can already serve as cosmological probes.","feed_headline":"LoTSS radio sky matches the standard model's clustering","feed_subtitle":"A clean 2 mJy sample of 19,719 sources shows clustering amplitude near 0.005 at one degree, matching Planck cosmology.","key_machinery":"The argument pivots on two objects. For the one-point statistics it is the counts-in-cell distribution and its moments, especially the clustering parameter $n_c=\\mathrm{Var}[k]/\\mathbb{E}[k]$, with a compound Poisson distribution (a Poisson number of clusters, each contributing a Poisson number of components) as the model that fits the data. For the two-point statistics it is the angular correlation function $w(\\theta)$, measured through the Landy-Szalay pair-counting estimator $\\hat w(\\theta)=(DD-2DR+RR)/RR$, where $DD$, $DR$, and $RR$ are normalized data-data, data-random, and random-random pair counts. The random catalogue is a noise-weighted mock built from a simulated radio sky and the actual LoTSS noise maps, so survey masks and completeness variations are folded into the null expectation. The theoretical comparison carries through a redshift-space window: the measured photo-$z$ histogram, the Planck 2018 parameters, and a bias-redshift relation are fed into a Boltzmann-solver code that produces the predicted $w(\\theta)$, with the finite-survey integral constraint subtracted.","core_discovery":"The paper's central claim is that the angular clustering of LoTSS-DR1 radio sources above a flux-density threshold of 2 mJy, restricted to cells with local rms noise below the survey median, is described by $w(\\theta)=A(\\theta/1\\,\\mathrm{deg})^{-\\gamma}$ with $A=(5.1\\pm0.6)\\times10^{-3}$ and $\\gamma=0.74\\pm0.16$, and that between 0.1 deg and 6 deg this matches the theoretical $w(\\theta)$ computed from the Planck 2018 best-fit cosmology, the observed photometric redshift distribution, and the bias relation $b(z)=1.6+0.85z+0.33z^2$, without adjusting any model parameter. The same analysis finds $w(\\theta)<10^{-2}$ at angular scales above 1 deg and a radio sky that is statistically isotropic at the percent level. A second, independent claim is that the counts-in-cells are non-Poissonian: a compound Poisson distribution fits the data well, with a clustering parameter $n_c$ that exceeds unity at low flux densities and approaches unity above roughly 1 mJy.","pith_inferences":["I infer that the agreement with the standard model is only as strong as the photo-$z$ window: since half the sources lack redshifts and the paper does not propagate photo-$z$ errors, a selection bias in which sources receive a redshift could shift the predicted $w(\\theta)$ by more than the quoted uncertainties.","I read the low-redshift overprediction as evidence that mixed AGN and star-forming samples need separate bias functions; a direct extension would fit bias per redshift bin after accounting for photo-$z$ scatter.","A testable corollary is that weighting sources by photo-$z$ uncertainties, or repeating the analysis with a spectroscopic subsample as it becomes available, should keep the predicted curve inside the measured band if the claimed agreement is robust."],"forward_implications":["The most reliable clustering measurement is the 2 mJy, low-noise sample, with amplitude $A=(5.1\\pm0.6)\\times10^{-3}$ and slope $\\gamma=0.74\\pm0.16$ at 1 deg.","Below 2 mJy, pointing-to-pointing flux-scale errors contaminate the correlation function; above 4 mJy, shot noise dominates.","The counts-in-cell distribution is non-Poissonian and well fitted by a compound Poisson distribution, with $n_c$ above unity at sub-mJy fluxes.","The radio sky is statistically isotropic at the percent level for angular scales above 1 deg.","With the larger sky coverage and better flux calibration of future LoTSS releases, the same statistical pipeline can begin to constrain cosmological parameters such as $\\sigma_8$ and the evolution of radio-source bias."],"supporting_citations":[{"why":"Supplies the LoTSS-DR1 images, the 325,694-source catalogue, and the 20 per cent flux-density scale uncertainty.","marker":"Shimwell