{"id":"96a1fd64-6006-4724-85b5-db3442691ee0","arxiv_id":"2412.00475","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Ten regions of the FAST All Sky HI Survey show two-point angular correlation functions consistent with cosmic isotropy within 2 sigma.","lead":"Using the first release of the FAST All Sky HI Survey, the authors test whether neutral hydrogen galaxies trace the universe isotropically across ten sky regions and find all regions consistent with isotropy within 2 sigma. The result validates the FASHI catalog for clustering studies, but the conclusion depends on a post hoc sensitivity cut and mock skies calibrated with the same data.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Null mocks omit the FASHI sensitivity map, so the per-region covariance and 2σ isotropic-consistency claim are not yet validated.","rationale":"The paper is a transparent null test, and the per-region 2PACF pipeline is standard. The reader's main concern is the post hoc sensitivity window and the use of the fitted b0 in the mocks; I agree that this is a weakness, but I think the more load-bearing and concrete issue is that the mocks do not appear to include the FASHI detection-sensitivity map at all. Evidence: after applying the [0.65,1.0] cut, the ten equal-area regions still have source densities from 1.07 to 5.32 deg^-2 (Table 2), a factor-of-five spread that cannot be Poisson noise and must be dominated by the survey's spatially varying sensitivity (the very effect the paper sets out to mitigate). Section 3.1 describes the mocks only as having the same number of sources and sky coverage, with no mention of a completeness or sensitivity map, so the covariance matrices from Eq. (9) will not include the dominant noise term in the real data. The 2σ consistency statement in Section 4 therefore rests on a null distribution that does not match the survey's selection function. This is testable: generate mocks with a position-dependent completeness map and see if the Table 3 deviations change. If they do not change, the isotropy claim is robust; if they do, the paper needs to be revised. Because this is an addressable modeling issue rather than a fatal error in the 2PACF methodology, the conditional verdict remains appropriate.","tokens_in":12550,"tokens_out":11647,"duration_ms":113523,"concrete_test":"Re-run the FLASK mock generation using a position-dependent completeness map derived from the FASHI RMS/sensitivity map (or the empirically measured detection probability versus position) rather than a uniform angular density, keeping the same total N, sky mask, and [0.65,1.0] cut; recompute the per-region 2PACF covariance and the Table 3 β deviations. If the maximum |β deviation| or the 2σ classifications change, the isotropy claim is not robust to the survey selection function; if they are unchanged, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"To support the 2σ isotropy claim, the null distribution for the 2PACF in each region must include all survey effects. The FLASK mocks in §3.1 are generated with 'the same number of sources and sky coverage' as the filtered sample, but the filtered sample still has large regional density variations: Table 2 shows source densities ranging from 1.07 deg^-2 (Area 1, 675 sources) to 5.32 deg^-2 (Area 7, 3342 sources) across equal-area 627 deg^2 regions. These variations are far larger than Poisson noise and reflect the residual FASHI sensitivity pattern that the [0.65,1.0] window does not remove. If the mocks assign sources uniformly over the footprint (or only follow the sky mask), they do not reproduce the per-region shot noise or the sensitivity-induced covariance, so the covariance matrices computed from Eq. (9) and the 'β std' values in Table 3 do not correspond to the actual noise in the data. The reported 2σ consistency is therefore not a valid test of isotropy until the sensitivity map is included in the mock generation. This is compounded by the post hoc choice of the window [0.65,1.0] to make b0 match the expected HI bias and by using that fitted b0=1.02 as input to the mocks.