Evidence for evolving Dark Energy from a new cosmic probe

Abstract

The $Λ$CDM concordance cosmological model provides a remarkably successful description of the formation and evolution of structure in the Universe. However, a growing discrepancy between measurements of the expansion rate $H_0$ from the near and distant Universe now appears to be significant at the ~4-7 $σ$ level. This inconsistency, known as the ``Hubble tension'', has arisen either due to unrecognized systematics in these measurements or new physics beyond the standard model, such as an evolving dark energy equation of state. Modeling ~20-year, multi-band optical light curves for 6992 active galactic nuclei (AGN), we find a tight relation linking the variability amplitude and characteristic timescale to their intrinsic luminosity. This empirical law enables us to construct an AGN-based Hubble diagram to z ~3.5. Joint inference with supernova distances reveals evidence for an evolving dark energy equation of state at the 3.8-3.9 $σ$ level over constant w models and 4.4-4.8 $σ$ over $Λ$CDM. Our results establish AGN light curves as a powerful tool for cosmography that could offer a novel pathway to test deviations from the standard $Λ$CDM expansion history.

Figures (16)

• Figure 1: Upper panel: An illustrative example of an AGN light curve from our sample and the fitted multi-band model. The light curve in each band is offset by a scale factor for visual clarity. Lower panel: Corresponding multi-band power spectral density (PSD) in angular frequencies from the model (magenta) and computed directly from the data using the Lomb-Scargle algorithm (black points; Lomb1976Scargle1982). The solid gray line is the noise floor level. The magenta error bar area is the 1$\sigma$ range of the angular frequency corresponding to the UV damping timescale.
• Figure 2: Empirical relation between the bias-corrected rest-frame 2500 Å monochromatic AGN luminosity, $L_{2500\,\text{\AA}}$, and a power-law combining the variability amplitude and timescale parameters. The best-fit model is shown in magenta along with the $1\sigma$ uncertainty band. The filled circles are AGN where rest-frame $2500$ Å is within the wavelength coverage of the spectrum. The open circles denote sources where rest-frame $2500$ Å is beyond the wavelength coverage of the spectrum, requiring an extrapolation of the AGN continuum. Open circles are not fitted, but are included in the figure for completeness. The slope of the best-fit relation with uncertainties from Suberlak2021 is shown as the cyan dashed line. The gray curves are 1 and 2$\sigma$ density contours.
• Figure 3: Bias-corrected Hubble diagram for Pantheon+ SN Ia sample (blue circle symbols) and our plotted AGN sample (black circle symbols; this work). The filled circles correspond to our fiducial fitting AGN sample, where rest-frame $2500$ Å is within the wavelength coverage of the spectrum. The open circles are sources where rest-frame $2500$ Å is beyond the wavelength coverage of the spectrum, requiring an extrapolation of the AGN continuum. Open circles are not fitted, but included on the figure for completeness. The flat $w_0w_a$CDM model fitted to the combined SN Ia + fiducial AGN sample is shown in magenta. The concordance $\Lambda$CDM and flat $w$CDM models are shown for comparison as the blue dotted and green dash–dotted lines, respectively. The Planck 2018 result (PlanckCollaboration2020; CMB power spectra, CMB lensing reconstruction, and BAO) is shown as the dashed yellow line, highlighting the Hubble tension. We overplot the AGN binned means and standard error on the means in red to guide the eye. The binned points are not used in the fit. The inset panel shows the Hubble diagram with redshift plotted using a logarithmic scale. The lower panel shows the binned residuals and fitted models relative to the best-fit flat $w$CDM model.
• Figure 4: Posterior distribution for flat $w_0w_a$CDM cosmological parameters for the Pantheon+ SN Ia Brout2022 (shown in light blue) and combined SN Ia + AGN fiducial fitting sample (this work, shown in gray). The posterior distributions for each parameter are shown along the diagonal. The text in the upper right gives the posterior medians and 16th and 84th percentiles (1$\sigma$ credible intervals). The 2D contour plots show the covariances between each pair of parameters. The darker and lighter shaded regions are the 1$\sigma$ and 2$\sigma$ contours, and dashed curves delimit the 3$\sigma$ contours.
• Figure 5: Same as Figure \ref{['fig:cornerw0waCDM']} but for the flat $w$CDM model where $w$ is not fixed to -1.