Pith. Sign in

REVIEW

Estimating the Power Spectrum of the Cosmic Microwave Background

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
J. R. Bond (1), A. H. Jaffe (2), L. Knox (1) ((1) CITA, U. Toronto, (2) CfPA, UC Berkeley)

Astrophysics · Submitted 22 Aug 1997 · 40 references

Pith's verdict: The fixed-point argument is short and right, the Fisher-derived filter is the durable contribution, and the paper is honest about the two places where the claim narrows — unimodality and negative bandpowers at high ℓ.

T0 · AI review

How this review was made: An AI referee (selfhosted/deepseek-v4.1-flash) read the paper on November 11, 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 →

· 2 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/9708203 (version 3)
Pith IDVPGIUOLB, Pith's permanent record of this paper and its review
Citation(2) CfPA, A. H. Jaffe (2), J. R. Bond (1), L. Knox (1) ((1) CITA, UC Berkeley), U. Toronto. "Estimating the Power Spectrum of the Cosmic Microwave Background." arXiv:astro-ph/9708203 (1997). https://pith.science/paper/VPGIUOLB
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

Iterating a minimum-variance quadratic estimator converges to the exact maximum-likelihood CMB power spectrum, making likelihood-quality bandpowers affordable for real chopping experiments.

The reading

Cosmic microwave background experiments do not measure the power spectrum directly. They record temperature differences between pairs of points on the sky, with noise, chopping patterns and a galactic cut, and the theorist wants the amplitude of fluctuations at each angular scale. The honest way to do this is to write down the Gaussian likelihood of the data as a function of the power spectrum and find its peak. That requires inverting a large covariance matrix many times, which is expensive and gets worse as the cube of the number of data points. The cheap way is a quadratic estimator: a fixed weighted sum over all pairs of data points, designed to have minimum variance. The folklore objection was that the quadratic estimator is only an approximation to the likelihood answer. This paper shows the objection is wrong. Both the direct likelihood search and the quadratic estimator have the same fixed point, because both corrections vanish exactly when the derivative of the log-likelihood vanishes. So you can start anywhere, apply the quadratic estimator, update the power spectrum, and repeat, and you converge on the true likelihood maximum. The authors demonstrate this on the COBE/DMR map (999 pixels after the galactic cut) and on the Saskatoon data (2928 time-stream differences compressed to 1200 signal-to-noise eigenmodes), recovering every multipole from the quadrupole to about ℓ = 35 for DMR and ten bands from ℓ = 19 to 499 for Saskatoon. They also show that the likelihood is strongly skewed at low multipoles, so the Gaussian error bars from the quadratic method are not the whole story, and they propose replacing the traditional window-function filter with one derived from the Fisher matrix.
Extended reading notes

Core claim

The central assertion, from the abstract and Section II D, is that 'applied iteratively, the quadratic estimator is not distinct from likelihood analysis, but is rather a rapid means of finding the power spectrum that maximizes the likelihood function.' The supporting argument is a fixed-point identity: the quadratic step is delta a_p = (1/2) sum (F-bar)^{-1} Tr[(Delta Delta^T - C) C^{-1} C_{T,p} C^{-1}], and the Fisher matrix F-bar is invertible, so delta a_p = 0 exactly when the gradient of ln L vanishes. If correct, a single O(N^3) factorization plus cheap quadratic evaluations replaces repeated full likelihood searches, and the resulting bandpowers and Fisher errors are the maximum-likelihood ones.

Load-bearing premise

The claim that the two schemes 'converge to the same C_ls' rests on the likelihood surface being well behaved enough that the iteration actually reaches the global maximum, not merely a stationary point. The fixed-point identity in Section II D guarantees only that a converged point is a stationary point. The paper asserts the rest from inspection: 'we have found that the space is sufficiently structureless that a simple iteration procedure works well' (Section II B), and later concedes that 'our iterative direct evaluation method for finding the peak may not converge to a unique maximum, as values oscillate between iterations' (Section III). Newton-Raphson carries 'the risk of getting trapped in a local extremum' (Section II C). No general convergence proof is offered; the evidence is two datasets and a handful of iterations.

