Abstract: We compute upper limits on cosmic microwave background cross-polarization by cross-correlating the PIQUE and Saskatoon experiments. We also discuss theoretical and practical issues relevant to measuring cross-polarization and illustrate them with simulations of the upcoming BOOMERANG 2002 experiment. We present a method that separates all six polarization power spectra (TT, EE, BB, TE, TB, EB) without any other "leakage" than the familiar EE-BB mixing caused by incomplete sky coverage. Since E and B get mixed, one might expect leakage between TE and TB, between EE and EB and between BB and EB—our method eliminates this by preserving the parity symmetry under which TB and EB are odd and the other four power spectra are even.
Introduction
By 2002, clear maps of the Cosmic Microwave Background (CMB) temperature fluctuations had been established, but its polarization remained an observational frontier. Theory predicted that polarization should exist at roughly 1% to 10% of the temperature signal's strength. This paper represents a foundational moment in observational cosmology: the first serious attempt to measure the "cross-talk" between the temperature of the CMB and its polarization.
The Challenge of Polarization and Leakage
Polarization is incredibly faint, making direct detection (autocorrelation) difficult. However, cosmologists realized they could look for how polarization correlates with the temperature map (cross-correlation). This $TE$ signal is statistically easier to extract from noise.
The primary technical hurdle is "leakage." Because telescopes only observe small patches of the sky, the mathematical process of separating $E$-modes (gradient-like) and $B$-modes (curl-like) often causes them to bleed into each other, creating false signals.
Methodology: Parity Symmetry and No-Leakage
The authors employed a Quadratic Estimator (QE) method to extract the power spectra. Their primary innovation was a geometric approach to handle the leakage problem:
- Parity Symmetry: In standard physics, the CMB should be "parity-even," meaning certain correlations ($TB$ and $EB$) should be zero.
- The No-Leakage Theorem: The authors proved that by choosing specific priors (assuming cross-correlations are zero during estimation), they could exploit this symmetry to eliminate 14 out of 15 possible types of leakage.
- Correlation Coefficients: Instead of raw power ($C_\ell$), they focused on a dimensionless correlation coefficient ($r_\ell$), which ranges from -1 to 1, making it easier to visualize the synchronization of temperature and polarization.
Datasets and Simulations
The study utilized two primary datasets:
- PIQUE (Princeton IQU Experiment): A ground-based experiment measuring polarization near the North Celestial Pole.
- Saskatoon (SK): Provided unpolarized temperature ($T$) data for the same region.
To validate their method, the authors simulated the then-upcoming BOOMERanG 2002 balloon flight. These simulations demonstrated that their quadratic estimator could successfully separate all six polarization power spectra ($TT, EE, BB, TE, TB, EB$) with minimal error, even in the presence of $B$-modes.
Key Insights and Results
- $TE$ as a "Smoking Gun": The $TE$ correlation is sensitive to primordial gravity waves, which tend to suppress the signal at large scales.
- Upper Limits: The study established 95% confidence limits for the $TE$ correlation ($X < 8.6 \mu K$) and the $TB$ correlation ($Y < 14.1 \mu K$).
- Consistency: Results were consistent with the "concordance model" ($\Lambda$CDM), which predicted signals just below these detection limits.
Conclusion
While instrumental noise in 2002 prevented a definitive discovery, this paper established the methodological "goalposts" for the field. The "no-leakage" approach and the focus on cross-correlation set the stage for the WMAP and Planck satellites, which eventually confirmed these predictions with high precision.
Summary updated