H0LiCOW - IX. Cosmographic analysis of the doubly imaged quasar SDSS 1206+4332 and a new measurement of the Hubble constant

TL;DR

This work advances time-delay strong lensing (TDSL) cosmography by performing a blind, self-consistent analysis of the doubly imaged quasar SDSS 1206+4332 using time delays, HST imaging, deflector kinematics, and line-of-sight data. By exploring a wide suite of lens and environment models within a Bayesian hierarchical framework and using an independent lensing code (lenstronomy), the authors derive a robust two-distance constraint and infer H_0 = 68.8^{+5.4}_{-5.1} km s^{-1} Mpc^{-1} for SDSS 1206+4332, which sharpens to H_0 = 72.5^{+2.1}_{-2.3} km s^{-1} Mpc^{-1} when combined with three previous H0LiCOW lenses. The analysis demonstrates that doubles can yield competitive cosmographic precision, mitigates lens-model systematics via multiple model choices and BIC-based weighting, and provides a publicly available likelihood to enable joint cosmographic analyses. The results independently corroborate the tension between local and CMB-based $H_0$ in certain cosmologies and underscore the value of diverse lens configurations for expanding the time-delay cosmography sample in coming surveys.

Abstract

We present a blind time-delay strong lensing (TDSL) cosmographic analysis of the doubly imaged quasar SDSS 1206+4332. We combine the relative time delay between the quasar images, Hubble Space Telescope imaging, the Keck stellar velocity dispersion of the lensing galaxy, and wide-field photometric and spectroscopic data of the field to constrain two angular diameter distance relations. The combined analysis is performed by forward modelling the individual data sets through a Bayesian hierarchical framework, and it is kept blind until the very end to prevent experimenter bias. After unblinding, the inferred distances imply a Hubble constant $H_0 = 68.8^{+5.4}_{-5.1}$ kms$^{-1}$Mpc$^{-1}$, assuming a flat Lambda cold dark matter cosmology with uniform prior on $Ω_{\rm m}$ in [0.05, 0.5]. The precision of our cosmographic measurement with the doubly imaged quasar SDSS 1206+4332 is comparable with those of quadruply imaged quasars and opens the path to perform on selected doubles the same analysis as anticipated for quads. Our analysis is based on a completely independent lensing code than our previous three H0LiCOW systems and the new measurement is fully consistent with those. We provide the analysis scripts paired with the publicly available software to facilitate independent analysis. The consistency between blind measurements with independent codes provides an important sanity check on lens modelling systematics. By combining the likelihoods of the four systems under the same prior, we obtain $H_0 = 72.5^{+2.1}_{-2.3}$kms$^{-1}$Mpc$^{-1}$. This measurement is independent of the distance ladder and other cosmological probes.

Figures (17)

• Figure 1: Drizzled HST-WFC3 image through filter F160W of the lens SDSS 1206+4332 . The doubly lensed quasar is embedded in a source galaxy, parts of which are quadruply lensed in a fold configuration. We label the different galaxies that we explicitly model. Prominently visible is a galaxy triplet in direction N-W, G2, and two other less massive nearby galaxies in direction E and N-E, G3 and G4.
• Figure 2: Example of a lens model drawn from the posterior sample and its ability to reconstruct the HST image. Upper left: Reduced HST image data. Upper middle: Reconstructed image within the chosen mask region. Upper right: Normalized residuals of the model compared to the data based on the noise map. Lower left: Reconstructed source from the model with a double Sersic profile and $n_{\text{max}}=8$ shapelet coefficients. Lower middle: Convergence of the lens model. Lower right: Magnification of the lens model and indicated image positions A and B as well as the intrinsic source position of the quasar (marked as a star).
• Figure 3: The same model as presented in Figure \ref{['fig:lens_model']} decomposed in its individual components. Upper panels: Model components without the instrumental convolution applied. Lower panels: Model components with the PSF convolution applied. Left: Lens light component as modelled by a double Sersic profile for G0 and a spherical Sersic profile for G1. Middle: Lensed extended source light, modelled with a double Sersic profile and $n_{\text{max}}=8$ shapelet coefficients. Right: Lensed source and lens light components combined. The lower panel also includes the components of the point sources.
• Figure 4: Measured time delay between images A and B of SDSS 1206+4332 from the data set of Eulaers:2013. Indicated are the mean and 1-$\sigma$ errors of the original analysis by Eulaers:2013 and two different updated re-analysis methods. We chose the equal weight marginalized measurement for this work.
• Figure 5: Microlensing time-delay maps and statistical distribution for the two quasar images A and B. The maps (right panel) are based on the magnification tensor of the lens model (Table \ref{['table:images']}), an estimate of the stellar initial mass function (IMF) and the normalization estimated from the stellar contribution to the lensing mass, and accretion disc properties summarized in Table \ref{['table:micro_lensling_specs']}. The distributions of the expected microlensing time-delay of the two images are shown on the left panels. The microlensing time delay is small compared to the measurement uncertainties of the relative time delay between the two images.