Pith. Sign in

REVIEW

Uncorrelated Modes of the Nonlinear Power Spectrum

PaperPDF from arXiv

data files

ReviewThe objections and verdict, as JSON Paper breakdownThe authors' claims, conditions and measurements, as JSON Signed receiptProof this review has not been altered
X Bluesky LinkedIn Reddit Hacker News
A. J. S. Hamilton (JILA, U. Colorado)

Astrophysics · Submitted 15 May 1999 · 2 references

Pith's verdict: A genuinely useful methods paper whose central numerical finding is real but rests on a hierarchical model the author himself calls wrong, and whose error-bar machinery quietly drops off-diagonal terms he quantified as non-negligible.

T0 · AI review

How this review was made: An AI referee (selfhosted/deepseek-v4.1-flash) read the paper on October 29, 2026 and answered a fixed list of questions about it. T0 is the starting level every paper gets. The levels describe how deeply Pith has checked the paper, never the authors.

  1. T0 AI review An AI read the paper and answered a fixed list of questions.
  2. T1 Audit Automatic checks that the references and files the paper relies on exist and match.
  3. T2 Compute The paper's code or calculations re-run with the same results.
  4. T3 Formal statement The claims written out in a formal math language a computer can read.
  5. T4 Proof Those claims proved, and the proof checked by computer.

How Pith checks papers →

· 4 major objections · 8 minor · Ask about this paper · The compiled record · Read the full review →
Challenge this review Comment (0) Read on arXiv ↗
Cite ▾
arXiv IDastro-ph/9905191 (version 4)
Pith ID6STQ3L2I, Pith's permanent record of this paper and its review
CitationA. J. S. Hamilton (JILA, U. Colorado). "Uncorrelated Modes of the Nonlinear Power Spectrum." arXiv:astro-ph/9905191 (1999). https://pith.science/paper/6STQ3L2I
Cite this reviewBibTeX for Pith's review of this paper
Claim paper

Why Pith says this

Highlight any text to comment on it.

Referee report

How Pith checks papers → · Full audit of this paper →

The paper

The paper claims

Prewhitening the nonlinear power spectrum — rescaling it so that shot noise becomes uniform — narrows the covariance of power enough that nonlinear modes become almost independent, restoring an FKP-style minimum-variance estimator on scales

The reading

Nonlinear gravitational evolution couples Fourier modes: the measured power at one wavenumber stops being independent of the power at another, so the tidy Gaussian machinery that treats each $k$ as its own measurement breaks down. This paper asks whether a change of variable can undo most of that coupling. The proposed transform, called prewhitening, is a pointwise rescaling of the correlation function, $X(r) = 2[1+xi(r)]^{1/2} - 2$, chosen so that the shot-noise part of the covariance becomes the identity matrix. Once that is done, the author finds the covariance of the transformed power to be far narrower in Fourier space, and the surviving four-point and three-point pieces of that covariance to be almost simultaneously diagonal with almost equal eigenvalues. If that holds, one can build a near-minimum-variance estimator and Fisher matrix for nonlinear power that mirrors, mode for mode, the Feldman-Kaiser-Peacock construction for linear power — which is exactly what a survey like 2dF or SDSS needs in order to quote honest error bars on nonlinear scales.

What carries the argument

The load-bearing object is the prewhitening map together with the decomposition it enables. In real space the prewhitened power is $X(r) = 2xi(r)/{1+[1+xi(r)]^{1/2}}$, equivalently $2[1+xi(r)]^{1/2}-2$, chosen because its first-order fluctuation is $Delta X = Delta xi / [1+xi]^{1/2}$ — the rescaling that whitens the shot-noise term. Writing the reduced pair covariance as $C = 2[K + (n_i^{-1}+n_j^{-1})J + n_i^{-1}n_j^{-1}H]$ and conjugating by $H^{-1/2}$ turns the two-point piece into the unit matrix and leaves prewhitened four-point and three-point matrices $M$ and $L$ that the author finds to be nearly diagonal in a common eigenbasis with nearly equal eigenvalues. That is the mechanism: two of the three covariance pieces are diagonalized exactly and simultaneously, and the third is diagonal only approximately, with the residual controlled by how close $lambda_alpha$ is to $mu_alpha$. The Fisher matrix of prewhitened power then takes the FKP-like form built from pair integrals weighted by $[mu_alpha + n(r)^{-1}]^{-1}$, with the eigenfunctions $phi_alpha(r)$ replacing spherical Bessel modes.

