Pith. sign in

REVIEW 2 major objections 4 minor 22 references

A Gaussian Covariance Matrix for Joint Pre- and Post-Reconstruction Full-Shape Power Spectrum Analysis

T0 review · 2 major / 4 minor · reviewed 2026-08-08 · deepseek-v4-flash

Pith's one-line read A semi-analytical Gaussian covariance matrix for the joint analysis of pre-reconstruction, post-reconstruction, and cross galaxy power spectra matches mock-based cosmological constraints over conservative fitting ranges.

desk verdict A solid methods paper: the new cross shot-noise model and estimator are useful, but the provenance of the bias input b=2.1 needs to be nailed down before the validation is fully convincing. read the letter →

arxiv 2608.05504 v1 pith:Z7FFPTDZ submitted 2026-08-06 astro-ph.CO

classification astro-ph.CO
keywords galaxypowerspectrumcovariancematrixBAOreconstructioncrossshotnoisedisplacementfieldfull-shapeanalysisredshift-spacedistortions
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper develops a semi-analytical covariance matrix for the joint full-shape analysis of the pre-reconstruction, post-reconstruction, and cross galaxy power spectra. It claims that a Gaussian covariance, built from measured power-spectrum multipoles plus a modeled, reconstruction-induced cross shot noise, captures the dominant covariance structure of the full data vector, so that cosmological constraints match those from mock-based numerical covariance on conservative scales. The paper also introduces a new estimator that measures the cross shot noise directly without splitting the galaxy catalogue. If correct, this gives a fast alternative to large mock suites for joint BAO and RSD analyses.

What carries the argument

The central object is the Gaussian covariance matrix for the nine-block data vector, whose entries are $C_{AB}(k,\mu) = (2/N_k)\tilde P_A \tilde P_B$ with $\tilde P$ including Legendre multipoles and shot noise, followed by Legendre projection to multipole space. The new ingredient is the cross shot-noise model: treating the reconstruction displacement field as Gaussian makes the exponential cumulant expansion truncate at second order, so the cross shot noise is $(1/\bar n)$ times the exponential of the displacement covariance tensor $\kappa_{ij}$, with $\kappa_{ij}$ computed from the Kaiser-model $P_{gg}$. A new estimator averages a shifted noise catalog to measure this term directly, avoiding the variance cost of splitting the galaxy catalogue.

What would settle it

Measure the actual displacement field's higher-order connected moments in the GLAM mocks; if the third or fourth cumulants of $k\cdot s$ contribute comparably to the second at $k \sim 0.12\,h\,{\rm Mpc}^{-1}$, the Gaussian truncation in Eq. (3.7) is violated and the modeled $P_{\rm cross}$ shot noise is biased.

Watch

Extended reading notes

Core claim

Using Wick's theorem, the covariance of the joint data vector $(P^{\rm pre}_\ell, P^{\rm post}_\ell, P^{\rm cross}_\ell)$ is expressed through products of 3D power spectra with shot-noise multipoles. The reconstruction-induced cross shot noise is modeled as $N_{\rm cross}(k,\mu) = (1/\bar n)\exp[-\tfrac{1}{2} k^2(\sigma_A^2(\mu) + (2f_{\rm fid}+f_{\rm fid}^2)\mu^2\sigma_B^2)]$, with displacement covariance tensors computed from a Kaiser-model galaxy power spectrum. A new estimator, $x_{\rm SN}[\delta(-k)R(k)] = \sum_i e^{-ik\cdot(x_{{\rm rec},i}-x_i)}$, measures this shot noise directly without splitting the catalogue. The paper validates the semi-analytical Gaussian covariance against GLAM mock catalogues: diagonal covariance elements are reproduced up to $k=0.25\,h\,{\rm Mpc}^{-1}$ for pre- and post-reconstruction spectra and up to $k=0.12\,h\,{\rm Mpc}^{-1}$ for the cross spectrum, and emulator-based joint fits with the Gaussian covariance yield cosmological constraints consistent with the mock-based covariance for $k_{\rm max}=0.18\,h\,{\rm Mpc}^{-1}$ (pre/post) and $k_{\rm max}=0.12\,h\,{\rm Mpc}^{-1}$ (cross).

Load-bearing premise

The load-bearing premise is that the reconstruction displacement field is Gaussian, so that $\langle e^{-ik\cdot s}\rangle$ is fixed by the two-point displacement covariance alone, and that this covariance is computed from a Kaiser-model galaxy power spectrum; if higher-order displacement cumulants or a biased Kaiser model matter, the modeled cross shot noise and the $P_{\rm cross}$ covariance block are wrong.

