{"id":"5a2a3bd7-44ca-4473-bc17-8fbe3570e9f3","arxiv_id":"2511.18430","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Semi-Gaussian redshift binning recovers the transverse BAO scale more accurately than top-hat or Gaussian bins in SKA-like forecasts, but the advantage depends on survey and bin width.","lead":"This paper uses forecasts to test how the shape of redshift bins — Gaussian, top-hat, or a hybrid 'semi-Gaussian' — affects measurements of the transverse BAO scale for the SKA and DESI surveys. It argues that semi-Gaussian bins recover the true BAO position better than top-hat bins at low redshift, despite containing fewer galaxies.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Semi-Gaussian advantage may be an artifact of hand-picked A=0.2 and an ambiguous width-matching rule; no sensitivity scan is provided.","rationale":"The strongest_claim is a comparative ranking among bin shapes. For that ranking to hold, the comparison must be fair (equal effective width) and the winner must not be a finely tuned point in shape-parameter space. The paper specifies the width-matching rule only as 'Gaussian σ_z is half σ̃_z' (§3), which does not equalize any standard measure such as FWHM, and the manuscript contains a numerical inconsistency between the stated σ̃_z=0.01 for the semi-Gaussian and the figure captions indicating 2σ_z=0.033. The semi-Gaussian uses A=0.2,B=0.8 with no exploration of A; since the semi-Gaussian's α lies between the top-hat's overestimate and the Gaussian's underestimate, A=0.2 may simply be the value where the biases cancel for the chosen z_c and σ_z. The FoM comparison is further compromised because §6 and Appendix B attribute the σ_z≈0.04 peak to bin centers moving into higher-N(z) regions, so the parameter-constraint part of the claim is not purely shape-driven. A sensitivity study over A and σ̃_z, plus an alternative width-matching choice, would settle whether the semi-Gaussian advantage is generic. The proposed correction in Eq. 5.1 is also asserted without derivation and may double-count the half-bin shifts, but this is secondary to the central ranking claim. This matches the reader's weakest_assumption, so the CONDITIONAL verdict is appropriate; no change is needed.","tokens_in":17518,"tokens_out":11325,"duration_ms":102690,"concrete_test":"Fix a representative SKA z_c=0.11 and σ_z=0.02 (or 0.0165) and recompute α (the peak ratio θ_fit/θ_BAO) for a grid of A ∈ [0,1] with B=1−A, and for σ̃_z ∈ [0.5σ_z, 2σ_z], using the same peak-finding procedure as in §5. If the semi-Gaussian's α stays closer to 1 than both the pure Gaussian and top-hat over the whole grid, the claim is robust; if α crosses 1 or the ordering changes with A, the central claim is an artifact of parameter choice. Additionally, re-run with an alternative width-matching rule (e.g., equal FWHM of the Gaussian component) to check whether the semi-Gaussian still wins.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the semi-Gaussian bin (Eq. 3.10 with A=0.2, B=0.8, σ̃_z=0.01) is the most accurate for transverse BAO rests on an unmatched comparison. Section 3 states 'We tried to keep the full width the same for all configurations, so the Gaussian σ_z is half σ̃_z,' but this rule does not equalize FWHM, effective area, or shot noise, and the text is internally inconsistent: §3 specifies σ̃_z=0.01 for the semi-Gaussian while the captions of Figures 2/3 show 2σ_z=0.033 for semi-Gaussian/top-hat and σ_z=0.0165 for Gaussian. More critically, A=0.2/B=0.8 is a single hand-picked hybrid. The α curves in Figures 10/11 show the semi-Gaussian lying between the overestimating top-hat and the underestimating Gaussian; choosing A=0.2 can simply cancel the two biases at a particular z_c and σ_z. No sensitivity scan over A, B, or σ̃_z is presented, so the claimed robustness of the 'semi-Gaussian' shape for σ_z<0.04 is not established. The FoM peaks at σ_z≈0.04 (Figures 14/15) are also acknowledged in §6 and Appendix B to arise from bin centers shifting into higher N(z) regions, not from shape alone, further confounding the parameter-constraint