Extended reading notes

Core claim

The paper's central claim is that prewhitening — transforming the power spectrum so that the noise covariance becomes proportional to the unit matrix — greatly narrows the covariance of the nonlinear power spectrum and produces a set of almost uncorrelated nonlinear modes analogous to the Fourier modes of a Gaussian field. Concretely, the reduced covariance of power is written as a sum of a four-point, a three-point, and a two-point piece; the two-point piece is made the identity by construction, and the author shows numerically that the prewhitened four-point and three-point matrices are nearly diagonal in a common basis and have nearly equal eigenvalues, $lambda_alpha approx mu_alpha$, across power-law spectra spanning the full admissible range $1 < gamma < 3$ and across realistic spectra (an observationally derived one and a nonlinearly evolved $Lambda$CDM model). That near-simultaneous diagonalization is what licenses a Fisher matrix and estimator of the prewhitened nonlinear power that take the same form as the FKP expressions for linear power, with the eigenfunctions of the prewhitened four-point matrix standing in for Fourier modes. The paper closes with three worked procedures — gourmet, fine, and fastfood — for extracting prewhitened nonlinear band-powers with error bars from a galaxy survey, and an appendix documenting FFTLog, a fast Hankel/Fourier transform for logarithmically spaced points, used in several of the manipulations.

Load-bearing premise

The near-simultaneous diagonalization of the prewhitened four-point and three-point matrices, and the near-equality of their eigenvalues, are established only within the hierarchical model with constant amplitudes (with the antisymmetric choice $R_b = -R_a$ adopted to keep the Schwarz inequality from being violated), a model the author states is certainly wrong at some level.

Editorial extensions

If this is right

  • Survey analyses gain a principled way to quote nonlinear band-powers with error bars that reflect mode coupling rather than pretending the modes are independent.
  • The Fisher matrix of prewhitened nonlinear power can be computed in advance from the selection function alone, so the constraining power of a planned survey on nonlinear scales can be forecast before any data are taken.
  • The near-equality of the four-point and three-point eigenvalues means a single effective weighting constant per mode suffices, collapsing the estimator to something close to the familiar FKP pipeline applied to a rescaled field.
  • The prewhitened nonlinear power spectrum turns out to sit curiously close to the linear power spectrum for realistic spectra, suggesting the transform absorbs much of what nonlinear evolution does to the shape.
  • FFTLog makes the real-space/Fourier-space transforms practical over the five or so decades of scale a large survey spans, without the $10^5 \times 10^5$ matrices a linearly spaced grid would demand.

Reading between the lines

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

  • If the prewhitened modes really are near-independent, the same transform could be applied to weak-lensing or 21-cm power spectra, where the same nonlinear mode coupling limits how finely a measured spectrum can be binned before neighbouring bins leak into one another.
  • The curious coincidence that $X(k)$ resembles $xi_L(k)$ suggests a diagnostic: a measured departure of prewhitened nonlinear power from the linear spectrum would flag either scale-dependent bias or a failure of the hierarchical ansatz, rather than merely nonlinear growth.
  • A cheap test of the whole construction is to ask whether the decorrelated band-powers from a mock catalogue recover the input prewhitened spectrum within the predicted $E^{-1}_{(alpha beta)}$ errors; a systematic tilt would indicate the three-point matrix is not as diagonal as assumed.
  • Because the estimator stays unbiased for any choice of the weighting matrix $B$ provided the Fisher matrix is read as the asymmetric expression, the practical risk is not bias but error bars that are too small — the failure mode to watch is over-optimistic variances, not shifted band-powers.
Original abstract

