Forecasts of constraining isotropic cosmic birefringence on AliCPT-1

TL;DR

This paper forecasts isotropic cosmic birefringence constraints using a semi-analytical, cross-frequency maximum-likelihood framework applied to AliCPT and Planck HFI simulations. By jointly fitting the birefringence angle $β$, per-band miscalibration angles $α_i$, and foreground $EB$ amplitude, the authors quantify how sky coverage, foreground complexity, and observing duration affect $σ(β)$, finding $σ(β)≈0.09^ ext{°}$ for one year and $σ(β)≈0.026^ ext{°}$ after four years (for favorable masks). They demonstrate that neglecting foreground $EB$ correlations can bias results by several tenths of a degree or more, while clean sky patches or robust foreground priors mitigate these biases. The results underscore AliCPT's potential to detect or tightly constrain parity-violating physics via cosmic birefringence, with about 11σ sensitivity to a nominal $β=0.3^ ext{°}$ signal in four years under realistic modeling. Overall, the work provides a practical, data-driven forecast for CB constraints in upcoming CMB polarization experiments and highlights the critical role of foreground treatment and sky coverage.

Abstract

Cosmic birefringence (CB) is a promising probe of parity-violating physics beyond the Standard Model, characterized by the rotation of the linear polarization plane of cosmic microwave background (CMB) photons. This effect, quantified by the birefringence angle $β$, generates non-zero $EB$ and $TB$ correlations that are otherwise absent in standard cosmology. However, instrumental miscalibration angles $α$ can mimic this signal, necessitating a joint estimation approach. In this work, we forecast the sensitivity of the AliCPT experiment, combined with Planck HFI data, on constraining the isotropic CB angle using a semi-analytical maximum-likelihood method. We simulate observations under various foreground complexities, rotation angles, and scanning strategies, and demonstrate that AliCPT can achieve an uncertainty of $σ(β)=0.09^\circ$ with one-year data, which will improve to $0.026^\circ$ after four years' observations. We also find that neglecting or mismodeling the foreground $EB$ correlation will introduce significant biases, which can be alleviated under a clean but small sky patch.

Figures (9)

• Figure 1: Masks in Equatorial coordinates. Left: the D30$\mu$K mask adopted for the deep-scan patch, with a sky fraction of 11%. Right: the W90, W80, W70, and W60 masks used for analyses of the wide survey region, indicated with colors from lightest to darkest red, with $f_{\rm sky}=38\%$, 32%, 28%, and 25%, respectively.
• Figure 2: $TB$ (upper) and $EB$ (lower) power spectra of low-complexity (solid) and high-complexity (dashed) foreground templates at 353 GHz for the two survey regions, D30$\mu$K (left blue) and W70 (right yellow). The black squares represent power spectra of the PR4 353 GHz map applied with the two masks respectively.
• Figure 3: Left: Bias of the measured parameters for HFI-only Dataset 1 simulations (low-FG, $\beta=\alpha_i=0$), with the 68% C.L. of the simulations' dispersion taken as the error bar. Right: Uncertainties of fitted parameters. The circles represent the dispersion of simulations, and the curves show the uncertainties predicted by the Fisher matrix.
• Figure 4: Bias of the measured parameters for AliCPT+HFI Dataset 1 (low-FG, $\beta=\alpha_i=0$) simulations.
• Figure 5: Uncertainties of the measured parameters for AliCPT+HFI Dataset 1 (low-FG, $\beta=\alpha_i=0$). The circles and triangles represent the dispersion of simulations, and the curves denote the uncertainties computed from the Fisher matrix.