part of the claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies how the choice of redshift-bin window shape (Gaussian, top-hat, and a hybrid 'semi-Gaussian') affects the measured transverse BAO angular scale and Fisher-forecast cosmological parameter constraints for SKA HI and DESI LRG surveys. The authors compute angular correlation functions and power spectra for each bin shape over a range of widths, extract the BAO peak position, compare it to the fiducial θ_BAO, and propose a correction based on fiducial shifts at the two half-bin edges. They find that the semi-Gaussian bin yields the most accurate BAO position for σ_z<0.04 and, in most cases, competitive or better parameter constraints than Gaussian or top-hat bins, despite higher shot noise. The paper concludes that semi-Gaussian binning is the most suitable choice for transverse BAO analyses, particularly for low-redshift surveys.","tokens_in":17850,"tokens_out":7102,"duration_ms":70805,"significance":"If robust, the paper identifies a practical and often ignored issue—bin-shape-dependent projection bias in transverse BAO measurements—and proposes an easily implementable correction. The qualitative asymmetry (Gaussian underestimates, top-hat overestimates) is physically plausible, and the internal comparison across two surveys and many widths is a useful survey-design exercise. The paper does not provide code, machine-checked proofs, or a derived formula for its correction, so its contribution is primarily a forecast/caveat. However, the central claims currently rest on an inconsistent width definition, a single hand-picked semi-Gaussian shape, and a confounded FoM comparison; with these issues fixed, the paper would be a useful contribution to the BAO forecasting literature.","major_comments":[{"comment":"The definition of the semi-Gaussian bin is internally inconsistent. The text states σ̃_z = 0.01 for the semi-Gaussian (and σ̃_z = 1 for the top-hat), but the captions of Figures 2 and 3 specify '2σ_z = 0.033' for both the semi-Gaussian and top-hat, with the Gaussian shown as σ_z = 0.0165. The Fisher forecast in §7 later refers to σ_z = 0.016. The 'full width same' rule in §3 is not satisfied by these numbers if σ̃_z ≠ 2σ_z for the semi-Gaussian. The authors must specify the exact widths used in every figure and calculation and demonstrate that the three shapes have equal effective support, or explain why the comparison is still fair. Without this, the ranking of bin shapes in Sections 5 and 6 may be an artifact of width mismatch rather than shape.","section":"§3, Eq. (3.10), Figs. 2/3"},{"comment":"The proposed correction θ_BAO = [1 − α_− − α_+] θ_fit is introduced without derivation. The α_± are defined as fractional shifts of the fiducial θ_BAO evaluated at the half-bin edges, but no argument is given for why the total bias is the sum of these two terms with equal weights, nor why a linear subtraction of this form corrects the peak location. Figures 12 and 13 show that α̃ − α is not zero and varies substantially with z_c and σ_z, yet no quantitative metric (e.g., residual RMS) or comparison to an independent estimator is provided. The correction should be derived from a model of the peak shift (e.g., a weighted average of the projection kernel) or validated on mock realizations; as it stands, Eq. (5.1) is an untested ad hoc formula.","section":"§5, Eqs. (5.1)–(5.3)"},{"comment":"The semi-Gaussian shape is specified by A=0.2, B=0.8 and a single σ̃_z, but no sensitivity scan over these parameters is presented. The central conclusion that the semi-Gaussian is the most accurate bin shape rests on this hand-picked hybrid. Figures 10 and 11 show that the semi-Gaussian α curve lies between the overestimating top-hat and the underestimating Gaussian; a different A/B choice would interpolate differently and could change the ranking. The authors should test a few values of A (e.g., 0.1, 0.3, 0.5) with B=1−A and σ̃_z fixed, and show whether the superiority for σ_z<0.04 persists. Without this, the claim is not robust.","section":"§3, Eq. (3.10); §5, Figs. 