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper tests cosmic isotropy using the first data release of the FAST All Sky HI Survey (FASHI). Because the survey is operated in a schedule-filler mode, its detection sensitivity is strongly position-dependent, and the authors argue that this induces spurious clustering. They restrict the catalog (z<0.09, 30°<DEC<64°) to a narrow sensitivity range [0.65,1.0] mJy, leaving 19,040 sources over 6,277 deg^2, and show via the angular power spectrum that this filtered sample has bias b0=1.02±0.06, consistent with external HI bias estimates, with shot noise consistent with ΔΩ/N. The footprint is then divided into ten regions of roughly 627 deg^2 each. In each region the two-point angular correlation function (2PACF) is measured over 0.5°<θ<10° and fitted with a power law ξ(θ)=(θ/θ0)^{-β}; the per-region fitted parameters are compared with distributions from 100 FLASK lognormal mock catalogs generated under homogeneity and isotropy with the same fitted b0. The paper concludes that all ten FASHI regions are consistent with statistical isotropy at the 2σ level, noting that Areas 5 and 10 show the largest deviations in β (1.34σ and -1.10σ, respectively) and that the unfiltered catalog shows larger regional fluctuations.","tokens_in":12796,"tokens_out":23863,"duration_ms":224643,"significance":"If the central claim is correct, the paper provides a new, independent HI-based consistency test of the Cosmological Principle and, more durably, a practical recipe for using the first FASHI release despite its strongly non-uniform sensitivity. Strengths: the analysis uses public data and public codes (NaMaster, PolSpice, FLASK, emcee, Colossus); the sensitivity-induced clustering is demonstrated directly in Figure 3 and Table 1; the shot-noise cross-check C_SN ≈ ΔΩ/N validates the power-spectrum fit; and the authors transparently report the unfiltered results in parallel with the filtered ones, including a cross-check of the anomalous Area 10 against ALFALFA Area-V. The isotropy claim itself is a null result consistent with the standard cosmological model, so the significance is moderate rather than high. The quantitative 2σ statement is, however, conditional on the mock construction and the sensitivity-window choice discussed in the major comments.","major_comments":[{"comment":"The null mocks as described do not reproduce the per-region noise of the data, so the reported uncertainties are not calibrated. The FLASK mocks are generated 'with the same number of sources and sky coverage as the real selected FASHI subsample,' but no position-dependent selection function or sensitivity map is listed among the inputs. Table 2 shows that the ten equal-area (627 deg^2) regions contain between 675 and 3,342 sources (densities 1.07-5.32 deg^-2), a factor-of-five range that cannot be Poisson noise; whether this reflects residual sensitivity selection or genuine large-scale structure, the mocks must reproduce it for the regional comparison to be valid. If sources are drawn uniformly over the footprint, each mock region contains roughly 1,900 sources, and the covariance matrix of Eq. (9) and the MCMC errors quoted in Table 3 and Figure 4 are computed for regions of mean density rather than for the data's actual density. In the sparse regions (e.g., Area 1, with 675 sources), the shot-noise contribution to the 2PACF variance is about an order of magnitude larger than in the mocks, so the per-region error bars and the 'β std' column of Table 3 are not a valid calibration for the data. The 2σ isotropy claim should be re-derived with mocks that include the actual FASHI sensitivity or an empirical per-region selection function, and the mock covariance should be validated against the data before it is used in the likelihood.","section":"§3.1, Eq. (9), Table 2"},{"comment":"The sensitivity window is selected post hoc, and the null hypothesis is calibrated with the resulting fit, making the test partially circular. Table 1 shows the fitted bias decreasing monotonically from b0=1.22±0.05 for the full sample to 0.51±0.09 for the range [0.70,0.93]; the adopted range [0.65,1.0] is the one that returns b0=1.02±0.06, matching the external expectation for HI sources, and the text states that this window is chosen 'to ensure accurate representation of the clustering properties.' A small shift of the window bounds changes the inferred clustering amplitude by a factor of two, so the analysis is extremely sensitive to the cut, and any genuine anisotropy that manifests as a shift in the apparent bias across windows would be absorbed by the selection itself. The 2PACF mocks are then generated with the very same fitted value, b0=1.02, so the comparison in Table 3 tests regional scatter around a data-calibrated mean rather than agreement with an externally predicted clustering amplitude. The paper should demonstrate that the 2σ conclusion is robust across the windows listed in Table 1 and should regenerate the null using an externally