Differentiable Strong Lensing: Uniting Gravity and Neural Nets through Differentiable Probabilistic Programming

Abstract

Since upcoming telescopes will observe thousands of strong lensing systems, creating fully-automated analysis pipelines for these images becomes increasingly important. In this work, we make a step towards that direction by developing the first end-to-end differentiable strong lensing pipeline. Our approach leverages and combines three important computer science developments: (a) convolutional neural networks, (b) efficient gradient-based sampling techniques, and (c) deep probabilistic programming languages. The latter automatize parameter inference and enable the combination of generative deep neural networks and physics components in a single model. In the current work, we demonstrate that it is possible to combine a convolutional neural network trained on galaxy images as a source model with a fully-differentiable and exact implementation of gravitational lensing physics in a single probabilistic model. This does away with hyperparameter tuning for the source model, enables the simultaneous optimization of nearly one hundred source and lens parameters with gradient-based methods, and allows the use of efficient gradient-based posterior sampling techniques. These features make this automated inference pipeline potentially suitable for processing a large amount of data. By analyzing mock lensing systems with different signal-to-noise ratios, we show that lensing parameters are reconstructed with percent-level accuracy. More generally, we consider this work as one of the first steps in establishing differentiable probabilistic programming techniques in the particle astrophysics community, which have the potential to significantly accelerate and improve many complex data analysis tasks.

13 pages, 9 figures. Updated to match version published in MNRAS

References in corpus (59)

• The Structure of Cold Dark Matter Halos
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Particle Dark Matter: Evidence, Candidates and Constraints
• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Machine learning and the physical sciences
• The Cosmic Evolution Survey (COSMOS) -- Overview
• Halo Formation in Warm Dark Matter Models
• The Dark Energy Survey: more than dark energy - an overview
• LSST Science Book, Version 2.0
• The COSMOS Survey: Hubble Space Telescope / Advanced Camera for Surveys (HST/ACS) Observations and Data Processing
• A New Era in the Quest for Dark Matter
• COSMOS : Hubble Space Telescope Observations
• The properties of warm dark matter haloes
• Strong Lensing by Galaxies
• Gravitational detection of a low-mass dark satellite at cosmological distance
• Jet Substructure at the Large Hadron Collider: A Review of Recent Advances in Theory and Machine Learning
• Detection of lensing substructure using ALMA observations of the dusty galaxy SDP.81
• Lenstronomy: multi-purpose gravitational lens modelling software package
• Detection of a Dark Substructure through Gravitational Imaging
• Bayesian Strong Gravitational-Lens Modeling on Adaptive Grids: Objective Detection of Mass Substructure in Galaxies
• Warm dark matter chills out: constraints on the halo mass function and the free-streaming length of dark matter with 8 quadruple-image strong gravitational lenses
• Fast Automated Analysis of Strong Gravitational Lenses with Convolutional Neural Networks
• A Bayesian analysis of regularised source inversions in gravitational lensing
• FIRE in the Field: Simulating the Threshold of Galaxy Formation
• Semi-linear gravitational lens inversion
• Edward: A library for probabilistic modeling, inference, and criticism
• The stellar mass-halo mass relation of isolated field dwarfs: a critical test of $\Lambda$CDM at the edge of galaxy formation
• Masked Autoregressive Flow for Density Estimation
• Improving Variational Inference with Inverse Autoregressive Flow
• Gravitational Lensing By Spiral Galaxies
• Gravitational lens modeling with basis sets
• The Third Gravitational Lensing Accuracy Testing (GREAT3) Challenge Handbook
• Fast calculation of a family of elliptical mass gravitational lens models
• Uncertainties in Parameters Estimated with Neural Networks: Application to Strong Gravitational Lensing
• Probing dark matter structure down to $10^7$ solar masses: flux ratio statistics in gravitational lenses with line of sight halos
• AutoLens: Automated Modeling of a Strong Lens's Light, Mass and Source
• Modelling the line-of-sight contribution in substructure lensing
• Low-mass halo perturbations in strong gravitational lenses at redshift z$\sim$0.5 are consistent with CDM
• Dissecting the Gravitational Lens B1608+656. I. Lens Potential Reconstruction
• Sylvester Normalizing Flows for Variational Inference
• CosmoGAN: creating high-fidelity weak lensing convergence maps using Generative Adversarial Networks
• Euclid Imaging Consortium Science Book
• Forward-Mode Automatic Differentiation in Julia
• Constraining sterile neutrino cosmologies with strong gravitational lensing observations at redshift z~0.2
• Diagnosing and Enhancing VAE Models
• Evaporating the Milky Way halo and its satellites with inelastic self-interacting dark matter
• Diversity in density profiles of self-interacting dark matter satellite halos
• Mining for Dark Matter Substructure: Inferring subhalo population properties from strong lenses with machine learning
• Substructure in the lens HE 0435-1223
• Deblending galaxy superpositions with branched generative adversarial networks
• Deep Learning the Morphology of Dark Matter Substructure
• Direct Detection of Dark Matter Substructure in Strong Lens Images with Convolutional Neural Networks
• Taming VAEs
• Stochastic Variational Inference
• Neural physical engines for inferring the halo mass distribution function
• Mesh-free free-form lensing I: Methodology and application to mass reconstruction
• Galaxy mass profiles from strong lensing I: The circular power-law model
• Modelling of the Complex CASSOWARY/SLUGS Gravitational Lenses
• Strong Lensing considerations for the LSST observing strategy

