The promising future of a robust cosmological neutrino mass measurement

TL;DR

This work forecasts how well future cosmological data can constrain the total neutrino mass $M_ u$ across a grid of cosmological models and experimental configurations. Using MCMC forecasts with mock likelihoods implemented in MontePython-CLASS, and conservatively excluding non-linear scale information, the authors compare 35 combinations of CMB and LSS data, including LiteBIRD, CORE-M5, CMB-S4, and PICO with DESI, Euclid, and SKA. In the minimal $oldsymbol{ ext{ΛCDM}}+M_ u$ setup, several combinations yield high-significance detections of $M_ u$ around $0.06$ eV, and including a precise $ au_{ ext{reio}}$ prior from 21 cm measurements can boost sensitivity substantially; extending to $N_{ ext{eff}}$, $w_0$, and $w_a$ introduces degeneracies that reduce, but do not erase, the neutrino mass constraints, with $N_{ ext{eff}}$ constraints benefiting especially from CMB-S4 and Euclid. The paper emphasizes the powerful complementarity of CMB and LSS data, provides a framework for comparing future experiments on neutrino mass sensitivity, and discusses caveats related to foregrounds and non-linear modeling that could affect real-data outcomes.

Abstract

We forecast the sensitivity of thirty-five different combinations of future Cosmic Microwave Background and Large Scale Structure data sets to cosmological parameters and to the total neutrino mass. We work under conservative assumptions accounting for uncertainties in the modelling of systematics. In particular, for galaxy redshift surveys, we remove the information coming from non-linear scales. We use Bayesian parameter extraction from mock likelihoods to avoid Fisher matrix uncertainties. Our grid of results allows for a direct comparison between the sensitivity of different data sets. We find that future surveys will measure the neutrino mass with high significance and will not be substantially affected by potential parameter degeneracies between neutrino masses, the density of relativistic relics, and a possible time-varying equation of state of Dark Energy.

Figures (4)

• Figure 1: The figure shows the fiducial model (black) compared to the model plus noise spectrum based on the minimum variance estimator, $C_l + N_l$, for each experiment (Planck in green, LiteBIRD in grey, CMB-S4 in purple, CORE-M5 in red, and PICO in blue), for temperature anisotropies (left), E-mode polarisation (middle), and CMB lensing potential (right). CMB-S4 compares favourably to PICO, but it is important to note that the noise depicted here does not include cosmic variance or uncertainties due to imperfect foreground removal, where PICO would have a clear advantage over CMB-S4 due to full-sky observations and a much larger number of channels.
• Figure 2: Neutrino mass sensitivity for each CMB experiment, alone and in combination with DESI, Euclid, Euclid + SKA1 IM, Euclid + SKA1 IM + $\tau_{\rm reio}$ prior. Each subplot corresponds to one CMB setup (LiteBIRD, LiteBIRD + CMB-S4, CORE-M5, CORE-M5 + CMB-S4, Planck, or PICO from top left to bottom right, where the desaturated symbols indicate the CMB-S4 sensitivity) and relevant combinations with large-scale structure surveys (reported on the x-axis). For each combination the sensitivity is depicted for four cosmological models: the minimal scenario $\Lambda$CDM $+M_\nu$, and three extensions $+N_{\rm eff}$, $+w_0$, and $+w_0+w_a$. The horizontal dashed lines show the thresholds for a 1 to 5$\sigma$ detection of $M_\nu=0.06$ eV.
• Figure 3: Neutrino mass sensitivity for each CMB experiment, alone and in combination with DESI, Euclid, Euclid + SKA1 IM, Euclid + SKA1 IM + $\tau_{\rm reio}$ prior. Each subplot corresponds to one CMB setup plus large-scale structure survey combination (CMB only, CMB + DESI, CMB + Euclid, CMB + Euclid + SKA1 IM, CMB + Euclid + SKA1 IM + $\tau_{\rm reio}$ prior from top left to bottom right). The horizontal dashed lines show the thresholds for a 1 to 5$\sigma$ detection of $M_\nu=0.06$ eV.
• Figure 4: Neutrino mass vs $\tau_{\rm reio}$ for the three configurations CMB-S4 + Euclid, CMB-S4 + Euclid + LiteBIRD, CMB-S4 + Euclid + SKA1 IM + $\tau_{\rm reio}$ prior in the minimal 7 parameter $\Lambda$CDM+$M_{\nu}$ model.