fixed bias (for instance b0=1.0 from Martin et al. 2012), quantifying how the β std values and the MCMC uncertainties change.","section":"§2.2-§2.3, §3.1, Table 1"}],"minor_comments":[{"comment":"The sentence '...resulting in a total of 1,000 regions for analysis' is attached to the definition of the power-law fit; it appears to be a leftover from the mock description (100 mock catalogs times 10 regions) and should be moved or deleted.","section":"§3.2"},{"comment":"The caption states '95% C.L.' while Section 3.2 describes 68% C.L. constraints, and the Mock row lists uncertainties twice as large as the 68% values quoted in the text; the convention should be made consistent and explicit.","section":"Table 3"},{"comment":"There is an unclosed parenthesis in 'downgrade the count map to a lower resolution of (Nside = 64 and apply an apodization method.'","section":"§2.2"},{"comment":"The bullet stating the angular separation range as '0° < θ < 10°' conflicts with the 0.5° lower bound used in Section 3.1 and the abstract.","section":"§4"},{"comment":"The 'β std' column is not defined: state whether σ is the scatter of the 1,000 mock region fits or the MCMC uncertainty, and specify the reference value used (global mock mean or per-region mean).","section":"Table 3"},{"comment":"Clarify the quantity to which the [0.65,1.0] cut is applied: the parameter RMS defined in Eq. (1) carries a high-precision numerical factor (9.529287035639559) whose origin and units are not explained, whereas the text's quoted median sensitivity of 0.76 mJy refers to σrms; the cut variable and its units should be stated unambiguously.","section":"§2.1, Eq. (1)"},{"comment":"The 19-bin covariance matrix is estimated from only 100 mock realizations and then inverted for the likelihood; the inverse covariance should be debiased with the Hartlap factor (or its stability tested), and this should be stated.","section":"§3.1"}],"recommendation":"major_revision","confidential_remarks":"The two major concerns are both repairable within the scope of the paper: rebuild the mocks so that they include the FASHI sensitivity/selection pattern, and demonstrate that the isotropy conclusion is robust to the window choice and to an externally fixed b0. If the revision addresses only presentation, I would not support acceptance, because the covariance calibration directly controls the quantitative 2σ statement. I have no concerns about novelty disclosure or citation practice: the dependence on Franco et al. (2023) is explicit and appropriate. The transparent parallel reporting of the unfiltered results is a genuine strength and should be kept."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the first isotropy test using the FASHI catalog, and the sensitivity-filtering recipe in Section 2.3 is the most durable contribution. Showing how the inferred HI bias changes from 1.22 down to 0.51 as you narrow the sensitivity window, and then picking the window that gives b0 = 1.02, is a practical warning for anyone using FASHI for clustering. The 2PACF analysis itself is standard, following Franco et al. (2023) closely, but that is not a flaw if the new data application is handled cleanly. It mostly is, and the paper is transparent about its choices.\n\nWhere it gets soft is the mock catalogs. The stress-test concern is on target: the FLASK mocks are generated with the same number of sources and sky coverage, but Table 2 shows per-region densities ranging from 1.07 to 5.32 deg^-2 across equal-area regions. Those fluctuations are far larger than Poisson noise and reflect the residual FASHI sensitivity pattern. If the mocks do not reproduce that pattern, the per-region covariance from Eq. (9) is too small, and the reported 2-sigma consistency is not a valid test of isotropy. That is a real, load-bearing problem, not a cosmetic one. On top of that, the sensitivity window [0.65, 1.0] is chosen post hoc to match the expected HI bias, and the same fitted b0 = 1.02 is fed into the mocks as the null hypothesis. That makes the null self-calibrated. The paper is honest about both choices, but honesty does not remove the circularity.\n\nI would stop short of calling the result wrong. The filtered data very likely are less contaminated than the raw catalog, and the null result is consistent with standard cosmology. But the strength of the claim — no statistically significant deviations — is conditional on the sensitivity map being irrelevant after filtering, which the paper does not demonstrate. The original data show larger fluctuations, and that is exactly what you would expect if selection effects are still present.