Editorial extensions

If this is right

  • Joint fits of $P_{\rm pre}+P_{\rm post}+P_{\rm cross}$ with the Gaussian covariance reproduce mock-covariance constraints at $k_{\rm max}=0.18\,h\,{\rm Mpc}^{-1}$ for pre/post and $0.12\,h\,{\rm Mpc}^{-1}$ for cross, so analyses on these scales can replace mock covariance with the analytic version.
  • The new cross shot-noise estimator matches the HS-HD method in the mean but has about 10% lower variance at $k=0.25\,h\,{\rm Mpc}^{-1}$, improving small-scale $P_{\rm cross}$ measurements without splitting catalogs.
  • Beyond these scales, differences between Gaussian and numerical covariances grow, especially for $P_{\rm cross}$, implying non-Gaussian covariance cannot be ignored in aggressive fitting ranges.
  • The framework reduces the computational cost of covariance estimation for high-dimensional joint data vectors, which is the main barrier for full-shape survey analyses.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Extension: The displacement-statistics shot-noise model could be carried over to other post-reconstruction statistics, such as correlation-function multipoles or the bispectrum, where mock covariances are even more expensive.
  • Extension: Because the new estimator does not split the catalogue, it should be straightforward to apply to surveys with complex masks once window-function effects are folded in, reducing small-scale variance compared with split-catalogue estimators.
  • Extension: A direct test of the Gaussian displacement assumption—measuring the third and fourth cumulants of the displacement field in N-body simulations—would indicate how far past $k=0.12\,h\,{\rm Mpc}^{-1}$ the cross-spectrum covariance can be trusted.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The paper constructs a semi-analytical Gaussian covariance matrix for the joint data vector of pre-reconstruction, post-reconstruction, and cross power-spectrum multipoles. The key new ingredient is an analytical model for the reconstruction-induced, scale-dependent cross shot noise derived from displacement-field cumulants (Eqs. 3.6-3.12), together with a new FFT-based estimator for this shot noise (Eqs. 4.4-4.6). The covariance is built from measured power spectra and modeled shot noise, then validated against 986 GLAM mocks, and finally used in emulator-based parameter fits to compare cosmological constraints with those from the numerical mock covariance. The authors report agreement for the diagonal covariance elements up to k=0.25 h/Mpc for the pre/post blocks and up to k=0.12 h/Mpc for the P_cross block, and near-identical joint parameter constraints for k_max=0.18 (pre/post) and 0.12 (cross) h/Mpc.

Significance. If accepted, the framework provides a fast, semi-analytical alternative to mock-based covariance estimation for the joint three-spectrum analysis that is becoming standard in DESI-like surveys. The paper's strengths are the explicit Wick-theorem/Legendre-projection derivation, the displacement-field model for the cross shot noise, the new xSN estimator that avoids catalog splitting, and the extensive validation against 986 independent mocks, including a controlled parameter-inference comparison. The central validation currently rests on model inputs whose provenance is not documented, so the significance is conditional on those inputs being fixed a priori and on a sensitivity analysis.

