Abstract
We present
GIGA-Lens
: a gradient-informed, GPU-accelerated Bayesian framework for modeling strong gravitational lensing systems, implemented in TensorFlow and JAX. The three components, ...optimization using multistart gradient descent, posterior covariance estimation with variational inference, and sampling via Hamiltonian Monte Carlo, all take advantage of gradient information through automatic differentiation and massive parallelization on graphics processing units (GPUs). We test our pipeline on a large set of simulated systems and demonstrate in detail its high level of performance. The average time to model a single system on four Nvidia A100 GPUs is 105 s. The robustness, speed, and scalability offered by this framework make it possible to model the large number of strong lenses found in current surveys and present a very promising prospect for the modeling of
(
10
5
)
lensing systems expected to be discovered in the era of the Vera C. Rubin Observatory, Euclid, and the Nancy Grace Roman Space Telescope.
Exploiting the powerful tool of strong gravitational lensing by galaxy clusters to study the highest-redshift Universe and cluster mass distributions relies on precise lens mass modelling. In this ...work, we aim to present the first attempt at modelling line-of-sight (LOS) mass distribution in addition to that of the cluster, extending previous modelling techniques that assume mass distributions to be on a single lens plane. We have focussed on the Hubble Frontier Field cluster MACS J0416.1–2403, and our multi-plane model reproduces the observed image positions with a rms offset of ~0.′′53. Starting from this best-fitting model, we simulated a mock cluster that resembles MACS J0416.1–2403 in order to explore the effects of LOS structures on cluster mass modelling. By systematically analysing the mock cluster under different model assumptions, we find that neglecting the lensing environment has a significant impact on the reconstruction of image positions (rms ~0.′′3); accounting for LOS galaxies as if they were at the cluster redshift can partially reduce this offset. Moreover, foreground galaxies are more important to include into the model than the background ones. While the magnification factor of the lensed multiple images are recovered within ~10% for ~95% of them, those ~5% that lie near critical curves can be significantly affected by the exclusion of the lensing environment in the models. In addition, LOS galaxies cannot explain the apparent discrepancy in the properties of massive sub-halos between MACS J0416.1–2403 and N-body simulated clusters. Since our model of MACS J0416.1–2403 with LOS galaxies only reduced modestly the rms offset in the image positions, we conclude that additional complexities would be needed in future models of MACS J0416.1–2403.
Full text
Available for:
FMFMET, NUK, UL, UM, UPUK
The dynamical modelling of integral field unit (IFU) stellar kinematics is a powerful tool to unveil the dynamical structure and mass build-up of galaxies in the local Universe, while gravitational ...lensing is nature’s cosmic telescope to explore the properties of galaxies beyond the local Universe. We present a new approach, which unifies dynamical modelling of galaxies with the magnification power of strong gravitational lensing, to reconstruct the structural and dynamical properties of high-redshift galaxies. By means of axisymmetric Jeans modelling, we create a dynamical model of the source galaxy, assuming a surface brightness and surface mass density profile. We then predict how the source’s surface brightness and kinematics would look when lensed by the foreground mass distribution and compare with the mock observed arcs of strong gravitational lensing systems. For demonstration purposes, we created and also analysed mock data of the strong lensing system RX J1131−1231. By modelling both the lens and source, we recover the dynamical mass within the effective radius of strongly lensed high-redshift sources within 5% uncertainty, and we improve the constraints on the lens mass parameters by up to 50%. This machinery is particularly well-suited for future observations from large segmented-mirror telescopes, such as the
James Webb
Space Telescope, which will yield high sensitivity and angular-resolution IFU data for studies on distant and faint galaxies.
Full text
Available for:
FMFMET, NUK, UL, UM, UPUK
Under the assumption of a flat LambdaCDM cosmology, recent data from the Planck satellite point toward a Hubble constant that is in tension with that measured by gravitational lens time delays and by ...the local distance ladder. Prosaically, this difference could arise from unknown systematic uncertainties in some of the measurements. More interestingly-if systematics were ruled out-resolving the tension would require a departure from the flat LambdaCDM cosmology, introducing, for example, a modest amount of spatial curvature, or a non-trivial dark energy equation of state. To begin to address these issues, we present an analysis of the gravitational lens RXJ1131-1231 that is improved in one particular regard: we examine the issue of systematic error introduced by an assumed lens model density profile. We use more flexible gravitational lens models with baryonic and dark matter components, and find that the exquisite Hubble Space Telescope image with thousands of intensity pixels in the Einstein ring and the stellar velocity dispersion of the lens contain sufficient information to constrain these more flexible models. The total uncertainty on the time-delay distance is 6.6% for a single system. We proceed to combine our improved time-delay distance measurement with the WMAP9 and Planck posteriors. In an open LambdaCDM model, the data for RXJ1131-1231 in combination with Planck favor a flat universe with Omega sub(k) = (ProQuest: Formulae and/or non-USASCII text omitted) (68% credible interval (CI)). In a flat wCDM model, the combination of RXJ1131-1231 and Planck yields w = (ProQuest: Formulae and/or non-USASCII text omitted) (68% CI).
