Constraints on primordial non-Gaussianity from Planck PR4 data

TL;DR

The paper analyzes Planck PR4 CMB data to constrain primordial non-Gaussianity (PNG) using bispectrum-based methods. It employs two estimators—the binned bispectrum and the integrated bispectrum—to test primordial shapes (local, equilateral, orthogonal) and secondary shapes from lensing, unclustered point sources, and the CIB, with validation on 600 simulations. The resulting constraints on the PNG amplitudes are $f_{\mathrm{NL}}^{\mathrm{local}} = -0.1 \pm 5.0$, $f_{\mathrm{NL}}^{\mathrm{equil}} = 6 \pm 46$, and $f_{\mathrm{NL}}^{\mathrm{ortho}} = -8 \pm 21$ (T+E), consistent with zero and representing the tightest Planck limits to date; lensing and PS bispectra are detected as expected, while the CIB remains undetected in the joint analysis. A systematic low lensing amplitude observed in the simulations does not impact the observational results, underscoring the robustness of the PR4 PNG constraints.

Abstract

We perform the first bispectrum analysis of the final Planck release temperature and E-polarization CMB data, called PR4. We use the binned bispectrum estimator pipeline that was also used for the previous Planck releases as well as the integrated bispectrum estimator. We test the standard primordial (local, equilateral and orthogonal) and secondary (lensing, unclustered point sources and CIB) bispectrum shapes. The final primordial results of the full T+E analysis are $f_\mathrm{NL}^\mathrm{local} = -0.1 \pm 5.0$, $f_\mathrm{NL}^\mathrm{equil} = 6 \pm 46$ and $f_\mathrm{NL}^\mathrm{ortho} = -8 \pm 21$. These results are consistent with previous Planck releases, but have slightly smaller error bars than in PR3, up to $12\%$ smaller for orthogonal. They represent the best Planck constraints on primordial non-Gaussianity. The lensing and point source bispectra are also detected, consistent with PR3. We perform several validation tests and find in particular that the 600 simulations, used to determine the linear correction term and the error bars, have a systematically low lensing bispectrum. We show however that this has no impact on our results.

Figures (2)

• Figure 1: The full and the noise-only power spectra determined from SEVEM PR3 and PR4 simulations, for temperature (left) and E-polarization (right).
• Figure 2: Estimated $\hat{f}_\mathrm{NL}^{local}$ (left) and its error bar $\sigma(\hat{f}_\mathrm{NL}^{local})$ (right) using different subsets of the PR4 simulations for linear correction and error bars, as a function of the lensing amplitude in the corresponding subset, determined using the integrated bispectrum estimator. Each subset consists of $100$ maps drawn randomly from the $600$SEVEM PR4 simulations. Note that the expected lensing bias of $4.3$ on the estimated $\hat{f}_\mathrm{NL}^{local}$ has not been subtracted.