Torsion cosmology in the light of DESI, supernovae and CMB observational constraints

TL;DR

This work tests a torsion-based cosmology within the Einstein-Cartan framework against state-of-the-art cosmological probes, combining DESI BAO, PantheonPlus and DESY5 SN Ia, and Planck 2018 CMB with lensing. The model introduces a torsion parameter $\alpha$ via $\phi(t) = -\tfrac{1}{2}\alpha H$, leading to a modified Friedmann equation $H^2(z) = \frac{H_0^2}{(1-\alpha)^2}[\Omega_m(1+z)^{3-\alpha}+\Omega_\Lambda]$ and reduces to ΛCDM as $\alpha \to 0$. Bayesian analysis with CLASS/Cobaya and Akaike Information Criterion reveals α = -0.00066 ± 0.00098 (full data), $H_0 ≈ 68.41$ km s$^{-1}$ Mpc$^{-1}$, and $S_8 ≈ 0.812$, with negative $\Delta$AIC values (−6.62 for the full dataset) signaling a statistically preferred fit over ΛCDM and a substantial reduction of the $S_8$ tension to ~0.1σ. The results suggest that torsion can simultaneously address key cosmological tensions while maintaining excellent agreement with diverse probes, though α remains consistent with zero within 1σ; future surveys will further test the predicted growth-modulation term $\alpha H$.

Abstract

In this work, we investigate a torsion-based cosmological model within the Einstei-Cartan framework, constrained by the latest combined datasets including DESI DR2 BAO, PantheonPlus and DESY5 supernovae, and the full Planck 2018 CMB measurements (temperature, polarization, and joint NPIPE PR4 + ACT DR6 lensing). The torsion parameter is constrained to $α= -0.00066 \pm 0.00098$ with the full dataset combination, consistent with zero at less than $1σ$, while yielding a Hubble constant $H_0 = 68.41 \pm 0.32$ km/s/Mpc and matter clustering amplitude $S_8 = 0.812 \pm 0.006$. The model shows notable potential in alleviating cosmological tensions, reducing the $S_8$ discrepancy with KiDS-1000 from $\sim 2.3σ$ in $Λ$CDM to only $0.1σ$. Model comparisons based on the Akaike information criterion show consistent improvements across all datasets, with $Δ{\rm AIC}$ values ranging from $-5.68$ to $-6.62$, indicating a statistically preferred fit for the torsion model. These results suggest that the torsion framework provides a physically well-motivated extension to $Λ$CDM, capable of simultaneously addressing key cosmological tensions while maintaining excellent agreement with diverse observational probes.

Figures (2)

• Figure 1: The one and two-dimensional marginalised posterior distributions in a $\Lambda$CDM cosmology from the different datasets.
• Figure 2: The one and two-dimensional marginalised posterior distributions in a torsion cosmology from the different datasets.