We present a detailed analysis of the inner mass structure of the Cosmic Horseshoe (J1148+1930) strong gravitational lens system observed with the Hubble Space Telescope (HST) Wide Field Camera 3 ...(WFC3). In addition to the spectacular Einstein ring, this systems shows a radial arc. We obtained the redshift of the radial arc counterimage zs, r = 1.961 ± 0.001 from Gemini observations. To disentangle the dark and luminous matter, we considered three different profiles for the dark matter (DM) distribution: a power law profile, the Navarro, Frenk, and White (NFW) profile, and a generalized version of the NFW profile. For the luminous matter distribution, we based the model on the observed light distribution that is fitted with three components: a point mass for the central light component resembling an active galactic nucleus, and the remaining two extended light components scaled by a constant mass-to-light ratio (M/L). To constrain the model further, we included published velocity dispersion measurements of the lens galaxy and performed a self-consistent lensing and axisymmetric Jeans dynamical modeling. Our model fits well to the observations including the radial arc, independent of the DM profile. Depending on the DM profile, we get a DM fraction between 60% and 70%. With our composite mass model we find that the radial arc helps to constrain the inner DM distribution of the Cosmic Horseshoe independently of the DM profile.
Full text
Available for:
FMFMET, NUK, UL, UM, UPUK
Abstract Supernova (SN) SN H0pe is a gravitationally lensed, triply imaged, Type Ia SN (SN Ia) discovered in James Webb Space Telescope imaging of the PLCK G165.7+67.0 cluster of galaxies. ...Well-observed multiply imaged SNe provide a rare opportunity to constrain the Hubble constant ( H 0 ), by measuring the relative time delay between the images and modeling the foreground mass distribution. SN H0pe is located at z = 1.783 and is the first SN Ia with sufficient light-curve sampling and long enough time delays for an H 0 inference. Here we present photometric time-delay measurements and SN properties of SN H0pe. Using JWST/NIRCam photometry, we measure time delays of Δ t ab = − 116.6 − 9.3 + 10.8 observer-frame days and Δ t cb = − 48.6 − 4.0 + 3.6 observer-frame days relative to the last image to arrive (image 2b; all uncertainties are 1 σ ), which corresponds to a ∼5.6% uncertainty contribution for H 0 assuming 70 km s −1 Mpc −1 . We also constrain the absolute magnification of each image to μ a = 4.3 − 1.8 + 1.6 , μ b = 7.6 − 2.6 + 3.6 , μ c = 6.4 − 1.5 + 1.6 by comparing the observed peak near-IR magnitude of SN H0pe to the nonlensed population of SNe Ia.
We present time-delay measurements for the new quadruple imaged quasar DES J0408−5354, the first quadruple imaged quasar found in the Dark Energy Survey (DES). Our result is made possible by ...implementing a new observational strategy using almost daily observations with the MPIA 2.2 m telescope at La Silla observatory and deep exposures reaching a signal-to-noise ratio of about 1000 per quasar image. This data qualityallows us to catch small photometric variations (a few mmag rms) of the quasar, acting on temporal scales much shorter than microlensing, and hence making the time delay measurement very robust against microlensing. In only seven months we very accurately measured one of the time delays in DES J0408−5354: Δt(AB) = −112.1 ± 2.1 days (1.8%) using only the MPIA 2.2 m data. In combination with data taken with the 1.2 m Euler Swiss telescope, we also measured two delays involving the D component of the system Δt(AD) = −155.5 ± 12.8 days (8.2%) and Δt(BD) = −42.4 ± 17.6 days (41%), where all the error bars include systematics. Turning these time delays into cosmological constraints will require deep Hubble Space Telescope (HST) imaging or ground-based adaptive optics (AO), and information on the velocity field of the lensing galaxy.
Full text
Available for:
FMFMET, NUK, UL, UM, UPUK
We investigate the mass content of galaxies in the core of the galaxy cluster Abell 611. We perform a strong lensing analysis of the cluster core and use velocity dispersion measurements for ...individual cluster members as additional constraints. Despite the small number of multiply-imaged systems and cluster members with central velocity dispersions available in the core of A611, the addition of velocity dispersion measurements leads to tighter constraints on the mass associated with the galaxy component, and as a result, on the mass associated with the dark matter halo. Without the spectroscopic velocity dispersions, we would overestimate the mass of the galaxy component by a factor of ~1.5, or, equivalently, we would underestimate the mass of the cluster dark halo by ~5 per cent. We perform an additional lensing analysis using surface brightness (SB) reconstruction of the tangential giant arc. This approach improves the constraints on the mass parameters of the five galaxies close to the arc by a factor up to ~10. The resulting parameters are in good agreement with the sigma -r sub( tr) scaling relation derived in the pointlike analysis. The galaxy velocity dispersions resulting from the SB analysis are consistent at the 1 sigma confidence level with the spectroscopic measurements. In contrast, the truncation radii for 2-3 galaxies depart significantly from the galaxy scaling relation and suggest differences in the stripping history from galaxy to galaxy.