\n\nWho gets value from this: anyone preparing to use FASHI for clustering, and people testing isotropy with radio surveys. It deserves a serious referee, but the referee should demand that the mocks include the FASHI sensitivity map, that the sensitivity window choice be justified without reference to the fitted bias, and that the isotropy result be reported with the mocks that actually match the regional noise. With those revisions, the paper would be a solid catalog-validation plus null test.","headline":"First FASHI isotropy test with a useful sensitivity-filtering recipe, but the null mocks omit the sensitivity map and the filtering window is post hoc; worth peer review, conditional result.","tokens_in":13357,"tokens_out":1930,"would_cite":false,"duration_ms":21837,"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 local Universe, traced by neutral hydrogen galaxies in the FAST All Sky HI Survey, shows no statistically significant deviation from cosmic isotropy.","keywords":["cosmic isotropy","two-point angular correlation function","neutral hydrogen galaxies","FASHI survey","Cosmological Principle","large-scale structure","angular power spectrum","MCMC parameter estimation"],"falsifier":"Measure the angular power spectrum dipole and quadrupole of the filtered 19,040-source FASHI sample: a statistically significant nonzero dipole would contradict the isotropy claim. Alternatively, re-run the ten-region 2PACF analysis with the sensitivity window fixed in advance and check whether the scatter in $\\beta$ across regions exceeds the mock scatter at more than $2\\sigma$.","tokens_in":12303,"feed_emoji":"🌌","tokens_out":5936,"duration_ms":49209,"temperature":0.7,"pith_summary":"This paper tests whether the local Universe is isotropic using the first release of the FAST All Sky HI Survey (FASHI), the largest catalog of extragalactic neutral hydrogen sources. Because FASHI's detection sensitivity varies widely across the sky, the authors restrict attention to the 19,040 sources with sensitivity in [0.65, 1.0] mJy, a window chosen so the measured linear bias matches the expected HI bias. They divide this sample into ten sky regions, measure the two-point angular correlation function in each, and fit a power law with MCMC. Comparing the ten regions against lognormal mock catalogs built under homogeneity and isotropy, they find all regions consistent with statistical isotropy within $2\\sigma$. The conclusion is that the HI-selected local Universe shows no statistically significant deviation from the Cosmological Principle.","feed_headline":"HI survey sees no break from cosmic isotropy","feed_subtitle":"Ten sky regions in the largest neutral-hydrogen catalog match isotropic mocks within 2 sigma, testing the Cosmological Principle.","key_machinery":"The central object is the two-point angular correlation function (2PACF), measured with the Szapudi\\textendash{}Szalay estimator over angular scales $0.5^\\circ < \\theta < 10^\\circ$, and fit to a power law $\\xi(\\theta) = (\\theta/\\theta_0)^{-\\beta}$. The load-bearing device is the sensitivity selection: restricting the catalog to detection sensitivities in [0.65, 1.0] mJy removes artificial clustering introduced by the survey's schedule-filler observing mode. The ten region-by-region power-law fits are compared against 100 lognormal mock catalogs generated under homogeneity and isotropy, and consistency is judged by the scatter of the fitted parameters.","core_discovery":"The paper's central claim is that the local Universe, as traced by neutral hydrogen galaxies in FASHI, is statistically isotropic: the best-fit power-law parameters ($\\theta_0$, $\\beta$) of the two-point angular correlation function in each of ten regions of roughly 600 square degrees agree with one another and with mock catalogs within $2\\sigma$. The median values $\\theta_0 \\approx 0.35^\\circ$ and $\\beta \\approx 1.36$ match the isotropic-mock reference values. After the sensitivity cut, the sample has linear bias $b_0 = 1.02 \\pm 0.06$, consistent with expected HI bias and with the ALFALFA survey. The authors note that Area-10's $\\beta$ deviates by more than $1\\sigma$ from the mock expectation, hinting at possible large-scale structure, but they do not treat this as a detection of anisotropy.","pith_inferences":["A stronger isotropy test would measure the dipole and quadrupole of the source count map directly; the 2PACF comparison here is sensitive to scale-dependent