Nonlinear evolution causes the galaxy power spectrum to become broadly correlated over different wavenumbers. It is shown that prewhitening the power spectrum - transforming the power spectrum in such a way that the noise covariance becomes proportional to the unit matrix - greatly narrows the covariance of power. The eigenfunctions of the covariance of the prewhitened nonlinear power spectrum provide a set of almost uncorrelated nonlinear modes somewhat analogous to the Fourier modes of the power spectrum itself in the linear, Gaussian regime. These almost uncorrelated modes make it possible to construct a near minimum variance estimator and Fisher matrix of the prewhitened nonlinear power spectrum analogous to the Feldman-Kaiser-Peacock estimator of the linear power spectrum. The paper concludes with summary recipes, in gourmet, fine, and fastfood versions, of how to measure the prewhitened nonlinear power spectrum from a galaxy survey in the FKP approximation. An Appendix presents FFTLog, a code for taking the fast Fourier or Hankel transform of a periodic sequence of logarithmically spaced points, which proves useful in some of the manipulations.

The compiled record

The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Compiled by classifier-1.13-20260917 via OpenRouter, question bank abstract-p1-20260923, 2026-09-23: 1 request, 5,073 input tokens, $0.0002 reported.

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.16 · human participants 0.02 · controlled comparison 0.06 · randomized assignment 0.01 · states quantitative result 0.07 · compares with prior results 0.81 · mentions code or data availability 0.30.

The abstract, sentence by sentence

arXiv listing abstract v4. Results and the main claim are underlined; .

Nonlinear evolution causes the galaxy power spectrum to become broadly correlated over different wavenumbers. It is shown that prewhitening the power spectrum - transforming the power spectrum in such a way that the noise covariance becomes proportional to the unit matrix - greatly narrows the covariance of power. The eigenfunctions of the covariance of the prewhitened nonlinear power spectrum provide a set of almost uncorrelated nonlinear modes somewhat analogous to the Fourier modes of the power spectrum itself in the linear, Gaussian regime. These almost uncorrelated modes make it possible to construct a near minimum variance estimator and Fisher matrix of the prewhitened nonlinear power spectrum analogous to the Feldman-Kaiser-Peacock estimator of the linear power spectrum. The paper concludes with summary recipes, in gourmet, fine, and fastfood versions, of how to measure the prewhitened nonlinear power spectrum from a galaxy survey in the FKP approximation. An Appendix presents FFTLog, a code for taking the fast Fourier or Hankel transform of a periodic sequence of logarithmically spaced points, which proves useful in some of the manipulations.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#sentencerolethe authors' own workstates the main conclusionstrength
s0 Nonlinear evolution causes the galaxy power spectrum to become broadly correlated over different wavenumbers. background 0.99 0.20 0.07 asserted 1.00
s1 It is shown that prewhitening the power spectrum - transforming the power spectrum in such a way that the noise covariance becomes proportional to the unit matrix - greatly narrows the covariance of power. result 0.43 0.85 0.18 asserted 1.00
s2 The eigenfunctions of the covariance of the prewhitened nonlinear power spectrum provide a set of almost uncorrelated nonlinear modes somewhat analogous to the Fourier modes of the power spectrum itself in the linear, Gaussian regime. result 0.67 0.88 0.47 asserted 0.99
s3 These almost uncorrelated modes make it possible to construct a near minimum variance estimator and Fisher matrix of the prewhitened nonlinear power spectrum analogous to the Feldman-Kaiser-Peacock estimator of the linear power spectrum. contribution 0.57 0.87 0.34 asserted 0.72
s4 The paper concludes with summary recipes, in gourmet, fine, and fastfood versions, of how to measure the prewhitened nonlinear power spectrum from a galaxy survey in the FKP approximation. method 0.63 0.72 0.14 asserted 0.97
s5 An Appendix presents FFTLog, a code for taking the fast Fourier or Hankel transform of a periodic sequence of logarithmically spaced points, which proves useful in some of the manipulations. method 0.51 0.74 0.02 asserted 0.98

What kind of paper the classifier reports this is: method or tool 0.97 · evidence from proof or derivation 0.82.