10/11"},{"comment":"The FoM comparison as a function of σ_z is confounded by the fact that the bin centers shift with σ_z, moving into regions of higher N(z) and reducing shot noise, as the paper itself acknowledges in §6 and Appendix B. Thus the FoM peaks at σ_z≈0.03–0.04 are driven by the survey selection function and bin placement, not by the bin shape. To claim that the semi-Gaussian improves parameter constraints, the authors should control for this: fix the bin-center positions while varying σ_z, hold the number of bins and effective volume constant, or compare at fixed shot-noise level. The '3 bins' analysis in Figures 8/9 is a step toward this, but the main σ_z-dependence plots in Figures 14/15 do not support a shape-only conclusion.","section":"§6, Figs. 14/15, Appendix B"}],"minor_comments":[{"comment":"The statement 'σ̃_z = 1 for the top-hat bins' is confusing; presumably this makes the exponential term negligible, giving an approximately constant window. Please clarify the top-hat limit explicitly.","section":"§3, Eq. (3.10)"},{"comment":"The footnote '130 bin schemes are shown in figure 17' appears as a stray, incomplete sentence; also, Figure 17 is in the appendix but the reference is unclear.","section":"§5, footnote"},{"comment":"The captions read 'Gaussian σ_z = 0.0165, σ_z = 0.0165' and 'Semi-gaussian 2σ_z = 0.033, 2σ_z = 0.033'; this duplicated notation is confusing and should be reconciled with the σ̃_z notation used in §3.","section":"Figs. 2/3 captions"},{"comment":"The abstract claims the semi-Gaussian 'best recovers the BAO signal for σ_z<0.04', but §7 states that for DESI, semi-Gaussian and Gaussian are comparable at σ_z=0.01. Please qualify the abstract to reflect the survey-dependent nuance.","section":"Abstract and §7"},{"comment":"The Fisher matrix expression appears to have missing summation over the i,j indices in CovM^{-1}; please clarify the indexing or define the covariance of the data vector explicitly.","section":"§4, Eq. (4.1)"}],"recommendation":"major_revision","confidential_remarks":"The paper's core idea is worth pursuing, but the current manuscript has internal inconsistencies in the bin-width definitions and several untested choices (the semi-Gaussian A/B parameters and the ad hoc correction formula) that undermine the robustness of the main claims. The FoM comparison is also confounded by the survey selection function. I would encourage the authors to make the bin-shape definitions and the analysis pipeline publicly available for independent verification. The paper is not ready for acceptance, but it does not warrant rejection if the authors can address these concerns."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper's real contribution is the first systematic comparison of Gaussian, top-hat, and semi-Gaussian redshift windows for the transverse BAO projection effect, across SKA-like and DESI-like Fisher forecasts. That fills a gap: prior work only varied bin width. The qualitative pattern is believable — at low redshift, top-hat bins overestimate theta_BAO, Gaussian bins underestimate, and the semi-Gaussian falls in between. The Fisher setup is standard, the survey-specific treatment is reasonable, and the paper is honest that the semi-Gaussian carries more shot noise.\n\nThe soft spots are substantive. The central claim is not robust as presented. Equation 3.10's A=0.2, B=0.8, sigma_tilde_z=0.01 is hand-picked, and there is no scan over A, B, or sigma_tilde_z. The alpha curves in Figs. 10/11 show the semi-Gaussian roughly intermediate between the other two; a different mix could move it to either side. The width-matching rule is also ambiguous: the text says the Gaussian sigma_z is half sigma_tilde_z, and the figure captions show 2 sigma_tilde_z = 0.033 and sigma_z = 0.0165, which is numerically consistent, but the effective shape widths are not really equalized and the relation is under-explained. The FoM peaks at sigma_z ~ 0.03–0.04 are explicitly attributed in Sec. 6 and App. B to bin centers shifting into higher-N(z) regions, so they are not evidence for the intrinsic advantage of the shape. Equation 5.1 is asserted without derivation and appears to combine the two half-bin fractional shifts additively with no weighting by bin occupation; as written it can cancel biases arithmetically. The validation is internal: the 'correct' BAO position is the fiducial theta_BAO, and the proposed correction uses that same fiducial cosmology. No code or data is provided, so the rankings are currently unreproducible.