REVIEW 3 major objections 4 minor
A joint analysis of galaxy clustering and 21-cm intensity mapping can pin the neutral hydrogen fraction to sub-percent accuracy, breaking the long-standing degeneracy with hydrogen bias.
Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →
A Fisher-matrix forecast says galaxy-21cm cross-correlation can measure Ω_HI at z~1 to sub-percent and improve growth constraints by ~2x, but the forecast's precision appears overstated.
T0 review reviewed 2026-08-03 challenge →
load-bearing objection The double-tracer Fisher formalism is a real step forward, but the sub-percent ΩHI claim is built on using the full DESI volume for a 1.4% sky 21-cm patch; with the true overlap volume the headline error quadruples and the paper needs revision. the 3 major comments →
Cosmological Implication of Cross-correlation between Galaxy Clustering and 21-cm Line Intensity Mapping
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
Core claim
The central claim is that the three power spectra measured jointly—galaxy auto-correlation, 21-cm auto-correlation, and the galaxy–21-cm cross-correlation—contain enough anisotropic information in the quasi-linear regime to determine xHI to 0.32% in the no-foreground case and to 2.6% even with strong foreground contamination. Because xHI enters the spectra as a pure overall multiplier while bHI modulates the angular dependence through the line-of-sight anisotropy of redshift-space distortions, the joint fit separates the two and breaks the ΩHI–bHI degeneracy that limits linear-regime analyses. The same joint fit improves the coherent-velocity constraint fσ8 by a factor of about two relative
What carries the argument
The machinery is a two-tracer extension of a nonlinear redshift-space distortion model. The anisotropic power spectra Pab(k, μ) are written as the sum of a perturbative part—including density–density, density–velocity-divergence, and velocity-divergence auto-spectra—plus a non-perturbative Finger-of-God damping factor, with higher-order correction terms calibrated in a fiducial cosmology and rescaled by growth functions. An information matrix then combines the three spectra with their full joint covariance, including galaxy shot noise, 21-cm shot noise, instrument noise, and foreground noise. This structure allows xHI, a multiplicative amplitude, to be separated from bHI, which changes the μ
Load-bearing premise
The load-bearing assumption is that the full galaxy-survey volume is available to the combined three-spectrum measurement, even though the 21-cm survey covers only a small sky patch; if the actual overlapping volume is used, the quoted sub-percent xHI precision weakens considerably.
What would settle it
Recompute the information-matrix forecast using the true overlap volume of the 21-cm survey (fsky≈0.014 at z≈1) instead of the full galaxy-survey volume; if σ(xHI)/xHI exceeds 1% in the no-foreground case, the headline sub-percent claim fails. A second check: run the identical forecast on mock catalogs with a known input xHI and see whether the 1σ scatter matches the predicted 0.32% error.
If this is right
- If the forecast is correct, the neutral hydrogen fraction xHI at z≈1 can be measured to roughly 0.3% with clean foreground subtraction, and to a few percent even with residual foregrounds—far beyond current stacked-emission constraints.
- The long-standing degeneracy between ΩHI and the hydrogen bias bHI, which has limited clustering-based estimates, would be broken by the anisotropic quasi-linear spectra.
- The growth-rate combination fσ8 would be constrained to about 1% and σ8 to about 5% from a single galaxy-plus-21-cm program, roughly doubling the fσ8 precision available from galaxy clustering alone.
- Distance constraints tied to geometric distortion would see little improvement from adding the 21-cm data, so this method complements rather than replaces galaxy BAO analyses.
- With finer redshift bins, the same analysis could map xHI(z) across z≈0–2 at roughly percent-level accuracy per bin.
Where Pith is reading between the lines
- The quoted sub-percent xHI error appears to use the full galaxy-survey volume for all three spectra; if only the overlapping sky area of the 21-cm survey (about 1.4% of the sky at z≈1) enters the calculation, the errors grow by roughly a factor of four, and the no-foreground constraint may rise above 1%. Verification of the overlap volume is the first thing to check.
- Because xHI is a pure amplitude, the same cross-correlation technique should extend to other line-intensity tracers at higher redshift, such as CO or [CII], to break analogous amplitude–bias degeneracies wherever a galaxy sample provides the cross-correlation.
