Probing neutrino masses with CMB lensing extraction

TL;DR

This paper assesses how well future CMB experiments can recover the large-scale structure power spectrum through lensing extraction using quadratic estimators, to constrain the total neutrino mass Mnu. Using a Fisher matrix framework that includes lensing power and estimator noise, it forecasts sensitivities for multiple experiments (Planck, SAMPAN, Inflation Probe, and ground-based arrays) under optimistic and pessimistic foreground treatments and for extended cosmological models. The results show substantial improvements in Mnu constraints from lensing alone, with Planck+SAMPAN achieving ~0.1 eV and an ambitious Inflation Probe potentially reaching ~0.035 eV, though discriminating neutrino mass ordering (normal vs inverted) remains challenging with CMB lensing alone. The work highlights the complementarity of CMB lensing with other cosmological probes and the importance of foreground control for robust neutrino mass inferences.

Abstract

We evaluate the ability of future cosmic microwave background (CMB) experiments to measure the power spectrum of large scale structure using quadratic estimators of the weak lensing deflection field. We calculate the sensitivity of upcoming CMB experiments such as BICEP, QUaD, BRAIN, ClOVER and PLANCK to the non-zero total neutrino mass M_nu indicated by current neutrino oscillation data. We find that these experiments greatly benefit from lensing extraction techniques, improving their one-sigma sensitivity to M_nu by a factor of order four. The combination of data from PLANCK and the SAMPAN mini-satellite project would lead to sigma(M_nu) = 0.1 eV, while a value as small as sigma(M_nu) = 0.035 eV is within the reach of a space mission based on bolometers with a passively cooled 3-4 m aperture telescope, representative of the most ambitious projects currently under investigation. We show that our results are robust not only considering possible difficulties in subtracting astrophysical foregrounds from the primary CMB signal but also when the minimal cosmological model (Lambda Mixed Dark Matter) is generalized in order to include a possible scalar tilt running, a constant equation of state parameter for the dark energy and/or extra relativistic degrees of freedom.

Figures (4)

• Figure 1: For six CMB experiments or combinations of experiments, we show the expected noise power spectrum $N_l^{dd}$ for the quadratic estimators $d(a,b)$ built out of pairs $ab \in \{TT, EE, TE, TB, EB\}$, and for the combined minimum variance estimator (mv). The thick line shows for comparison the signal power spectrum $C_l^{dd}=\langle d_l^m d_l^{m*} \rangle$. The sum of the two curves $N_l^{dd}+C_l^{dd}$ represents the expected variance of a single multipole $d(a,b)_l^m$.
• Figure 2: For the same six CMB experiments or combinations of experiments as in figure \ref{['fig_binned_errors']}, we show the expected binned error on the reconstructed power spectra: from top to bottom, $C_l^{dd}$ (using the minimum variance quadratic estimator), $C_l^{TT}$ and $C_l^{EE}$. The curves represent the power spectra of the fiducial model described in section \ref{['sec:results']}.
• Figure 3: Logarithmic derivatives of the lensing power spectrum $C_l^{dd}$ with respect to each cosmological parameter. The derivatives with respect to $\omega_{\rm b}$ and $\omega_{\rm m}$ have been rescaled in order to fit inside the figure.
• Figure 4: 1-$\sigma$ confidence limits on the pairs ($M_{\nu}$, $\theta_i$), for each parameter $\theta_i$ in our eleven-dimensional model. The red solid (green dashed) contours are those expected for Planck (Inflation Probe). For each case, the smaller (larger) ellipse corresponds to the forecasts with (without) lensing extraction.