et al. 2019"},{"why":"Constructs the value-added source catalogue, including component grouping and artefact removal, which this analysis measures.","marker":"Williams et al. 2019"},{"why":"Provides the photometric redshift estimates ('z_best') used for the redshift window and redshift-split samples.","marker":"Duncan et al. 2019"},{"why":"Fixes the cosmological parameters ($H_0$, $\\Omega_b h^2$, $\\Omega_c h^2$, $A_s$, $n_s$) used in the theoretical prediction.","marker":"Planck Collaboration et al. 2018a,b"},{"why":"Introduces the bias parametrisation $b(z)=1.6+0.85z+0.33z^2$ used in the theoretical $w(\\theta)$.","marker":"Nusser & Tiwari 2015"},{"why":"Extends and applies the NVSS-based bias model adopted in the comparison.","marker":"Tiwari & Nusser 2016"},{"why":"Supplies the minimal-variance estimator used for all angular correlation measurements.","marker":"Landy & Szalay 1993"},{"why":"Provides the Boltzmann-solver code that converts cosmology, $N(z)$, and bias into the predicted angular correlation.","marker":"Challinor & Lewis 2011"},{"why":"Supplies the method for building noise-weighted mock catalogues from simulated skies and LoTSS rms noise maps.","marker":"Hale et al. 2018"},{"why":"Gives the SKADS simulated sky used to assign realistic flux densities to the mock catalogue.","marker":"Wilman et al. 2008"}],"fun_headline_variants":["LoTSS radio clustering matches Planck universe","No free parameters: LOFAR clustering matches Planck","Compound Poisson fits LOFAR counts, clustering fits Planck","Radio sky clustering matches standard model to percent","LOFAR sources cluster as Planck cosmology predicts"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The standard-model comparison stands or falls on the assumption that the photometric redshifts for the roughly half of sources that have them faithfully represent the true redshift distribution of the 2 mJy sample, since the paper does not propagate redshift errors or account for the missing half.","fun_headline_variants_meta":{"raw":{"variants":["LoTSS radio clustering matches Planck universe","No free parameters: LOFAR clustering matches Planck","Compound Poisson fits LOFAR counts, clustering fits Planck","Radio sky clustering matches standard model to percent","LOFAR sources cluster as Planck cosmology predicts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000876,"raw_usage":{"total_tokens":3888,"prompt_tokens":1139,"completion_tokens":2749,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":755,"completion_tokens_details":{"reasoning_tokens":2677}},"tokens_in":755,"tokens_out":2749,"duration_ms":19485,"temperature":1.0,"reasoning_tokens":2677,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:45:53.526675+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the predicted $w(\\theta)$ for the 2 mJy low-noise sample using a spectroscopic redshift distribution rather than the photometric one, and check whether the prediction still lies inside the measured 1-$\\sigma$ band between 0.2 and 2 degrees; if it drifts outside, the claimed agreement depended on the assumed $N(z)$.","supporting_citations":[{"cited_title":"L., Hardcastle , M","cited_arxiv_id":null,"evidence_quote":"Constructs the value-added source catalogue, including component grouping and artefact removal, which this analysis measures."},{"cited_title":"J., Sabater , J., R \\\"o ttgering , H","cited_arxiv_id":null,"evidence_quote":"Provides the photometric redshift estimates ('z_best') used for the redshift window and redshift-split samples."},{"cited_title":"& Tiwari , P","cited_arxiv_id":null,"evidence_quote":"Introduces the bias parametrisation $b(z)=1.6+0.85z+0.33z^2$ used in the theoretical $w(\\theta)$."},{"cited_title":"L., Jarvis , M","cited_arxiv_id":null,"evidence_quote":"Supplies the method for building noise-weighted mock catalogues from simulated skies and LoTSS rms noise maps."},{"cited_title":"J., Miller , L., Jarvis , M","cited_arxiv_id":null,"evidence_quote":"Gives the SKADS simulated sky used to assign realistic flux densities to the mock catalogue."}],"review_version":1}