What was examined, and what was not

sources read
  • arXiv listing abstract v4
not examined
  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers
levels run
  • P abstract
not run
  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

Figures
Figure 1. Schematic illustration of the 4-point, 3-point, and 2- point contributions to the covariance Cijkl of pairs ij with other pairs kl. The 3-point and 2-point contributions are shot-noise contributions in which one or both galaxies of the pair ij are the same as one or both of the pair kl. ucts of overdensities, equation (36), is, in the real represen￾tation with no implicit summation, Cijkl = ξikξjl + ξilξjk + ηijkl +
Figure 2. Correlation coefficient K(kα,kβ)/[K(kα,kα)K(kβ, kβ)]1/2 of the 4-point contribution K(kα, kβ) to the covari￾ance of the power (i.e. the covariance without shot-noise) in the case of a power law power spectrum with correlation function ξ(r) = (r/5 h −1Mpc)−1.8 . Each line is the correlation coefficient for a fixed kβ, and each line peaks at kα = kβ, whereat the value is unity. The hierarchical amplitudes are Ra = −Rb = 4.195. The resolution is 128 points per decade, ∆ log k = 1/128. the (unprewhitened) reduced covariance matrix Cαβ, equa￾tion (42), for the case of a power law power…
Figure 4. Correlation coefficients (top) K(kα, kβ)/[K(kα, kα)K(kβ, kβ)]1/2 of the covariance, and (bottom) M(kα, kβ)/[M(kα, kα) M(kβ, kβ)]1/2 of the prewhitened covariance, of the power spectrum in four different models of the power spectrum. Each curve is the correlation coefficient at fixed kβ = 1 h Mpc−1 , plotted as a function of kα. The three sets of panels starting from the left are for power law power spectra with correlation functions ξ(r) = (r/5 h −1Mpc)−γ with indices γ = 1.1, 1.8, and 2.9, while the rightmost panel is for the ΛCDM power spectrum of Eisenstein & Hu (1998) with ΩΛ =…
Figure 5. Correlation coefficient Lαβ/(LααLββ) 1/2 of the pre￾whitened 3-point covariance Lαβ in the representation of eigen￾functions φα of the prewhitened 4-point covariance Mαβ, for a power law power spectrum with correlation function ξ(r) = (r/5 h−1Mpc)−1.8 . Each line is the correlation coefficient at a fixed nominal wavenumber kβ, plotted against the nominal wavenumber kα, which labels the 4-point eigenfunctions φα or￾dered by eigenvalue. Each line peaks at kα = kβ, whereat the correlation coefficient is unity. The upper panel is for 4-point hi￾erarchical amplitudes Rb = −Ra; the lower…
Figure 6. Correlation coefficient Lαβ/(LααLββ) 1/2 of the prewhitened 3-point covariance Lαβ in the representation of eigenfunctions φα of the prewhitened 4-point covariance Mαβ, at a representative nominal wavenumber kβ = 1 h Mpc−1 . The horizontal axis is the nominal wavenumber kα, which labels the 4-point eigenfunctions φα ordered by eigenvalue. The three sets of panels starting from the left are for power law power spectra with correlation functions ξ(r) = (r/5 h −1Mpc)−γ with indices γ = 1.1, 1.8, and 2.9, while the rightmost panel is for the ΛCDM power spectrum of Eisenstein & Hu (1998)…
Figure 7. Ratio µα/ξ(kα) of the eigenvalue µα of the 4-point prewhitened covariance matrix M to the nonlinear power spec￾trum ξ(kα), as a function of the nominal wavenumber kα, which labels the 4-point eigenfunctions φα ordered by eigenvalue, for a power law power spectrum with correlation function ξ(r) = (r/5 h −1Mpc)−1.8 . The relation between eigenvalue and nominal wavenumber varies with resolution. The low resolution case has ∆ log k = 1/32; the solid line is for Rb = −Ra, the long-dashed line for Rb = Ra. The high resolution case has ∆ log k = 1/128; here the (dashed) Rb = −Ra and…
Figure 9. Sample of (discretized) eigenfunctions φα(k)[4πk3 ∆ ln k/(2π) 3 ] 1/2 of the prewhitened 4-point covariance Mαβ for a power law power spectrum with correlation function ξ(r) = (r/5 h−1Mpc)−1.8 , and Rb = −Ra, at resolutions of (top) ∆ log k = 1/32 and (bottom) ∆ log k = 1/128. The eigenfunc￾tions have nominal wavenumbers kα of 0.1, 1, and 10 h Mpc−1 . In the hierarchical regime, the eigenfunctions grow wigglier as the resolution increases. Examination of the eigenfunctions of the 4-point and 3-point matrices M and L reveals at least part of the rea￾son why their eigenvalues seem to…
Figure 10. Linear power spectrum ξL(k), nonlinear power spectrum ξ(k), and prewhitened nonlinear power spectrum X(k) for (left) the Ωm = 0.3 power spectrum derived from observations by Peacock (1997), and (right) the COBE-normalized ΛCDM power spectrum from the fitting formulae of Eisenstein & Hu (1998), with parameters as listed on the graph. The nonlinear power spectra were computed from the linear power spectra according to the formula of Peacock & Dodds (1996). The ΛCDM power spectrum is the one used in Figures 4, 6, 8, 11, and 12. are attached to the model, not to the data’, to quote…
Figure 11. shows the ratio µ(k)/ξ(k) of the effective FKP constant µ(k) to the nonlinear power spectrum ξ(k) for sev￾eral different power spectra. The ratios plotted in
Figure 12. shows the correlation function ξ(r) computed by FFTLog for the nonlinear ΛCDM power spectrum of Eisen￾stein & Hu (1998) shown in
Figure 13. illustrates these steps for the unbiased (q = 0) Hankel transform, equation (155), of order µ = −1/2 of a function that is Gaussian in the log a(r) = exp[−(ln r) 2 /2] . (188) Truncation of the function a(r) leads to ringing of its transform ˜a(k) at high frequencies k, as seen in the middle right panel of

