Willem Elbers, Carlos S. Frenk, Adrian Jenkins, Baojiu Li and Silvia Pascoli                          

PRD 2024, in press ; arXiv:2407.10965

One of the most exciting new results in studies of the large-scale structure of the Universe is the tentative finding of an inconsistency between the standard ΛCDM model of cosmology and the measurement of the baryon acoustic oscillations in the DESI survey of approximately 6 million galaxy, quasars and Lyman-a forest lines. Depending on which additional dataset are combined with DESI (CMB, supernovae, etc), the discrepancy with ΛCDM can be as large as 3.9σ. The data are, instead, consistent with a time-varying dark energy, for example, with a model in which the equation of state of dark energy is parametrized as w=w₀ + w_a (1-a),  where a is the expansion factor and w₀ and w_a are constants. This is referred to as the w₀w_aCDM model.

Another major DESI target is to set the most stringent astrophysical constraints on the sum of the masses of neutrinos, Σm_ν, which is possible because massive neutrinos affect both the shape of the power spectrum of density perturbations and the expansion history of the Universe, both of which can be measured with DESI. However, in the context of the ΛCDM model, the marginalized posterior distribution of Σm_ν peaks at the lower edge of the prior, Σm_ν =0 and resembles the tail of a distribution with a central value in the negative mass range. 

In this study, we introduced the concept of an “effective neutrino mass” that extends consistently to negative values. Within ΛCDM, the posterior distribution of this mass peaks at a negative value and, furthermore, is inconsistent with upper limits from neutrino oscillation measurements. By contrast, within w₀w_aCDM, the posterior peaks at a positive value, m_n,eff    = 0.06 (+0.15  -0.10) eV, and is perfectly consistent with laboratory data. (A similar behaviour is found for “mirage” dark energy models). These results provide further, independent evidence against ΛCDM and in favour w₀w_aCDM. Our analysis was underpinned by a set of very large-volume cosmological simulations of the formation of large-scale structure using a novel method we developed and previously introduced (Elbers et al 2021, MNRAS, 501, 2614), the δf method that accurately resolves neutrinos down to small scales and dramatically reduces the shot noise compared with the previous standard method.

Figure 1 A large cosmological hydrodynamics simulation, performed on COSMA-8, including massive neutrinos followed with the δf method. Each panel represents a different mass component in the simulation, as indicated, in a thin slab. The size of the computational cube is 2.8Gpc.