Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A model-independent assessment of the late-time dark energy density evolution

Rayff de Souza, Agripino Sousa-Neto, Javier E. González, Jailson Alcaniz

TL;DR

The paper investigates whether late-time dark energy evolution is required by current distance data, arguing that standard EoS parameterizations can bias conclusions. It introduces a model-independent reconstruction of the dark energy density $X(z)$ using Gaussian Process regression, linking $X(z)$ to distance measures $D_M(z)$ and $D_H(z)$ derived from DESI DR2 BAO and SN Ia datasets (Pantheon+, Union3, DESY5) while incorporating Planck distance priors for $\Omega_m$. By comparing the GP-derived $X(z)$ with ΛCDM, a thawing quintessence, and CPL parameterizations, the study finds all models are compatible with the GP reconstruction at the 95% CL, with the strongest tension for ΛCDM in the DESY5 low-redshift regime but not statistically decisive. The work demonstrates a robust data-driven path to assess late-time DE behavior, reducing model dependence and highlighting the need for more precise future data to break residual degeneracies among viable dark energy scenarios.

Abstract

Combined measurements of Baryon Acoustic Oscillations (BAO) from the Dark Energy Spectroscopic Survey (DESI), the Cosmic Microwave Background (CMB) and Type Ia Supernovae (SN Ia), have recently challenged the $Λ$-Cold Dark Matter ($Λ$CDM) paradigm, indicating potential evidence for a dynamical dark energy component. These results are usually obtained in the context of the dark energy equation-of-state (EoS) parameterizations, generally implying in phantom-crossing at intermediate redshifts. However, a general mapping between these parameterizations that yields approximately the same background observables clouds the inference of the true nature of dark energy in the context of these parametric methods. In this work, we propose a model-independent reconstruction of the dark energy density, which is more directly constrained than its EoS, based on the Gaussian Process (GP) regression method with the use of DESI DR2 BAO data and the Pantheon+, Union3 and DESY5 SN Ia samples. In addition, we perform a statistical comparison between the energy densities of $Λ$, a non-phantom thawing quintessence-type dark energy, and the Chevallier-Polarski-Linder parameterization with the reconstructed function. We find that all models agree with the GP reconstruction at 95\% C.L., with the largest discrepancy coming from $Λ$CDM with DESY5 at low redshifts. Even in this case, our findings suggest that it may be premature to claim statistically significant evidence for evolving or phantom dark energy with current DESI and SN Ia measurements.

A model-independent assessment of the late-time dark energy density evolution

TL;DR

The paper investigates whether late-time dark energy evolution is required by current distance data, arguing that standard EoS parameterizations can bias conclusions. It introduces a model-independent reconstruction of the dark energy density using Gaussian Process regression, linking to distance measures and derived from DESI DR2 BAO and SN Ia datasets (Pantheon+, Union3, DESY5) while incorporating Planck distance priors for . By comparing the GP-derived with ΛCDM, a thawing quintessence, and CPL parameterizations, the study finds all models are compatible with the GP reconstruction at the 95% CL, with the strongest tension for ΛCDM in the DESY5 low-redshift regime but not statistically decisive. The work demonstrates a robust data-driven path to assess late-time DE behavior, reducing model dependence and highlighting the need for more precise future data to break residual degeneracies among viable dark energy scenarios.

Abstract

Combined measurements of Baryon Acoustic Oscillations (BAO) from the Dark Energy Spectroscopic Survey (DESI), the Cosmic Microwave Background (CMB) and Type Ia Supernovae (SN Ia), have recently challenged the -Cold Dark Matter (CDM) paradigm, indicating potential evidence for a dynamical dark energy component. These results are usually obtained in the context of the dark energy equation-of-state (EoS) parameterizations, generally implying in phantom-crossing at intermediate redshifts. However, a general mapping between these parameterizations that yields approximately the same background observables clouds the inference of the true nature of dark energy in the context of these parametric methods. In this work, we propose a model-independent reconstruction of the dark energy density, which is more directly constrained than its EoS, based on the Gaussian Process (GP) regression method with the use of DESI DR2 BAO data and the Pantheon+, Union3 and DESY5 SN Ia samples. In addition, we perform a statistical comparison between the energy densities of , a non-phantom thawing quintessence-type dark energy, and the Chevallier-Polarski-Linder parameterization with the reconstructed function. We find that all models agree with the GP reconstruction at 95\% C.L., with the largest discrepancy coming from CDM with DESY5 at low redshifts. Even in this case, our findings suggest that it may be premature to claim statistically significant evidence for evolving or phantom dark energy with current DESI and SN Ia measurements.

Paper Structure

This paper contains 13 sections, 22 equations, 6 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Dark energy density reconstruction
  3. Gaussian Processes
  4. Observational Data
  5. DESI BAO
  6. Type IA Supernovae
  7. Pantheon+
  8. Union3
  9. DESY5
  10. Cosmic Microwave Background
  11. Results
  12. Summary and Conclusions
  13. Matter density parameter from CMB distance priors

Figures (6)

  • Figure 1: Total equation of state of the Universe at low-redshifts, obtained via Gaussian Processes regression with DESI DR2 and Pantheon+ data.
  • Figure 2: Left: Solid curves show the mapping between the dark energy densities for several common parameterizations, a cosmological constant and the $w_T(z)$ thawing quintessence. The dashed curve and the shaded region display the best-fit energy density and the 95% C. L for CPL using the DESI BAO DR2+Planck 2018+DESY5 data combination, respectively. Right: The upper panel shows the predictions for $D_H(z)$ and $D_M(z)$ in units of $c/H_0$ for each of the solid curves depicted in the left plot, where all the curves, except for the $\Lambda$CDM prediction in black, are superimposed on top of each other. In the lower panel we highlight the relative difference between the models' distances and the best-fit prediction for the thawing quintessence parameterization. The grey shaded band denotes the region where the relative differences stay below the sub-percent threshold.
  • Figure 3: Gaussian Process reconstructions of $\tilde{D}_M$ (top) and $\tilde{D}^\prime_M (=\tilde{D}_H)$ (bottom) for Pantheon+, Union3, and DESY5 combined with DESI BAO.
  • Figure 4: GP dark energy density reconstruction and the statistical distribution of $X(z)$ for each parameterization. Each panel shows the results assuming the CMB+DESI DR2+SNe sample data combination. All shaded regions display the 68% C.L.
  • Figure 5: Distribution of the standard deviation of each model relative to the GP reconstruction as a function of redshift, for each SNe sample used in the data combination.
  • ...and 1 more figures