\n\nThese are fixable rather than fatal. The broad takeaway — smooth-edged bins are safer than top-hats for low-redshift angular BAO, and thin bins are not automatically better — is worth having and is supported by the qualitative behavior. The quantitative claims about semi-Gaussian optimality and the correction formula need more work.\n\nThis paper is for people designing redshift binning for angular BAO or tomographic analyses with SKA/DESI-like surveys. A serious referee should engage with it, mainly to require a sensitivity scan over the shape parameters, a derivation or numerical demonstration of Eq. 5.1, and code or data release. I would not desk-reject it.","headline":"A useful first systematic comparison of bin shapes for transverse BAO forecasts, but the headline semi-Gaussian recommendation rests on a hand-picked shape and no sensitivity scan, so it is a plausible guideline, not an established result.","tokens_in":18343,"tokens_out":1782,"would_cite":true,"duration_ms":20127,"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 paper shows that the shape of the redshift bin—not just its width—controls how faithfully the transverse BAO scale is recovered, and that a hybrid semi-Gaussian bin outperforms Gaussian and top-hat bins for low-redshift surveys.","keywords":["baryon acoustic oscillations","transverse BAO","angular power spectrum","redshift binning","projection effect","Fisher forecast","SKA","DESI"],"falsifier":"A focused calculation varying the semi-Gaussian mixing fraction A from 0 to 1 (and the width σ̃_z) while keeping the total bin width fixed: if the recovered θ_BAO deviation α stops being closest to 1 for some intermediate A (or if top-hat overtakes it), the paper's central conclusion loses support. Equivalently, applying the three bin schemes to a realistic mock catalog with known cosmology and checking which returns the input θ_BAO would settle it.","tokens_in":17337,"feed_emoji":"📏","tokens_out":5055,"duration_ms":43901,"temperature":0.7,"pith_summary":"This paper asks whether the way we slice a galaxy survey into redshift bins changes the measured angular scale of baryon acoustic oscillations and the cosmological parameters derived from it. Using Fisher forecasts for the SKA HI and DESI LRG surveys, it compares Gaussian, top-hat, and a hybrid 'semi-Gaussian' bin shape. The authors find that for σ_z < 0.04 the semi-Gaussian bin recovers the fiducial BAO scale with the smallest bias and yields competitive parameter constraints even though it carries more shot noise than a top-hat. They also propose a semi-statistical correction that subtracts the BAO contribution from the lower- and higher-redshift halves of a bin. The takeaway is that bin shape choice, previously ignored, is a first-order systematic in transverse BAO analyses.","feed_headline":"Semi-Gaussian bins recover the BAO scale best","feed_subtitle":"A hybrid bin shape beats Gaussian and top-hat slicing for low-z surveys, despite higher shot noise.","key_machinery":"The comparison hinges on three bin window functions: the Gaussian bin n_i(z) of Eq. 3.9 with width σ_z; the top-hat and semi-Gaussian bins ñ_i(z) of Eq. 3.10, which are flat at height A with a Gaussian bump of width σ̃_z on top; the semi-Gaussian case uses A=0.2, B=0.8, σ̃_z=0.01, and the widths are matched so the Gaussian σ_z is half σ̃_z. The correction mechanism is the adjacent-redshift formula in Eq. 5.1: θ_BAO = (1−α⁻−α⁺) θ_fit, where α⁻ and α⁺ are fractional deviations of the BAO scale at the lower and upper half-bin redshifts from the central value.","core_discovery":"The central claim is that a redshift bin whose selection function