- The forecast's realism hinges on how many large-scale line-of-sight modes survive foreground removal; the paper parametrizes this with a foreground noise term, but real spectrally smooth foregrounds may remove a broader set of modes than a simple noise term captures.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This paper develops a Fisher-matrix forecast for combining DESI ELG galaxy clustering auto-power, 21-cm line-intensity mapping auto-power, and their cross-power at z = 1.0–1.2. The model includes a nonlinear RSD treatment with Finger-of-God damping, higher-order bias terms, and the Alcock–Paczynski effect. The central claims are that the joint analysis constrains the neutral hydrogen fraction xHI (and hence ΩHI) to sub-percent accuracy—σ(xHI)/xHI = 0.32% in the no-foreground case—while improving growth constraints by roughly a factor of two over galaxy clustering alone. Appendix C presents an analytic reduction of the xHI Fisher element as a sanity check.
Significance. If the forecast were correct as stated, the claimed sub-percent ΩHI measurement at z ≈ 1 would be a substantial advance over current stacked and clustering constraints, which have order-unity uncertainties, and would demonstrate a practical route to breaking the ΩHI–bHI degeneracy. The two-tracer extension of the nonlinear RSD formalism is nontrivial, and the analytic Fisher reduction in Appendix C is a useful internal consistency check. However, the headline quantitative claim depends on a survey-volume assignment that appears internally inconsistent with the stated 21-cm survey geometry. The proposed methodology is plausible, but the numerical results presented in Table 2 and the abstract do not follow from the experimental setup described in §2.3.
major comments (3)
- [§2.3, Eq. (34), Table 1, Table 2, §3.3] The central forecast uses inconsistent survey volumes. Section 2.3 specifies the 21-cm survey as a roughly 24°×24° patch with fsky≈0.014, corresponding to a comoving volume of about 0.3–0.5 h^-3 Gpc^3 at z = 1.0–1.2. The Fisher mode count in Eq. (34) is instead evaluated with V_survey = 8.4 h^-3 Gpc^3, the full DESI ELG volume quoted in Table 1. Since Eq. (31) is the covariance of Pgg, PgH, and PHH for the same Fourier modes, all three spectra should use the overlap volume, or the non-overlapping DESI-only modes must be separated block-diagonally. The present implementation overcounts modes by a factor of roughly 15–25, so all Fisher errors are underestimated by a factor of about 4–5. In particular, the headline σ(xHI)/xHI = 0.32% in Table 2 becomes approximately 1.3–1.6% in the no-foreground case, and the Nfore = 10^4 entry becomes approximately 10–13%, not 2.6%. The abstract’s sub-perc
- [§2.5, Eq. (31)] If the 21-cm patch is embedded inside the DESI footprint, the galaxy auto-spectrum can in principle use the full DESI volume, while the gH and HH spectra are limited to the overlap volume and the gg auto-spectrum in the overlap region has a different effective mode count. The current single-Np implementation does not specify which of these cases is intended and therefore cannot be interpreted as either an optimistic or conservative forecast. A corrected Fisher matrix must either set V_survey to the overlap volume for the full 3×3 block or construct a block-diagonal covariance that treats overlap and non-overlap DESI modes separately. This is a necessary correction regardless of the final numerical values.
- [Appendix B, §4.2] The claimed breaking of the bHI–xHI degeneracy relies on the nonlinear RSD terms A, B, T, and F, which are calibrated from the authors’ previous N-body work (Zheng & Song 2016; Zheng et al. 2019). The paper does not provide an independent validation of these templates for a HI tracer at z ≈ 1, nor does it propagate their calibration uncertainty into the Fisher forecast. Because the factor-of-several improvement over the HI auto-spectrum comes precisely from these template terms, the robustness of Table 2 should be tested, e.g. by varying the template amplitudes by their reported calibration uncertainty or by comparing with an independent HI simulation. This is a concrete correctness-risk concern and should be addressed even after the volume issue is corrected.
minor comments (4)
- [Table 2, §3.3] The column label “galaxy-21cm” is ambiguous. It should state explicitly whether the column is the full joint analysis (gg + gH + HH) or only the cross-spectrum contribution; the text sometimes refers to the joint analysis as “cross+auto.”
- [§4.2] Typographical issue: “xHI parameter is most degenerate with galaxy bias bHI” should read “HI bias bHI,” not galaxy bias.
- [Eq. (36)] The integral upper limit is written as MHI, but the intended upper limit is presumably the maximum hydrogen clump mass (or infinity). Please clarify the notation.
