How to measure CMB polarization power spectra without losing information

Max Tegmark, A. de Oliveira‐CostaPhysical 2001 · DOI · arXiv

Abstract: We present a method for measuring CMB polarization power spectra given incomplete sky coverage and test it with simulated examples such as Boomerang 2001 and MAP. By augmenting the quadratic estimator method with an additional step, we find that the E and B power spectra can be effectively disentangled on angular scales substantially smaller than the width of the sky patch in the narrowest direction. We find that the basic quadratic and maximum-likelihood methods display an unneccesary sensitivity to systematic errors when $T\ensuremath{-}E$ cross-correlation is involved, and show how this problem can be eliminated at negligible cost in increased error bars. We also test numerically the widely used approximation that sample variance scales inversely with sky coverage, and find it to be an excellent approximation on scales substantially smaller than the sky patch.

Introduction

The mapping of Cosmic Microwave Background (CMB) temperature fluctuations has provided a wealth of cosmological data. However, the next major challenge is the detection of its much fainter polarization. Polarization is decomposed into two components: E-modes (curl-free) and B-modes (divergence-free). While E-modes provide insights into the early plasma, B-modes are considered the "holy grail" of cosmology, potentially carrying the signature of primordial gravitational waves from the Big Bang.

A significant mathematical hurdle arises because the transformation from raw sky maps to E and B modes is non-local. While separation is perfect on a full sky, real-world experiments typically observe only a "patch" of the sky. This "cut sky" geometry induces leakage, where B-modes contaminate E-mode estimates and vice versa. This paper introduces an optimized "quadratic estimator" pipeline designed to extract polarization power spectra from incomplete maps without information loss.

Methodology

The authors refine the Quadratic Estimator (QE) and Maximum Likelihood (ML) frameworks through two key innovations:

  1. The Disentangling Step: Standard estimators often mix E and B signals. The authors introduce an additional step in the QE process by choosing a weight matrix that forces the leakage matrix to be the identity. This ensures that, on average, the E-mode estimate is not contaminated by B-mode power.
  2. Cross-Correlation Decoupling: Temperature ($T$) and E-modes are physically correlated. Standard algorithms use $T$ data to improve $E$ signal estimation. However, systematic errors in $T$ maps can leak into polarization results. The authors propose setting the $T-E$ cross-correlation to zero in the prior model used for weighting, effectively insulating polarization measurements from temperature-based systematics at a negligible cost (approx. 1%) in error bar size.

Key Insights and Results

The study evaluated these methods using simulated models based on experiments like COBE, MAP (WMAP), Planck, and Boomerang (B2001).

Geometry and Scale

A primary finding is that E/B leakage depends almost entirely on the geometry of the sky patch rather than detector sensitivity. Leakage is most pronounced at large angular scales (low multipole $l$) and decreases significantly as $l$ increases relative to the patch width.

Leakage as a function of multipole l for different sky geometries.

As shown in Figure 6, the leakage for the B2001 geometry drops rapidly, whereas a 1D "Circle" geometry retains high leakage across scales. This confirms that E/B separation is highly effective for angular scales significantly smaller than the narrowest width of the sky patch.

Performance of the Estimator

The quadratic estimator is demonstrated to be "lossless," distilling cosmological information from millions of pixels into band-powers without wasting signal-to-noise. The window functions for these estimators show sharp peaks at the target multipoles with minimal leakage at high $l$.

Window functions for E and B modes at l=70 for the B2001 experiment.

In Figure 1, the window function for the B2001 experiment at $l=70$ illustrates this successful isolation. The E-mode estimate (top panel) is strongly peaked at the target scale, while the B-mode leakage (bottom panel) remains minimal.

Conclusion

The proposed method provides a robust framework for upcoming CMB polarization missions. It validates the $1/f_{sky}$ rule for sample variance and offers an analytical shortcut for error bar forecasting that matches expensive numerical simulations. While pixel effects and potential non-Gaussianity remain areas for future refinement, the "disentangled" quadratic estimator stands as an optimal tool for the first definitive detections of CMB polarization.

Summary updated

Abstract

We present a method for measuring CMB polarization power spectra given incomplete sky coverage and test it with simulated examples such as Boomerang 2001 and MAP. By augmenting the quadratic estimator method with an additional step, we find that the E and B power spectra can be effectively disentangled on angular scales substantially smaller than the width of the sky patch in the narrowest direction. We find that the basic quadratic and maximum-likelihood methods display an unneccesary sensitivity to systematic errors when $T\ensuremath{-}E$ cross-correlation is involved, and show how this problem can be eliminated at negligible cost in increased error bars. We also test numerically the widely used approximation that sample variance scales inversely with sky coverage, and find it to be an excellent approximation on scales substantially smaller than the sky patch.
BibTeX
@inproceedings{tegmark2001how,
  author = {Max Tegmark and A. de Oliveira‐Costa},
  title = {{How to measure CMB polarization power spectra without losing information}},
  booktitle = {Physical},
  year = {2001},
  doi = {10.1103/physrevd.64.063001},
}

Loading related content...

Figure