Cited by in corpus (29)

• The DAWES review 10: The impact of deep learning for the analysis of galaxy surveys
• Hierarchical Inference With Bayesian Neural Networks: An Application to Strong Gravitational Lensing
• Statistical strong lensing. I. Constraints on the inner structure of galaxies from samples of a thousand lenses
• Sensitivity of strong lensing observations to dark matter substructure: a case study with Euclid
• GIGA-Lens: Fast Bayesian Inference for Strong Gravitational Lens Modeling
• Using wavelets to capture deviations from smoothness in galaxy-scale strong lenses
• Strong Lensing by Galaxies
• Strong lens modelling: comparing and combining Bayesian neural networks and parametric profile fitting
• Extracting the Subhalo Mass Function from Strong Lens Images with Image Segmentation
• SLITronomy: towards a fully wavelet-based strong lensing inversion technique
• Fast and robust Bayesian Inference using Gaussian Processes with GPry
• Efficient Gravitational Wave Template Bank Generation with Differentiable Waveforms
• One never walks alone: the effect of the perturber population on subhalo measurements in strong gravitational lenses
• First results of the glitching pulsars monitoring program at the Argentine Institute of Radioastronomy
• Finding Strong Gravitational Lenses Through Self-Attention
• Kernel Phase and Coronagraphy with Automatic Differentiation
• Machine Learning and Cosmology
• Strong-lensing source reconstruction with variationally optimised Gaussian processes
• Quantifying the structure of strong gravitational lens potentials with uncertainty-aware deep neural networks
• Modeling lens potentials with continuous neural fields in galaxy-scale strong lenses
• Targeted Likelihood-Free Inference of Dark Matter Substructure in Strongly-Lensed Galaxies
• Dark Matter Subhalos, Strong Lensing and Machine Learning
• Vela Pulsar: Single Pulses Analysis with Machine Learning Techniques
• Analytic auto-differentiable $\Lambda$CDM cosmography
• Transformers as Strong Lens Detectors- From Simulation to Surveys
• A compact group lens modeled with GIGA-Lens: Enhanced inference for complex systems
• Embedded-model flows: Combining the inductive biases of model-free deep learning and explicit probabilistic modeling
• Differentiable probabilistic programming for strong gravitational lensing
• Determining the Dark Matter distribution in galaxies with Deep Learning