major comments (2)
  1. [Sec. 5 / Eq. (3.11) / Fig. 2] The value b=2.1 used in the analytical cross shot-noise model appears only in the Fig. 2 caption, and the smoothing scale Sigma_s entering K_s(q) in Eq. (3.3) is not stated anywhere in the validation section. Because sigma_A^2 and sigma_B^2 in Eq. (3.10) are quadratic in b through P_gg, and N_cross in Eq. (3.9) depends exponentially on them, the agreement shown in Fig. 2 is not an independent validation unless b and Sigma_s are fixed a priori by a stated procedure. Please specify how b is measured (e.g., from the large-scale pre-reconstruction monopole of the same mocks, or from the HOD) and report a sensitivity test of the relevant covariance entries to, say, a +/-10% change in b and to the adopted Sigma_s. This is load-bearing because the cross shot noise enters the P_cross covariance and the cross-covariance blocks.
  2. [Sec. 5 / Eqs. (2.5), (4.2), (4.7) / Fig. 6] The Gaussian covariance is constructed with eP_cross = P_cross_sub + N_cross_model, i.e., the total cross power including shot noise. The numerical covariances it is compared against are for shot-noise-subtracted estimators: the HS-HD estimator in Eq. (4.2) and the direct-subtraction estimator in Eq. (4.7). Subtracting a scale-dependent shot-noise estimate is not a constant shift and can change the variance, as the paper itself shows by the ~10% variance difference between the two estimators at k=0.25 h/Mpc. The manuscript should explain why the Gaussian covariance for the total P_cross is applicable to the subtracted data vectors; currently the only justification is the empirical agreement shown in Fig. 6, which is limited to k<=0.12 h/Mpc. A brief derivation or a quantitative statement of when the subtraction effect is negligible would make the validation self-consistent.
minor comments (4)
  1. [Sec. 5] The text says the covariances are rescaled to an effective survey volume 'three times that of the GLAM mock, i.e., 3(h^-1 Mpc)^3'; the unit should presumably be (h^-1 Gpc)^3.
  2. [Sec. 5 / Figs. 7-11] The parameter-inference test uses the emulator both as the mock observation vector and as the theoretical model. This is a valid controlled covariance comparison, but the text should state explicitly that these are noiseless Fisher-type forecasts that isolate the covariance approximation, not actual constraints from the GLAM mocks.
  3. [Fig. 5 caption / Sec. 6] There are small language errors: 'between between' in the Fig. 5 caption, 'For the the joint analysis' in Sec. 6, and 'Both the measurements agree well the analytical model' in the discussion of Fig. 2. These should be corrected.
  4. [Sec. 3 / Eq. (3.10)] The integrals in Eq. (3.10) define sigma_A^2(μ) with a factor 1/2 in front of the (2π)^-2; the corresponding value of sigma_A in the equal-bias limit is stated in Eq. (3.13). It would help the reader if the reduction from Eq. (3.10) to Eq. (3.13) were sketched in one line, since Eq. (3.13) is the expression most easily compared with the literature.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the covariance is a forward model evaluated from measured power spectra and an independent analytic shot-noise model, then validated against mock-based covariance.

full rationale

The paper's central object, the semi-analytical Gaussian covariance, is not derived from the quantity it is claimed to predict. Eq. (2.3)-(2.6) is the standard Wick-theorem Gaussian covariance in terms of the 3D power spectra, and the paper explicitly states that the 'measured power-spectrum multipoles' are used as inputs together with the analytic cross shot-noise model of Sec. 3. The validation in Sec. 5 compares this forward-modelled covariance with the numerical covariance estimated from 986 GLAM mock catalogues. No parameter of the Gaussian covariance is fit to the mock covariance; the only adopted numeric values (b=2.1, b_fid=1.824, f=f_fid=0.778) enter the independent cross shot-noise model, not the covariance calibration. The emulator-based parameter inference uses the same emulator as both the mock observation vector and the theoretical model ('We adopt the fiducial cosmology prediction from the emulator [14] as the mock observation vector. The theoretical model is evaluated using the same emulator'), which is self-referential as a validation design and limits the test to covariance-shape comparisons rather than an end-to-end recovery of independent data, but it does not make the derivation circular because the Gaussian covariance itself is not obtained from those parameter constraints. The provenance of b=2.1 is not stated, which is a reproducibility and robustness concern, but there is no quoted evidence that it was tuned to match the mock N_cross, so it cannot be classified as a fitted input renamed as a prediction under the hard-evidence rule. The central derivation chain is self-contained: the Gaussian covariance formula is independent of the numerical covariance it is tested against, and the agreement is a genuine forward-model check rather than a tautology.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The model rests on standard large-scale structure assumptions (Gaussian fields, Kaiser RSD, plane-parallel). The numerical inputs (b, f, sigma_s, alpha) are adopted from the mock setup and reconstruction convention; none are fitted to the covariance target in this paper.

