HOLISMOKES -- IX. Neural network inference of strong-lens parameters and uncertainties from ground-based images

Abstract

Modeling of strong gravitational lenses is a necessity for further applications in astrophysics and cosmology. Especially with the large number of detections in current and upcoming surveys such as the Rubin Legacy Survey of Space and Time (LSST), it is timely to investigate in automated and fast analysis techniques beyond the traditional and time consuming Markov chain Monte Carlo sampling methods. Building upon our convolutional neural network (CNN) presented in Schuldt et al. (2021b), we present here another CNN, specifically a residual neural network (ResNet), that predicts the five mass parameters of a Singular Isothermal Ellipsoid (SIE) profile (lens center $x$ and $y$, ellipticity $e_x$ and $e_y$, Einstein radius ${\theta }_E$) and the external shear (${\gamma }_{ext,1}$, ${\gamma }_{ext,2}$) from ground-based imaging data. In contrast to our CNN, this ResNet further predicts a 1$\sigma$ uncertainty for each parameter. To train our network, we use our improved pipeline from Schuldt et al. (2021b) to simulate lens images using real images of galaxies from the Hyper Suprime-Cam Survey (HSC) and from the Hubble Ultra Deep Field as lens galaxies and background sources, respectively. We find overall very good recoveries for the SIE parameters, while differences remain in predicting the external shear. From our tests, most likely the low image resolution is the limiting factor for predicting the external shear. Given the run time of milli-seconds per system, our network is perfectly suited to predict the next appearing image and time delays of lensed transients in time. Therefore, we also present the performance of the network on these quantities in comparison to our simulations. Our ResNet is able to predict the SIE and shear parameter values in fractions of a second on a single CPU such that we are able to process efficiently the huge amount of expected galaxy-scale lenses in the near future.

17 pages, including 11 figures, published with A&A

References in corpus (35)

• Array Programming with NumPy
• The NumPy array: a structure for efficient numerical computation
• Large Magellanic Cloud Cepheid Standards Provide a 1% Foundation for the Determination of the Hubble Constant and Stronger Evidence for Physics Beyond LambdaCDM
• The Hubble Ultra Deep Field
• Cosmic Distances Calibrated to 1% Precision with Gaia EDR3 Parallaxes and Hubble Space Telescope Photometry of 75 Milky Way Cepheids Confirm Tension with LambdaCDM
• A Bayesian approach to strong lensing modelling of galaxy clusters
• H0LiCOW V. New COSMOGRAIL time delays of HE0435-1223: $H_0$ to 3.8% precision from strong lensing in a flat $\Lambda$CDM model
• Multiple Images of a Highly Magnified Supernova Formed by an Early-Type Cluster Galaxy Lens
• Fast Automated Analysis of Strong Gravitational Lenses with Convolutional Neural Networks
• The Halos of Satellite Galaxies: the Companion of the Massive Elliptical Lens SL2S J08544-0121
• The CFHTLS Strong Lensing Legacy Survey: I. Survey overview and T0002 release sample
• The MUSE Hubble Ultra Deep Field Survey: II. Spectroscopic redshifts and comparisons to color selections of high-redshift galaxies
• The SL2S Galaxy-scale Lens Sample. V. Dark Matter Halos and Stellar IMF of Massive Early-type Galaxies out to Redshift 0.8
• Uncertainties in Parameters Estimated with Neural Networks: Application to Strong Gravitational Lensing
• The weirdest SDSS galaxies: results from an outlier detection algorithm
• RingFinder: automated detection of galaxy-scale gravitational lenses in ground-based multi-filter imaging data
• The Sloan Lens ACS Survey. XIII. Discovery of 40 New Galaxy-Scale Strong Lenses
• PyAutoLens: Open-Source Strong Gravitational Lensing
• Arcfinder: An algorithm for the automatic detection of gravitational arcs
• VDES J2325-5229 a z=2.7 gravitationally lensed quasar discovered using morphology independent supervised machine learning
• STRIDES: Automated uniform models for 30 quadruply imaged quasars
• HOLISMOKES. VI. New galaxy-scale strong lens candidates from the HSC-SSP imaging survey
• Automated galaxy-galaxy strong lens modelling: no lens left behind
• HOLISMOKES. VIII. High-redshift, strong-lens search in the Hyper Suprime-Cam Subaru Strategic Program
• The use of convolutional neural networks for modelling large optically-selected strong galaxy-lens samples
• TDCOSMO. IX. Systematic comparison between lens modelling software programs: time delay prediction for WGD 2038-4008
• CHITAH: Strong-gravitational-lens hunter in imaging surveys
• A Neural Network Gravitational Arc Finder based on the Mediatrix filamentation Method
• GIGA-Lens: Fast Bayesian Inference for Strong Gravitational Lens Modeling
• TDCOSMO X. Automated Modeling of 9 Strongly Lensed Quasars and Comparison Between Lens Modeling Software
• Large-Scale Gravitational Lens Modeling with Bayesian Neural Networks for Accurate and Precise Inference of the Hubble Constant
• Nested sampling cross-checks using order statistics
• Constraining the multi-scale dark-matter distribution in CASSOWARY 31 with strong gravitational lensing and stellar dynamics
• Auto-identification of unphysical source reconstructions in strong gravitational lens modelling
• The dynamical structure of broken power-law and double power-law models for dark matter haloes

Cited by in corpus (12)

• The CosmoVerse White Paper: Addressing observational tensions in cosmology with systematics and fundamental physics
• Strong Lensing by Galaxies
• Streamlined Lensed Quasar Identification in Multiband Images via Ensemble Networks
• HOLISMOKES -- X. Comparison between neural network and semi-automated traditional modeling of strong lenses
• LeMoN: Lens Modelling with Neural networks -- I. Automated modelling of strong gravitational lenses with Bayesian Neural Networks
• HOLISMOKES -- XI. Evaluation of supervised neural networks for strong-lens searches in ground-based imaging surveys
• Exploiting the diversity of modeling methods to probe systematic biases in strong lensing analyses
• HOLISMOKES XIII: Strong-lens candidates at all mass scales and their environments from the Hyper-Suprime Cam and deep learning
• Lens Modeling of STRIDES Strongly Lensed Quasars using Neural Posterior Estimation
• Accelerating lensed quasar discovery and modeling with physics-informed variational autoencoders
• HOLISMOKES XVI: Lens search in HSC-PDR3 with a neural network committee and post-processing for false-positive removal
• A compact group lens modeled with GIGA-Lens: Enhanced inference for complex systems