clustering but not to a uniform dipole anisotropy.","The sensitivity window [0.65, 1.0] mJy was chosen after seeing that it yields the expected bias; an a priori choice or a scan over many windows with corrected significance would make the isotropy claim more robust.","Combining FASHI with ALFALFA data (which the paper says overlap near declination 30 degrees) would double the sky coverage and allow isotropy tests on scales beyond 10 degrees.","The full FASHI survey, with over 100,000 predicted sources, will shrink the covariance matrices enough to turn the current $2\\sigma$ consistency into a much tighter constraint, or expose a genuine deviation."],"forward_implications":["The filtered FASHI subsample, with bias $b_0 = 1.02 \\pm 0.06$, behaves like a standard HI tracer and is suitable for further cosmological clustering analyses.","The $2\\sigma$ consistency of all ten regions supports the Cosmological Principle at the scales and depths probed by FASHI (redshift $z < 0.09$).","The Area-10 $\\beta$ deviation, though below $2\\sigma$, points toward a possible super-void or super-cluster near that region that future surveys could confirm or refute.","The strong dependence of the angular power spectrum on the sensitivity window means future FASHI analyses must apply a similar selection to avoid false clustering signals."],"supporting_citations":[{"why":"Supplies the FASHI first-release catalog of 41,741 extragalactic HI sources and the sensitivity measurements used throughout the analysis.","marker":"Zhang et al. (2023)"},{"why":"Provides the ten-region 2PACF methodology and the reference values for $\\theta_0$ and $\\beta$ under the isotropy hypothesis.","marker":"Franco et al. (2023)"},{"why":"Establishes the expected HI bias near unity that motivates the choice of the sensitivity window.","marker":"Martin et al. (2012)"},{"why":"Supplies the Planck 2018 cosmological parameters used for theoretical spectra and mock catalog generation.","marker":"Aghanim et al. (2020)"},{"why":"Provides the FLASK code used to generate the 100 lognormal isotropic mock catalogs.","marker":"Xavier et al. (2016)"},{"why":"Provides PolSpice, the software used to estimate the edge- and noise-corrected 2PACF.","marker":"Chon et al. (2004)"},{"why":"Gives the bias-plus-shot-noise model used to fit the angular power spectra and derive $b_0$ and $C_{\\rm SN}$.","marker":"Xia et al. (2009)"},{"why":"Provides the emcee MCMC sampler used for all parameter fitting.","marker":"Foreman-Mackey et al. (2013)"}],"fun_headline_variants":["FAST HI survey finds no deviation from cosmic isotropy","Local Universe stays isotropic in FAST neutral hydrogen data","Ten sky regions in FASHI show isotropic clustering","Cosmic isotropy holds within 2-sigma in FAST HI survey","No anisotropy detected in FAST HI survey's local Universe"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that cutting the catalog to detection sensitivities in [0.65, 1.0] mJy removes survey-induced clustering without removing real anisotropy, and that this window can be chosen after inspecting the measured bias.","fun_headline_variants_meta":{"raw":{"variants":["FAST HI survey finds no deviation from cosmic isotropy","Local Universe stays isotropic in FAST neutral hydrogen data","Ten sky regions in FASHI show isotropic clustering","Cosmic isotropy holds within 2-sigma in FAST HI survey","No anisotropy detected in FAST HI survey's local Universe"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000772,"raw_usage":{"total_tokens":3438,"prompt_tokens":983,"completion_tokens":2455,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":599,"completion_tokens_details":{"reasoning_tokens":2375}},"tokens_in":599,"tokens_out":2455,"duration_ms":16789,"temperature":1.0,"reasoning_tokens":2375,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T05:21:25.831966+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the angular power spectrum dipole and quadrupole of the filtered 19,040-source FASHI sample: a statistically significant nonzero dipole would contradict the isotropy claim. Alternatively, re-run the ten-region 2PACF analysis with the sensitivity window fixed in advance and check whether the scatter in $\\beta$ across regions exceeds the mock scatter at more than $2\\sigma$.","supporting_citations":[{"cited_title":"2009, JCAP, 09, 003, 10.1088/1475-7516/2009/09/003","cited_arxiv_id":null,"evidence_quote":"Gives the bias-plus-shot-noise model used to fit the angular power spectra and derive $b_0$ and $C_{\\rm SN}$."}],"review_version":1}