Aims. We perform a comprehensive study of the total mass distribution of the galaxy cluster RXC J2248.7−4431 (z = 0.348) with a set of high-precision strong lensing models, which take advantage of ...extensive spectroscopic information on many multiply lensed systems. In the effort to understand and quantify inherent systematics in parametric strong lensing modelling, we explore a collection of 22 models in which we use different samples of multiple image families, different parametrizations of the mass distribution and cosmological parameters. Methods. As input information for the strong lensing models, we use the Cluster Lensing And Supernova survey with Hubble (CLASH) imaging data and spectroscopic follow-up observations, with the VIsible Multi-Object Spectrograph (VIMOS) and Multi Unit Spectroscopic Explorer (MUSE) on the Very Large Telescope (VLT), to identify and characterize bona fide multiple image families and measure their redshifts down to mF814W ≃ 26. A total of 16 background sources, over the redshift range 1.0−6.1, are multiply lensed into 47 images, 24 of which are spectroscopically confirmed and belong to ten individual sources. These also include a multiply lensed Lyman-α blob at z = 3.118. The cluster total mass distribution and underlying cosmology in the models are optimized by matching the observed positions of the multiple images on the lens plane. Bayesian Markov chain Monte Carlo techniques are used to quantify errors and covariances of the best-fit parameters. Results. We show that with a careful selection of a large sample of spectroscopically confirmed multiple images, the best-fit model can reproduce their observed positions with a rms scatter of 0.̋3 in a fixed flat ΛCDM cosmology, whereas the lack of spectroscopic information or the use of inaccurate photometric redshifts can lead to biases in the values of the model parameters. We find that the best-fit parametrization for the cluster total mass distribution is composed of an elliptical pseudo-isothermal mass distribution with a significant core for the overall cluster halo and truncated pseudo-isothermal mass profiles for the cluster galaxies. We show that by adding bona fide photometric-selected multiple images to the sample of spectroscopic families, one can slightly improve constraints on the model parameters. In particular, we find that the degeneracy between the lens total mass distribution and the underlying geometry of the Universe, which is probed via angular diameter distance ratios between the lens and sources and the observer and sources, can be partially removed. Allowing cosmological parameters to vary together with the cluster parameters, we find (at 68% confidence level) Ωm = 0.25+ 0.13-0.16 and w = −1.07+ 0.16-0.42 for a flat ΛCDM model, and Ωm = 0.31+ 0.12-0.13 and ΩΛ = 0.38+ 0.38-0.27 for a Universe with w = −1 and free curvature. Finally, using toy models mimicking the overall configuration of multiple images and cluster total mass distribution, we estimate the impact of the line-of-sight mass structure on the positional rms to be 0.̋3 ± 0. We argue that the apparent sensitivity of our lensing model to cosmography is due to the combination of the regular potential shape of RXC J2248, a large number of bona fide multiple images out to z = 6.1, and a relatively modest presence of intervening large-scale structure, as revealed by our spectroscopic survey.
Full text
Available for:
FMFMET, NUK, UL, UM, UPUK
Strong gravitational lensing is a powerful technique for probing galaxy mass distributions and for measuring cosmological parameters. Lens systems with extended source-intensity distributions are ...particularly useful for this purpose since they provide additional constraints on the lens potential (mass distribution). We present a pixelated approach to modeling the lens potential and source-intensity distribution simultaneously. The method makes iterative and perturbative corrections to an initial potential model. For systems with sources of sufficient extent such that the separate lensed images are connected by intensity measurements, the accuracy in the reconstructed potential is solely limited by the quality of the data. We apply this potential reconstruction technique to deep Hubble Space Telescope observations of B1608+656, a four-image gravitational lens system formed by a pair of interacting lens galaxies. We present a comprehensive Bayesian analysis of the system that takes into account the extended source-intensity distribution, dust extinction, and the interacting lens galaxies. Our approach allows us to compare various models of the components of the lens system, which include the point-spread function (PSF), dust, lens galaxy light, source-intensity distribution, and lens potential. Using optimal combinations of the PSF, dust, and lens galaxy light models, we successfully reconstruct both the lens potential and the extended source-intensity distribution of B1608+656. The resulting reconstruction can be used as the basis of a measurement of the Hubble constant. As an illustration of the astrophysical applications of our method, we use our reconstruction of the gravitational potential to study the relative distribution of mass and light in the lensing galaxies. We find that the mass-to-light ratio for the primary lens galaxy is (2.0 ± 0.2)h M L -1 B, within the Einstein radius (3.9 h -1 kpc), in agreement with what is found for noninteracting lens galaxies at the same scales.