Original abstract

We develop two methods for estimating the power spectrum, C_l, of the cosmic microwave background (CMB) from data and apply them to the COBE/DMR and Saskatoon datasets. One method involves a direct evaluation of the likelihood function, and the other is an estimator that is a minimum-variance weighted quadratic function of the data. Applied iteratively, the quadratic estimator is not distinct from likelihood analysis, but is rather a rapid means of finding the power spectrum that maximizes the likelihood function. Our results bear this out: direct evaluation and quadratic estimation converge to the same C_ls. The quadratic estimator can also be used to directly determine cosmological parameters and their uncertainties. While the two methods both require O(N^3) operations, the quadratic is much faster, and both are applicable to datasets with arbitrary chopping patterns and noise correlations. We also discuss approximations that may reduce it to O(N^2) thus making it practical for forthcoming megapixel datasets.

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, 5,555 input tokens, $0.0002 reported.

What the abstract itself states, as probabilities: claims first 0.03 · states limitation 0.75 · human participants 0.01 · controlled comparison 0.04 · randomized assignment 0.01 · states quantitative result 0.18 · compares with prior results 0.76 · mentions code or data availability 0.02.

The abstract, sentence by sentence

arXiv listing abstract v3. Results and the main claim are underlined; .

We develop two methods for estimating the power spectrum, C_l, of the cosmic microwave background (CMB) from data and apply them to the COBE/DMR and Saskatoon datasets. One method involves a direct evaluation of the likelihood function, and the other is an estimator that is a minimum-variance weighted quadratic function of the data. Applied iteratively, the quadratic estimator is not distinct from likelihood analysis, but is rather a rapid means of finding the power spectrum that maximizes the likelihood function. Our results bear this out: direct evaluation and quadratic estimation converge to the same C_ls. The quadratic estimator can also be used to directly determine cosmological parameters and their uncertainties. While the two methods both require O(N^3) operations, the quadratic is much faster, and both are applicable to datasets with arbitrary chopping patterns and noise correlations. We also discuss approximations that may reduce it to O(N^2) thus making it practical for forthcoming megapixel datasets.

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 We develop two methods for estimating the power spectrum, C_l, of the cosmic microwave background (CMB) from data and apply them to the COBE/DMR and Saskatoon datasets. contribution 1.00 0.95 0.09 asserted 1.00
s1 One method involves a direct evaluation of the likelihood function, and the other is an estimator that is a minimum-variance weighted quadratic function of the data. method 0.95 0.79 0.05 asserted 1.00
s2 Applied iteratively, the quadratic estimator is not distinct from likelihood analysis, but is rather a rapid means of finding the power spectrum that maximizes the likelihood function. interpretation 0.62 0.47 0.23 asserted 1.00
s3 Our results bear this out: direct evaluation and quadratic estimation converge to the same C_ls. result 0.98 0.95 0.44 asserted 1.00
s4 The quadratic estimator can also be used to directly determine cosmological parameters and their uncertainties. contribution 0.74 0.47 0.10 asserted 0.55
s5 While the two methods both require O(N^3) operations, the quadratic is much faster, and both are applicable to datasets with arbitrary chopping patterns and noise correlations. result 0.31 0.77 0.10 asserted 1.00
s6 We also discuss approximations that may reduce it to O(N^2) thus making it practical for forthcoming megapixel datasets. future work 0.49 0.57 0.04 hedged possibility 1.00

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from analysis of existing data 0.98.

