We are pleased to announce that our paper “Neural network prediction of model parameters for strong lensing samples from Hyper Suprime-Cam Survey” has been accepted to Monthly Notices of the Royal Astronomical Society (MNRAS).
Priyanka Gawade, Anupreeta More, Surhud More, Akisato Kimura, Alessandro Sonnenfeld, Masamune Oguri, Naoki Yoshida, “Neural network prediction of model parameters for strong lensing samples from Hyper Suprime-Cam Survey,” Monthly Notices of the Royal Astronomical Society (MNRAS), 2025.
Gravitational lensing can result in the formation of multiple images of sources which are sufficiently well aligned with a lensing potential. Observing and analyzing such strong gravitational lensing effects, for example, the distortions of the background sources, the positions and magnifications of the multiple images, and the time delays in the case of transient sources, can provide us crucial information about the properties of the lens and the source.
Forward lens modelling techniques are used estimate the parameters that describe the mass distribution in such individual lens systems. Lens modelling is computationally expensive, often requires attention to individual systems, in addition to development of sophisticated lens modelling codes. Some of the steps that require human attention are identifying lensed features, masking out contaminants, and finding an adequate initial guess for the lens model parameters, and finding an adequate initial guess for the lens model parameters. Finding more efficient ways to study gravitational lenses is the need of the hour given large upcoming surveys.
In this work, we build and train a simple convolutional neural network with an aim of rapidly predicting model parameters of gravitational lenses. We focus on the inference of the Einstein radius, and ellipticity components of the mass distribution. We train our network on a variety of simulated data with increasing degree of realism and compare its performance on simulated test data in a quantitative manner.
We also model 182 gravitational lenses from the HSC survey using YattaLens pipeline to infer their model parameters, which allow a benchmark to compare the predictions of the network. Given all considerations, we conclude that the network trained on simulated samples with lensed sources injected in empty HSC cutouts is the most robust, reproducing Einstein radii with an accuracy of about 10 − 20 percent, a bias less than 5 percent, and an outlier fraction of the order of 10 percent.
We argue in favor of the subtraction of the lens light before modelling the lens mass distribution. Our comparisons of the inferred parameters of 10 HSC lenses previously modelled in the literature, demonstrate agreement on the Einstein radius. However, the ellipticity components from the network as well as the individual modelling methods, seem to have systematic uncertainties beyond the quoted errors.
More details can be checked at the preprint that has already been uploaded to ArXiv https://arxiv.org/abs/2404.18897 .