Periodic Reporting for period 4 - COSMICLENS (Cosmology with Strong Gravitational Lensing)
Reporting period: 2023-04-01 to 2023-09-30
There is a growing discrepancy between measurements of the Hubble constant, H0, performed with cosmological probes in the early Universe with Cosmic Microwave Background (CMB) and probes in the late universe such as pulsating stars and supernovae. Deciding on whether this discrepancy, called the Hubble tension, is real or not has profound implications for the current cosmological paradigm and can very well imply that new physics beyond the standard model must be considered. However, all cosmological probes currently in use do rely on each other at some level and we lack of truly independent verification. Strong lensing time delays, the topic of COSMICLENS, consists in the only other independent probe that can address the Hubble tension. COSMICLENS implements the whole chain of analysis needed to successfully measure and use time delays in strongly lensed to measure H0 precisely enough. The principle of the method is simple: time delays can be measured using the photometric variations of the lensed images in multiply imaged quasars. As the optical paths to each image are different in length, and as they intersect the lens plane at different impact parameters, the wavefronts along each path reach the observer at different times. As a consequence, the photometric variations in each lensed image of the quasar are seen shifted in time. This time delay, due to both geometry and gravity is related to H0 with only little sensitivity to other cosmological parameters. More precisely, the geometrical part of the time delay contains H0 and can be constrained with a mass model for the lens galaxy(ies). H0 depends on the slope of the lensing potential at the position of the lensing images. This means that models should be as accurate as possible as to turn time delays into reliable values of H0, which is the second goal of COSMICLENS, by designing new modeling tools where the lens and source are described on a basis of wavelets rather than in an analytical way, as done traditionally. A third part of the project is to understand what could possibly go wrong with modeling and to confront the standard and new modeling tools to detailed simulations of lenses and therefore to provide crash-tests and benchmarking for our tools. This uses analytical and well-controlled simulations of lenses. Finally, one has to find as many lenses as possible to measure new time delays. Lenses are rare and not all can be used for time delay cosmography. We have therefore massively exploited public sky surveys to find new objects. New techniques have been devised for this work, which will also benefit lens searches in upcoming space-based and ground-based all sky surveys such as Euclid and Rubin-LSST.
COSMICLENS is organized around 4 main sub-projects that are the 4 vital ingredients to carry out cosmography with strong lensing time delays: 1- excellent time delay measurements, 2- excellent models for the lens and its environment, 3- excellent control of the systematic errors, 4- lensed quasars in sufficient numbers. COSMICLENS has been successful on all fronts. Telescope time has been available in large part thanks to ERC funding to measure time delays with 3 telescopes in both hemispheres. About 40 new time delay measurements have been done and are useful for cosmography and H0 measurements. This triples the number of time delays known before the project, hence providing long-lasting measurements from high-cadence and high-SNR light curves that are unlikely to be surpassed by Rubin-LSST. We have also measured delays in doubly images quasars, not just quadruples, and opened a new direction of research with doubles, which constitute 80% of the population of lensed quasars. Excellent HST and JWST data have been obtained for some of these lenses, which highlights the importance of the field of time delays among the community. In addition, we have developed a full lens modelling toolbox that we made public and documented, along with a proposal of a new standard for lens modelling that will ensure better cross-talks between different codes. Our lens search has produced 130 new lensed quasars, some of which are being monitored for time delay measurements. This doubles the number of known lensed quasars. The search was particularly efficient from the Gaia survey, meaning that COSMICLENS made a significant contribution to the Gaia mission itself, even though this was not planned at the beginning of the project. Also not planned originally was the development of a new public and interactive database of lensed objects, now containing over 15’000 entries (including lensed galaxies and lensed quasars), and featuring most of the literature available for each entry. The novelty of this database is that it is flexible and interactive, so that it can be updates by any registered users. This database is now an accepted in-kind contribution of EPFL to the Rubin-LSST survey.
COSMICLENS has allowed to establish lensing time delays as a reliable field and has contributed to build the TDCOSMO collaboration, now gathering most of the key-players in this field and gathering all required expertise to turn time delays into cosmology. During the course of COSMICLENS, criticisms were made in the literature about the degeneracies inherent to lens models, that may prevent from reaching a few percent accuracy on H0. One family of potential systematic errors could arise from deviation of the shape of the lensing galaxy from a perfect ellipsoid. We have shown that the uncertainty introduced by this effect is random and already accounted for in the uncertainty budget of TDCOSMO analyses. The other degeneracy is related to the uncertainty on the radial distribution of the lensing galaxy. To break that degeneracy, TDCOSMO and COSMICLENS proposed two different mitigation strategies that combine time delays and kinematics of lens galaxies. The first approach is a hierarchical Bayesian analysis that combines dynamics of lenses with no time delays to lensing constraint from systems that have measured time delays from COSMICLENS and coarse kinematic information. The second approach relies on combining lens modeling with direct spatially-resolved kinematics of lens galaxies that have time delays. In the latter case the kinematics is obtained from approved time on JWST, or ground-based data benefiting of adaptive optics. The current best value of H0 from COSMICLENS+TDCOSMO strengthens the current Hubble tension but upcoming observations and analyses will establish this on a firmer ground. In this case, the current cosmological paradigm will have to be revised.