Measuring Cosmic Neutrino Masses Independently of Dark Energy

Technical Summary: Measuring Cosmic Neutrino Masses Independently of Dark Energy

Problem Statement
Cosmological data within the standard $\Lambda$CDM framework currently provide the tightest constraints on the sum of neutrino masses, $\sum m_\nu$, yielding $\sum m_\nu < 0.056$ eV (95% CL) when combining Planck CMB, CMB lensing, and DESI DR2 BAO. This result is in $2\text{--}3\sigma$ tension with the inverted-ordering floor of $0.10$ eV derived from neutrino oscillation experiments. However, this tight bound relies heavily on the assumption of a cosmological constant ($\Lambda$). Current data mildly prefer an evolving dark energy equation of state, and relaxing the dark energy model to a dynamical form (e.g., $w_0, w_a$) significantly degrades the neutrino mass constraint. The central problem addressed is the model dependence of these bounds: it is unclear how much of the tension with oscillation data arises from neutrino physics versus the assumption of a specific late-time expansion history.

Methodology
The authors deconstruct the sensitivity of various cosmological probes (primary CMB, CMB lensing, and BAO) to late-time physics to identify routes that are robust to dark energy variations. They define a robustness metric $R \equiv \sigma(\sum m_\nu)_{DE} / \sigma(\sum m_\nu)_{\Lambda}$, where $R \approx 1$ indicates independence from the dark energy model.

The study employs two distinct analytical routes:

1. Dark-Energy-Marginalized Route: Retains all data (CMB, lensing, BAO) and marginalizes over a parametrized dark energy history, typically $w(a) = w_0 + w_a(1-a)$. The authors test the stability of this bound against increasingly flexible parametrizations (constant $w$, binned $w(a)$, and cubic $w(a)$).
2. Late-Universe-Free Route: Constructs a combination of probes that, by design, removes dependence on the late-time expansion history ($z < z_{DE} \approx 0.7$). This involves:
   • Using the unlensed primary CMB by marginalizing over the lensing amplitude $A_{\text{lens}}$ to remove the two-point lensing channel, which is sensitive to late-time distances.
   • Utilizing the four-point lensing reconstruction ($C^{\kappa\kappa}_L$), which measures the matter power spectrum directly and is shown to be insensitive to low-redshift expansion.
   • Excluding absolute BAO distances ($D_M, D_H$) which encode the full expansion history, or using differenced distances ($\Delta D_M$) that remove the low-redshift integral, though the latter adds negligible constraining power beyond the lensing baseline.

The analysis utilizes Planck PR4/PR3, ACT DR6, SPT3G, and DESI DR2 data, alongside mock datasets generated with a fiducial $\Lambda$CDM cosmology to isolate intrinsic probe sensitivities from statistical fluctuations.

Key Contributions and Results

• Probe Sensitivity Decomposition:

  • Primary CMB: The unlensed primary CMB alone provides a weak bound ($\sum m_\nu < 0.78$ eV) but is highly robust ($R \approx 1.02$) to dark energy variations. The sensitivity to dark energy in standard CMB analyses arises primarily from the lensing smoothing of acoustic peaks.
  • CMB Lensing: The two-point lensing channel (smoothing of peaks) is sensitive to dark energy ($R \approx 1.22$). In contrast, the four-point lensing channel (reconstructed $C^{\kappa\kappa}_L$) is insensitive to dark energy ($R \approx 0.98$) because its Fisher information peaks at high multipoles ($L \approx 300$) and high source redshifts ($z \approx 2.2$), where dark energy is negligible.
  • BAO: BAO distances constrain $\sum m_\nu$ primarily through the determination of the physical matter density $\omega_m$. This determination relies on comparing distances across the matter- and dark-energy-dominated eras. Consequently, any BAO information that tightens the bound beyond the four-point lensing floor reintroduces dark energy sensitivity. Differencing BAO distances to remove the low-redshift integral restores robustness ($R \approx 1$) but eliminates the constraining power on $\omega_m$ provided by the absolute scale.

• Route I: Dark-Energy-Marginalized Bound:

  • Marginalizing over $(w_0, w_a)$ yields $\sum m_\nu < 0.152$ eV (95% CL) on real data.
  • Crucially, the authors demonstrate that this bound saturates near 0.152 eV even when moving to more flexible models (binned and cubic $w(a)$). The data resolve 3–4 independent modes of $w(a)$, but these additional modes are nearly orthogonal to the $\omega_m$ direction constrained by the CMB. Thus, $(w_0, w_a)$ marginalization is sufficient for smooth, matter-conserving dark energy models.

• Route II: Late-Universe-Free Bound:

  • Combining the unlensed primary CMB (with $A_{\text{lens}}$ marginalized) and the reconstructed four-point lensing spectrum yields a bound of $\sum m_\nu < 0.41$ eV (95% CL) on real data.
  • This bound is independent of the dark energy parametrization, remaining stable at $0.40\text{--}0.43$ eV across $\Lambda$CDM, $w_0w_a$CDM, and binned/cubic models ($R \approx 1$).
  • Unlike the standard bound, which tightens from mock to real data due to CMB-BAO tension, the late-universe-free bound remains stable between mock and real data, confirming it does not rely on the disputed low-redshift distance information.

• Future Projections:

  • With future data (Simons Observatory lensing and Spec-S5 BAO), the marginalized route tightens to $\sigma(\sum m_\nu) \approx 43$ meV ($\sum m_\nu < 0.153$ eV), while the late-universe-free route improves to $\sigma(\sum m_\nu) \approx 82$ meV ($\sum m_\nu < 0.28$ eV) at the cosmic-variance limit.
  • The authors also analyze future LSST + Simons Observatory cross-correlations (6 $\times$ 2-point), finding that including CMB lensing cross-correlations reduces the dark energy degradation ($R \approx 1.37$) compared to cosmic shear alone ($R \approx 1.52$).

Significance and Claims
The paper claims to provide a rigorous quantification of the model dependence of current neutrino mass constraints. It establishes that the tension between cosmological bounds and the inverted-ordering floor is largely driven by the assumption of a cosmological constant. By constructing a "late-Universe-free" bound, the authors demonstrate that current data independently support $\sum m_\nu < 0.41$ eV without assuming a specific dark energy model.

This relaxed bound ($\sim 0.41$ eV) places the neutrino mass scale within the sensitivity range of next-generation laboratory experiments like Project 8 (targeting $m_{\nu_e} \sim 0.1$ eV), whereas current terrestrial experiments like KATRIN ($m_{\nu_e} < 0.45$ eV) are not expected to detect a signal unless the mass is near the upper limit of the cosmological bound. The work motivates vital synergies between future cosmological surveys and terrestrial neutrino measurements to independently verify the neutrino mass hierarchy and absolute mass scale, free from the degeneracies of late-time cosmology. The authors emphasize that while the late-universe-free bound is weaker by a factor of $\sim 2.7$ in $\sigma$ compared to the marginalized bound, it trades statistical power for model independence, serving as a crucial consistency check against potential systematic errors or new physics in the dark sector.