is flat over 80% of the distribution and Gaussian-peaked over the remaining 20%—the 'semi-Gaussian' scheme with A=0.2, B=0.8, σ̃_z=0.01—recovers the transverse BAO peak closest to the fiducial value for SKA-like low-redshift surveys, and gives tighter or comparable parameter constraints than pure Gaussian or top-hat bins, despite higher shot noise. For DESI, semi-Gaussian is tightest below σ_z=0.04 and marginal otherwise. The proposed correction, θ_BAO = (1−α⁻−α⁺)θ_fit, uses the fiducial BAO angular scale of the two adjacent half-bin redshifts to remove the projection effect.","pith_inferences":["Editorial inference: The A=0.2/B=0.8 split is not derived from first principles; a systematic scan over A and B would reveal whether the semi-Gaussian advantage is a genuine feature of the shape or an artifact of this one parameter choice.","Editorial inference: The adjacent-redshift correction (Eq. 5.1) assumes the projection bias can be captured by just two half-bin redshifts; for very wide bins or steep N(z), a full integral over the bin may be needed, and the correction should be tested on mocks before use.","Editorial inference: If confirmed, the result suggests that photometric surveys with large redshift uncertainties could improve BAO constraints by engineering selection functions closer to a semi-Gaussian, rather than relying on top-hat photo-z shells.","Editorial inference: The Fisher framework assumes Gaussian likelihood and ignores non-Gaussian covariance; a full simulation-based covariance could alter the FoM ranking."],"forward_implications":["For low-redshift surveys like SKA, adopting semi-Gaussian bins should recover the transverse BAO scale with less bias than top-hat or Gaussian bins, at σ_z < 0.04.","The statistical correction based on adjacent redshift halves can be applied to existing and future angular BAO measurements to remove part of the projection bias.","Gaussian binning, though common, gives the least precise parameter constraints among the three schemes for SKA, and is weak for DESI at small σ_z.","The optimal bin width for BAO detection depends on z_c, so fixed thin bins are not automatically optimal; this supports the recommendation to choose bin width based on the survey's redshift range."],"fun_headline_variants":["Hybrid binning wins for BAO recovery","Semi-Gaussian bins beat pure shapes for BAO","Optimal bin shape for transverse BAO found","Semi-Gaussian slicing recovers BAO best","Bin shape matters: semi-Gaussian top for BAO"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire ranking assumes that the three bin shapes have the same effective full width (Gaussian σ_z set to half σ̃_z) and that the hand-picked semi-Gaussian parameters A=0.2, B=0.8, σ̃_z=0.01 are the right hybrid; if that width-matching rule or parameter choice is changed, the ordering of bin shapes could change.","fun_headline_variants_meta":{"raw":{"variants":["Hybrid binning wins for BAO recovery","Semi-Gaussian bins beat pure shapes for BAO","Optimal bin shape for transverse BAO found","Semi-Gaussian slicing recovers BAO best","Bin shape matters: semi-Gaussian top for BAO"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000203,"raw_usage":{"total_tokens":1187,"prompt_tokens":674,"completion_tokens":513,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":418,"completion_tokens_details":{"reasoning_tokens":437}},"tokens_in":418,"tokens_out":513,"duration_ms":4616,"temperature":1.0,"reasoning_tokens":437,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T20:44:44.738807+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A focused calculation varying the semi-Gaussian mixing fraction A from 0 to 1 (and the width σ̃_z) while keeping the total bin width fixed: if the recovered θ_BAO deviation α stops being closest to 1 for some intermediate A (or if top-hat overtakes it), the paper's central conclusion loses support. Equivalently, applying the three bin schemes to a realistic mock catalog with known cosmology and checking which returns the input θ_BAO would settle it.","supporting_citations":[],"review_version":1}