- [Abstract, §3.2] The abstract states that the cosmic-expansion constraint is “slightly improved,” while §3.2 concludes that DA is “not much improved” and H^-1 is actually more difficult to constrain. The wording should be harmonized to avoid overstating the AP benefit.
Circularity Check
No significant circularity: the Fisher forecast is self-contained; xHI/Omega_HI is a free parameter whose error is projected from the model, not fitted from the claimed result.
full rationale
The paper's central forecast is a Fisher-matrix projection: model power spectra (Eqs. 6-30) are differentiated with respect to parameters and combined with the covariance (Eqs. 31-35). The target xHI/Omega_HI is a free parameter in the signal temperature (Eq. 3) with a fiducial input value; the quoted sigma(xHI) is a Cramer-Rao error, not a value fitted to data. The nonlinear RSD kernels cited from Zheng & Song (2016) and Song et al. (2018) are N-body-calibrated templates; this is external evidence (stated in App. B: 'measured from N-body simulations according to Zheng & Song (2016)'), not an invocation of the paper's own target result, so it does not constitute circularity under the stated rules. Appendix C algebraically checks the xHI Fisher element but inherits the same mode-count normalization, so it is a consistency check rather than an independent validation. The main substantive concern is internal consistency of the survey volume: Eq. (34) uses V_survey from Table 1 (8.4 h^-3 Gpc^3 for the ELG bin) while Sec. 2.3 specifies a 24x24 deg^2 21-cm patch with fsky~0.014 (~0.5 h^-3 Gpc^3 at z~1); if the true overlap volume is smaller, quoted errors inflate by roughly a factor of four. This is a correctness/robustness issue, not a circularity.
Axiom & Free-Parameter Ledger
free parameters (4)
- σp (FoG velocity dispersion)
- xHI (neutral hydrogen fraction) =
0.027
- bH1 (HI linear bias) =
1.1
- Nfore (foreground residual amplitude) =
10^2-10^4 h^-3 Mpc^3
axioms (4)
- domain assumption RegPT two-loop perturbation theory plus N-body-calibrated A/B/T/F correction terms accurately describe the nonlinear RSD power spectrum.
- standard math The Fisher matrix formalism gives unbiased forecasts with Gaussian covariance and no systematic errors.
- domain assumption No velocity bias; all tracers share the same velocity field.
- ad hoc to paper Foreground residuals behave as an additive, noise-like term Nfore uncorrelated with galaxy noise.
Cite this review
Pith. "Pith review of Cosmological Implication of Cross-correlation between Galaxy Clustering and 21-cm Line Intensity Mapping." pith.science (2026). https://pith.science/paper/4HKJRY3L
@misc{pith2026251211353,
author = {Pith},
title = {Pith review of: Cosmological Implication of Cross-correlation between Galaxy Clustering and 21-cm Line Intensity Mapping},
year = {2026},
howpublished = {\url{https://pith.science/paper/4HKJRY3L}},
note = {Machine review of arXiv:2512.11353}
}
read the original abstract
The apparent anisotropies of galaxy clustering and 21-cm mapping in redshift space offer a unique opportunity to simultaneously probe cosmic expansion and gravity on cosmological scales through the Alcock-Paczynski (AP) effect and redshift-space distortions (RSD). Although improved theoretical models exist for anisotropic clustering, their applicability is limited by the non-perturbative smearing effect caused by the randomness of relative velocities. Here, we consider an alternative approach using the statistical power of cross-correlation between galaxy clustering and 21-cm line intensity mapping. Based on Fisher matrix analysis, fully incorporating nonlinear RSD, we estimate the benefit of combining both observables. We find that, for spectroscopy surveys like DESI combined with 21-cm line-intensity mapping surveys, constraints on the growth of structure and the cosmic expansion rate are improved by a factor of two relative to the galaxy auto-correlation. Crucially, such an observation can strongly constrain the neutral hydrogen (HI) content Omega_HI to a sub-percent level. This level of precision unlocks the potential of this method to probe post-reionization astrophysics with enhanced precision. It would far surpass existing constraints from stacked 21-cm emission and break the degeneracy between Omega_HI and the HI bias b_HI inherent in the linear-regime power-spectrum analysis. This cross-correlation approach effectively compensates for the loss of constraining power when using galaxy clustering alone.
Figures
This paper was first reviewed by deepseek-v4-flash on August 3, 2026.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.