Cosmic birefringence from a joint analysis of ACT and Planck

Technical Summary: Cosmic Birefringence from a Joint Analysis of ACT and Planck

Problem and Motivation
The standard $\Lambda$CDM cosmological model is parity-symmetric, yet extensions involving ultra-light pseudoscalar fields (such as axion-like particles) coupled to electromagnetism via a Chern-Simons term predict a parity-violating rotation of the plane of linearly polarized light from the Cosmic Microwave Background (CMB). This phenomenon, known as cosmic birefringence, is characterized by an angle $\beta$. A non-zero $\beta$ would mix intrinsic CMB E-modes and B-modes, generating a non-zero EB power spectrum ($C^{EB}_{\ell}$), which is otherwise zero in $\Lambda$CDM.

Recent independent analyses of Planck Data Release 4 (DR4) and Atacama Cosmology Telescope (ACT) Data Release 6 (DR6) have hinted at a non-zero $\beta$. However, measuring $\beta$ is complicated by instrumental miscalibration angles ($\alpha$), which rotate the observed polarization plane. Without independent constraints, $\alpha$ and $\beta$ are degenerate ($\beta_{obs} = \alpha + \beta$). While Planck has utilized polarized foreground emission to break this degeneracy, and ACT has relied on instrumental priors derived from optics models, a joint analysis has not yet been performed to leverage the complementary strengths of both datasets.

Methodology
This work presents a joint analysis of the polarized maps from ACT DR6 and Planck DR4 (NPIPE release). The methodology involves:

1. Data Combination: The analysis utilizes auto-power spectra (ACT$\times$ACT, Planck$\times$Planck) and cross-power spectra (ACT$\times$Planck) across overlapping multipole ranges ($600 \lesssim \ell \lesssim 2000$). The cross-spectra are critical because instrumental systematics and noise between the two telescopes are uncorrelated.
2. Likelihood Framework: The author employs a likelihood function (Eq. 3) that includes a multi-band generalization of the observed EB spectrum. This framework incorporates:
   • A filamentary dust model for Galactic foreground EB correlations, parameterized by amplitudes $A_\ell$ and a misalignment angle $\psi_\ell$ derived from 353 GHz dust-dominated spectra.
   • Gaussian priors on the miscalibration angles $\alpha_i$ for both experiments. For ACT, these priors are derived from pre-flight optics models (Ref. [26]). For Planck, priors are derived from pre-flight focal plane orientation limits, though the analysis primarily relies on foreground emission to constrain $\alpha$ for Planck.
3. Pipeline: A custom pipeline calculates power spectra and covariances. While the ACT-only block uses the official DR6 covariance, the cross-spectra and Planck-only blocks utilize the author's pipeline to ensure consistency.
4. Robustness Tests: To mitigate potential biases from foreground modeling or instrumental priors, the author performs a robustness test masking the Galactic plane (30% cut) and restricting analysis to high multipoles ($\ell > 600$) where CMB dominates over foregrounds.

Key Contributions

• First Joint Analysis: This is the first study to jointly analyze ACT and Planck data for isotropic cosmic birefringence, breaking the $\alpha + \beta$ degeneracy using two independent methods simultaneously (foreground emission for Planck and instrumental priors for ACT).
• Cross-Spectrum Constraints: The work demonstrates the constraining power of the ACT$\times$Planck cross-power spectra, where the datasets overlap, providing a measurement less susceptible to the specific systematic vulnerabilities of either telescope alone.
• Systematic Mitigation: By combining datasets with different systematic dependencies, the author shows that recovering $\beta = 0^\circ$ would require simultaneous failures of both the dust EB modeling and the instrumental priors.

Results

• Primary Measurement: The baseline joint analysis yields a cosmic birefringence angle of $\beta = 0.277^\circ \pm 0.057^\circ$. This result excludes the null hypothesis ($\beta = 0^\circ$) with a statistical significance of 4.8$\sigma$.
• Subsets:
  • ACT-only (using pipeline): $\beta = 0.207^\circ \pm 0.073^\circ$.
  • Planck-only (using filamentary dust model): $\beta = 0.387^\circ \pm 0.094^\circ$.
  • ACT$\times$Planck cross-spectra: $\beta = 0.178^\circ \pm 0.078^\circ$ (2.3$\sigma$).
• Robustness: When masking the Galactic plane and removing foreground-dominated bands (30, 44, 353 GHz) to rely on high-$\ell$ CMB data and ACT priors, the result is $\beta = 0.236^\circ \pm 0.067^\circ$, excluding zero at 3.5$\sigma$. This confirms the signal persists even when foreground modeling is minimized.
• Goodness of Fit: The joint analysis yields a probability to exceed (PTE) of 9.7%. The low PTE is driven largely by the ACT$\times$ACT block (PTE = 0.1%), which the author attributes to a known poor fit in the ACT data around $\ell \approx 1800-1900$. The Planck$\times$Planck and ACT$\times$Planck blocks show excellent fits (PTE > 58%).

Significance and Claims
The paper claims that the detection of a non-zero cosmic birefringence angle is robust against the specific systematic uncertainties inherent to either dataset alone. The result suggests a parity-violating extension to the standard model, potentially pointing to ultra-light pseudoscalar fields in the dark sector.

However, the author maintains a modest tone regarding cosmological conclusions. They explicitly state that "unresolved systematics in the data must be understood before we can draw strong cosmological conclusions." The signal cannot yet be definitively attributed to new physics because unknown systematics in neither Planck nor ACT can be fully excluded. The author emphasizes that confirming the cosmological origin of this signal requires a deeper understanding of both datasets, particularly the instrumental systematics in ACT and the foreground modeling in Planck. The work serves as a strong statistical hint rather than a definitive discovery, highlighting the need for future data and improved systematic characterization.