Showing 11 of the paper's 13 figures. See all figures in the PDF ↗ Figures reproduced from arXiv:astro-ph/9905191 by the authors.

References
How later papers used it

133 later papers cite it

Evidence: substantial. 133 of the citing papers share no author with it, and 107 of those carry a confident label from Pith's citation classifier.

Pith read 21 of the 189 citing sentences, choosing only those whose citation markers point at this paper alone, and wrote the paraphrases from them. The summary, the counts, the evidence level, the links and the quotes come from code; the summary counts the papers listed below. A later paper is listed under a label when Pith's citation classifier gives one of its sentences that label at probability 0.8 or more (0.95 for support), Pith's reading finds that the citing paper itself did it, and a second check confirms the paraphrase against the sentence. This section lists uses only. It does not list criticism or disagreement, so their absence here says nothing about how the paper has been received. Papers that share an author with this one are listed apart, under "By the same authors". A quote is the citing authors' own sentence, shown when their paper's licence is CC BY, CC BY-SA or CC0. JSON

Uses its method or tool 12 of 107 papers with this label
How the literature uses it

85 arXiv papers cite this in 128 sentences

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

Papers citing it · 30 sentences without a confident label

Builds on it 1 paper
Uses its method or tool 66 papers
Cites it as background 3 papers

Discussion

Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.