What was examined, and what was not

sources read
  • arXiv listing abstract v3
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. FIG. 1. A one-dimensional example of quadratic estimation.
Figure 2. FIG. 2. Maximum-likelihood power spectra from iterative
Figure 3. FIG. 3. Iterative quadratic estimation. The curve is the ze
Figure 4. FIG. 4. We compare the results of the quadratic and direct
Figure 5. FIG. 5. Comparison of different groups’ power spectrum
Figure 6. FIG. 6. Probability Distributions for individual
Figure 8. FIG. 8. Quadratic estimates of the power in 10 bins, de
Figure 9. FIG. 9. Probability Distributions for the power in bands,
Figure 10. FIG. 10. DMR with ortho-rebinning
Figure 11. FIG. 11. SK with ortho-rebinning
Figure 13. FIG. 13. The Fisher matrix sum,
Figure 14. FIG. 14. Slices of the SK 26 band covariance matrix at
Figure 15. FIG. 15. Comparison of forecasts for the two balloon ex
Figure 16. FIG. 16. The forecasted data with error bars are shown in

Showing 14 of the paper's 16 figures. See all figures in the PDF ↗ Figures reproduced from arXiv:astro-ph/9708203 by the authors.

References
How later papers used it

83 later papers cite it

Evidence: substantial. 78 of the citing papers share no author with it, and 27 of those carry a confident label from Pith's citation classifier.

Pith read 13 of the 108 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 3 of 27 papers with this label
How the literature uses it

52 arXiv papers cite this in 71 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 · 32 sentences without a confident label

Uses its method or tool 20 papers
  • The correction to the parameter $a_p$ can then be written as [ref] δa_p = 1/2∑_p' (F^-1)_pp' Tr[ ( Δ Δ^T - )( ^-1 _,p' ^-1 ) ].

    Data compression of measurements of peculiar velocities of Supernovae Ia · Optimal quadratic estimator · 0.93

  • This is described in [ref] , and yields the iterative algorithm: [equation] where, in Eq. [ref] there is an implicit summation over $j$, the sub-index $_,i$ implies differentiation with respect to $q_i$, $$ is the Fisher matrix, $q_i^n$ is the $n-$th iteration of the solution for $q_i$ and the previous iteration $q_i^n-1$ is used to compute $$ and $_,i$ in the second term.

    Calibrating photometric redshifts with intensity mapping observations · Forecasting formalism · 0.97

  • Assuming the data $d̃^k$, with $k$ running over frequencies $f_k$, to be Gaussian variables with zero means and described only by their variance $C_IJ(f_k,)$, the log-likelihood can be written as [ref] [equation] where $$ denotes the set of model parameters, including the coefficients $c_ℓm$ parametrizing the anisotropies in the GWB.

    Pulsar timing array sensitivity to anisotropies in the gravitational wave background · Fisher forecast of sensitivities · 0.91

Cites it as background 14 papers
  • There is a large literature of unbiased power spectrum estimators that are optimal, have well-understood statistical properties, and robustly meet common observational challenges such as partial sky coverage, linear sky contamination, and inhomogeneous noise [ref] .

    Catalog-based pseudo-$C_\ell$s · Introduction · 0.86

  • The basis of our discussion will be the quadratic estimator formalism, which has a long history in the CMB and galaxy survey literature (e.g., Refs. [ref] ), and was explicitly adapted for $21 cm$ cosmology in Refs. [ref] .

    The Epoch of Reionization Window: I. Mathematical Formalism · Quadratic estimator formalism · 0.87

  • These include a variety of (quasi-)optimal quadratic power spectrum estimators [ref] (which saw significant application in the early 2000s [ref] ), approximate power spectrum weighting schemes [ref] , sub-optimal but efficient bispectrum estimators [ref] , as well as extension to anisotropy [ref] , and the analogous methods for two-sphere observables (including [ref] for the two-point function and [ref] for higher-point functions).

    PolyBin3D: A Suite of Optimal and Efficient Power Spectrum and Bispectrum Estimators for Large-Scale Structure · Introduction · 0.97

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.