STRIDES: a 3.9 per cent measurement of the Hubble constant from the strong lens system DES J0408-5354

TL;DR

This study delivers a blind time-delay cosmography measurement of the Hubble constant from the strong lens DES J0408$-$5354, achieving $H_0 = 74.2^{+2.7}_{-3.0}$ km s$^{-1}$ Mpc$^{-1}$ (3.9% precision) within a flat $\Lambda$CDM framework. It employs a comprehensive lens modelling framework that jointly uses high-resolution HST imaging, measured time delays, deflector kinematics, and external convergence priors, while exploring 24 mass-model combinations and marginalizing over source/perturber configurations. The analysis accounts for mass-sheet degeneracy via kinematic probes and multilens-plane treatment of line-of-sight perturbers, producing a two-cosmological-distance posterior $\{D_{\Delta t}^{\rm eff}, D_d\}$ with their covariance. The result is consistent with previous H0LiCOW measurements and local distance-ladder estimates, reinforcing the existing tension with early-Universe inferences and underscoring the value of independent, high-precision lens-based $H_0$ constraints for cosmology. The work also emphasizes methodological robustness via independent blind analyses and outlines avenues to improve precision with larger lens samples and more flexible mass models in future studies.

Abstract

We present a blind time-delay cosmographic analysis for the lens system DES J0408$-$5354. This system is extraordinary for the presence of two sets of multiple images at different redshifts, which provide the opportunity to obtain more information at the cost of increased modelling complexity with respect to previously analyzed systems. We perform detailed modelling of the mass distribution for this lens system using three band Hubble Space Telescope imaging. We combine the measured time delays, line-of-sight central velocity dispersion of the deflector, and statistically constrained external convergence with our lens models to estimate two cosmological distances. We measure the "effective" time-delay distance corresponding to the redshifts of the deflector and the lensed quasar $D_{Δt}^{\rm eff}=3382^{+146}_{-115}$ Mpc and the angular diameter distance to the deflector $D_{\rm d}=1711^{+376}_{-280}$ Mpc, with covariance between the two distances. From these constraints on the cosmological distances, we infer the Hubble constant $H_0 = 74.2^{+2.7}_{-3.0}$ km s$^{-1}$ Mpc$^{-1}$ assuming a flat $Λ$CDM cosmology and a uniform prior for $Ω_{\rm m}$ as $Ω_{\rm m} \sim \mathcal{U}(0.05, 0.5)$. This measurement gives the most precise constraint on $H_0$ to date from a single lens. Our measurement is consistent with that obtained from the previous sample of six lenses analyzed by the $H_0$ Lenses in COSMOGRAIL's Wellspring (H0LiCOW) collaboration. It is also consistent with measurements of $H_0$ based on the local distance ladder, reinforcing the tension with the inference from early Universe probes, for example, with 2.2$σ$ discrepancy from the cosmic microwave background measurement.

Figures (17)

• Figure 1: RGB color composite of the lens systems DES J0408$-$5354. The three HST filters used to create the RGB image are F160W (red), F814W (green), and F475X (blue). The relative amplitudes between the three filters are adjusted in this figure for better visualization by achieving a higher contrast. We label different components of the lens system. G1 is the main deflector galaxy and G2 is its satellite galaxy. In addition to the lensed arcs from the extended quasar host galaxy, this lens system has extra source components S2 and S3. The source component S2 is doubly imaged and forms an extended arc inside the Einstein radius. S3 forms another fainter extended arc on the North-East outside the Einstein radius without a noticeable counterimage. Four nearby perturbers G3--G6 along the line of sight are marked with the dashed, white circles.
• Figure 2: Impact of varying $\Omega_{\rm m}$ in the $\Lambda\text{CDM}$ cosmology on the angular diameter distance ratios between the lens and source planes. All the distance ratios except for the black line changes less than 1 per cent for a wide range of $\Omega_{\rm m}$. The black line corresponds to the distance ratio involving S2's lens plane. As the S2's Einstein radius is small ($\sim$0.002 arcsec), the change in the black line only has a small effect on the effective Fermat potential [cf. equation (\ref{['eq:double_plane_fermat_potential']})]. Therefore, fixing the distance ratios for the fiducial cosmology with $\Omega_{\rm m} = 0.3$ is not a strong assumption in our analysis. See Appendix \ref{['app:effect_cosmology']} for tests validating this point.
• Figure 3: Observed and estimated properties of the line-of-sight galaxies G3--G6. Top left: velocity dispersions derived from the MUSE integral field spectra. Top right: SIS Einstein radius distributions obtained from the observed velocity dispersions. Middle left: estimated stellar masses from BuckleyGeer20. Middle right: halo mass $M_{200}$ inferred from the estimated stellar mass using the stellar mass--halo mass relation from Behroozi19. Bottom left: halo concentration parameter $c_{200}$ obtained using a halo mass--concentration relation for our fiducial cosmology Diemer19. Bottom right: scale radius of the NFW profile in angular unit for our fiducial cosmology from the $M_{200}$ and $c_{200}$ priors. The intrinsic scatter and uncertainties of the adopted scaling relations are accounted for at each conversion step. We use the SIS Einstein radius distributions as priors for the SIS model and the $M_{200}$ and $c_{200}$ distributions as priors for the NFW model for G4--G6.
• Figure 4: The most likely lens model and reconstructed image of DES J0408$-$5354 using the power-law model. The top row shows the observed RGB image, reconstructed RGB image, the convergence profile, and the magnification model in order from the left-hand side to the right-hand side. The next three rows show the observed image, the reconstructed image, the residual, the reconstructed source in order from the left-hand side to the right-hand side for each of the HST filters. The three filters are F160W (second row), F814W (third row), and F475X (fourth row). All the scale bars in each plot correspond to 1 arcsec. The patchy or ring-like artefacts in the source reconstruction translate to lensed features below the noise level in the image, thus they do not affect our lens model.
• Figure 5: The most likely lens model and reconstructed image of DES J0408$-$5354 using the composite model. The top row shows the observed RGB image, reconstructed RGB image, the convergence profile, and the magnification model in order from the left-hand side to the right-hand side. The next three rows show the observed image, the reconstructed image, the residual, the reconstructed source in order from the left-hand side to the right-hand side for each of the HST filters. The three filters are F160W (second row), F814W (third row), and F475X (fourth row). All the scale bars in each plot correspond to 1 arcsec. The patchy or ring-like artefacts in the source reconstruction translate to lensed features below the noise level in the image, thus they do not affect our lens model.