free parameters (5)
  • b (linear bias in Kaiser model) = 2.1
    Used in Eq. (3.11) to compute P_gg for the displacement covariance. Adopted from the mock galaxy sample, but its origin is not stated in the paper.
  • b_fid (fiducial bias in reconstruction) = 1.824
    Input to the reconstruction displacement field in Eq. (3.4).
  • f = f_fid (linear growth rate) = 0.778
    Used both as the true and fiducial growth rate in the displacement model.
  • sigma_s (reconstruction smoothing scale) = not stated
    Appears in the Gaussian smoothing kernel K_s(q) in Eq. (3.10). The numerical value used in the mocks is not given in the paper, which is a reproducibility gap.
  • alpha (random catalogue ratio) = 0.05
    Used in the Poisson shot noise for P_post, Eq. (3.1).
assumptions (5)
  • domain assumption Gaussian density fluctuations (Wick's theorem applies)
    The covariance in Eq. (2.3) is derived assuming Gaussian fields. This is the foundation of the Gaussian covariance model.
  • domain assumption Gaussian displacement field for reconstruction
    Eq. (3.7) truncates the cumulant expansion of <e^{ik*s}> at second order, assuming all higher cumulants vanish.
  • domain assumption Kaiser model for the redshift-space galaxy power spectrum P_gg
    Used in Eq. (3.11) to compute the displacement covariance tensor; ignores nonlinear corrections beyond Kaiser RSD.
  • domain assumption Plane-parallel approximation for the line of sight
    The displacement field and shot noise are derived assuming a global line-of-sight direction z, valid for the cubic mock geometry.
  • domain assumption Measured mean power spectra from mocks represent the true underlying spectra
    The Gaussian covariance is built using the mean P from 986 mocks; any bias in those means propagates into the covariance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Gaussian Covariance Matrix for Joint Pre- and Post-Reconstruction Full-Shape Power Spectrum Analysis." pith.science (2026). https://pith.science/paper/Z7FFPTDZ

@misc{pith2026260805504,
  author       = {Pith},
  title        = {Pith review of: A Gaussian Covariance Matrix for Joint Pre- and Post-Reconstruction Full-Shape Power Spectrum Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z7FFPTDZ}},
  note         = {Machine review of arXiv:2608.05504}
}
abstract

We apply the Gaussian covariance formalism to develop a semi-analytical covariance model for the joint analysis of pre-reconstruction, post-reconstruction, and cross full-shape galaxy power spectra. We model the reconstruction-reduced, scale-dependent cross shot noise using displacement-field statistics and introduce a new estimator that directly measures this term. Using the measured power spectra and the modeled shot-noise predictions as inputs, we construct the Gaussian covariance while accounting for correlations between the pre- and post-reconstruction density fields. We validate the resulting semi-analytical Gaussian covariance against mock catalogues. Using emulator-based parameter inference, we demonstrate that the semi-analytical Gaussian covariance adequately captures the dominant contribution to the covariance structure of the full data vector ($P_{\ell}^{\rm pre}, P_{\ell}^{\rm post}, P_{\ell}^{\rm cross}$). For the joint fit to these three power spectra, it yields cosmological constraints consistent with those obtained using the mock-based numerical covariance over the adopted fitting ranges: $k_{\rm max}=0.18\,h\,{\rm Mpc}^{-1}$ for $P_{\rm pre}$ and $P_{\rm post}$, and $k_{\rm max}=0.12\,h\,{\rm Mpc}^{-1}$ for $P_{\rm cross}$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 3 canonical work pages

  1. [1]

    Eisenstein, H.-j

    D.J. Eisenstein, H.-j. Seo, E. Sirko and D. Spergel,Improving Cosmological Distance Measurements by Reconstruction of the Baryon Acoustic Peak,Astrophys. J.664(2007) 675 [astro-ph/0604362]

  2. [2]

    Gil-Mar ´ ın,How to optimally combine pre-reconstruction full shape and post-reconstruction BAO signals,JCAP05(2022) 040 [2203.05581]

    H. Gil-Mar ´ ın,How to optimally combine pre-reconstruction full shape and post-reconstruction BAO signals,JCAP05(2022) 040 [2203.05581]

  3. [3]

    M. Maus, A. Baleato Lizancos, M. White, A. de Mattia and S.-F. Chen,An analytic approximation to the covariance between pre- and post-reconstruction galaxy two-point statistics,JCAP05(2026) 078 [2602.12343]

  4. [4]

    Wang et al.,Extracting high-order cosmological information in galaxy surveys with power spectra,Commun

    Y. Wang et al.,Extracting high-order cosmological information in galaxy surveys with power spectra,Commun. Phys.7(2024) 130 [2202.05248]

  5. [5]

    Y. Wang et al.,Nonlinear information from desi luminous red galaxies: An emulator-based analysis of pre- and post-reconstruction power spectra,arXiv e-prints(2026) arXiv:2603.25693 [2603.25693]

  6. [6]

    Covariance of the redshift-space matter power spectrum after reconstruction

    C. Hikage, R. Takahashi and K. Koyama,Covariance of the redshift-space matter power spectrum after reconstruction,Phys. Rev. D102(2020) 083514 [2007.13998]

  7. [7]

    R. Zhao, K. Koyama, Y. Wang and G.-B. Zhao,Modeling the Covariance Matrix for the Power Spectra Before and After the BAO Reconstruction,Res. Astron. Astrophys.24(2024) 125015 [2410.18524]

  8. [8]

    White, Y.-S

    M. White, Y.-S. Song and W.J. Percival,Forecasting Cosmological Constraints from Redshift Surveys,Mon. Not. Roy. Astron. Soc.397(2008) 1348 [0810.1518]

Show all 22 references
  1. [9]

    Blake et al.,Galaxy And Mass Assembly (GAMA): improved cosmic growth measurements using multiple tracers of large-scale structure,Mon

    C. Blake et al.,Galaxy And Mass Assembly (GAMA): improved cosmic growth measurements using multiple tracers of large-scale structure,Mon. Not. Roy. Astron. Soc.436(2013) 3089 [1309.5556]

  2. [10]

    Taruya, T

    A. Taruya, T. Nishimichi and S. Saito,Baryon Acoustic Oscillations in 2D: Modeling Redshift-space Power Spectrum from Perturbation Theory,Phys. Rev. D82(2010) 063522 [1006.0699]

  3. [11]

    Chen et al.,Extensive analysis of reconstruction algorithms for DESI 2024 baryon acoustic oscillations,JCAP05(2026) 001 [2411.19738]

    X. Chen et al.,Extensive analysis of reconstruction algorithms for DESI 2024 baryon acoustic oscillations,JCAP05(2026) 001 [2411.19738]

  4. [12]

    Sugiyama,Developing a theoretical model for the resummation of infrared effects in the postreconstruction power spectrum,Phys

    N. Sugiyama,Developing a theoretical model for the resummation of infrared effects in the postreconstruction power spectrum,Phys. Rev. D110(2024) 063528 [2402.06142]

  5. [13]

    S. Ando, A. Benoit-L´ evy and E. Komatsu,Angular power spectrum of galaxies in the 2MASS Redshift Survey,Mon. Not. Roy. Astron. Soc.473(2018) 4318 [1706.05422]

  6. [14]

    Wang et al.,Emulating Power Spectra for Prereconstructed and Postreconstructed Galaxy Samples,Astrophys

    Y. Wang et al.,Emulating Power Spectra for Prereconstructed and Postreconstructed Galaxy Samples,Astrophys. J.966(2024) 35 [2311.05848]

  7. [15]

    Klypin and F

    A. Klypin and F. Prada,Dark matter statistics for large galaxy catalogues: power spectra and covariance matrices,Mon. Not. Roy. Astron. Soc.478(2018) 4602 [1701.05690]

  8. [16]

    Guo et al.,The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: modelling of the luminosity and colour dependence in the Data Release 10,Mon

    H. Guo et al.,The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: modelling of the luminosity and colour dependence in the Data Release 10,Mon. Not. Roy. Astron. Soc.441(2014) 2398 [1401.3009]

  9. [17]

    Percival et al.,The Clustering of Galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: Including covariance matrix errors,Mon

    W.J. Percival et al.,The Clustering of Galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: Including covariance matrix errors,Mon. Not. Roy. Astron. Soc.439 (2014) 2531 [1312.4841]

  10. [18]

    Scoccimarro, M

    R. Scoccimarro, M. Zaldarriaga and L. Hui,Power spectrum correlations induced by nonlinear clustering,Astrophys. J.527(1999) 1 [astro-ph/9901099]. – 14 –

  11. [19]

    Sugiyama, S

    N.S. Sugiyama, S. Saito, F. Beutler and H.-J. Seo,Perturbation theory approach to predict the covariance matrices of the galaxy power spectrum and bispectrum in redshift space,Mon. Not. Roy. Astron. Soc.497(2020) 1684 [1908.06234]

  12. [20]

    Hu and A.V

    W. Hu and A.V. Kravtsov,Sample variance considerations for cluster surveys,Astrophys. J. 584(2003) 702 [astro-ph/0203169]

  13. [21]

    Takada and W

    M. Takada and W. Hu,Power Spectrum Super-Sample Covariance,Phys. Rev. D87(2013) 123504 [1302.6994]

  14. [22]

    Y. Li, S. Singh, B. Yu, Y. Feng and U. Seljak,Disconnected Covariance of 2-point Functions in Large-Scale Structure,JCAP01(2019) 016 [1811.05714]. – 15 –

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.