Keywords
deep learning
strong gravitational lenses
Einstein's radio
image reconstruction
convolutional neural networks
How to Cite
Abstract
This study addresses the modeling of strong gravitational lenses (SGL) through convolutional neural networks (CNN), aiming to improve the inference of physical parameters associated with the singular isothermal ellipsoid (SIE) model. SGL are essential tools for probing dark matter and dark energy, as well as for obtaining insights into the internal structure of galaxies. However, the increasing volume of astronomical data challenges the efficiency of traditional modeling methods. A synthetic dataset of 76 396 galaxy-galaxy lensing images was
generated from the Chinese Space Station Telescope (CSST) catalog using the Lenstronomy package. The CNN model was based on the AlexNet architecture,
with modifications including additional convolutional layers and batch normalization to optimize feature extraction and reduce computational cost. The network predicts four parameters of the SIE Emodel: Einstein radius (θE), axis ratio (f), and ellipticity components (εx, εy). Results demonstrate high predictive performance, with determination coefficients (R²) ranging from 0.95 to 0.97 and average relative errors between 2.6% and 5.1%. Furthermore, reconstructed images achieved a fidelity of 37 dB in PSNR. These findings confirm the effectiveness of lightweight CNN architectures for the automatic modeling of gravitational lenses across massive astronomical datasets.
References
P. Schneider, C. S. Kochanek, y J. Wambsganss, Gravitational lensing: strong, weak and micro. Alemania: Springer Berlin, Heidelberg, 2006, DOI:10.1007/978-3-540-30310-7.
P. Schneider, Extragalactic Astronomy and Cosmology: An Introduction. Alemania: Springer Berlin, Heidelberg, 2015, DOI: 10.1007/978-3-642-54083-7.
Y. D. Hezaveh et al., “Detection of c substructure using ALMA observations of the dusty galaxy SDP.81”, arXiv:1601.01388, 2016, DOI: 10.48550/arXiv.1601.01388.
M. Meneghetti, Introduction to gravitational lensing: with Python examples, Suiza: Springer Nature, 2021, DOI: 10.1007/978-3-030-73582-1.
A. J. Shajib et al., “Strong Lensing by Galaxies”, Space Science Reviews, vol. 220, núm. 87, 2024, DOI: 10.1007/s11214-024-01105-x.
T. Treu, “Strong Lensing by Galaxies”, Annual Review of Astronomy and Astrophysics, vol. 48, pp. 87-125, 2010, DOI: 10.1146/annurev-astro-081309-130924.
Euclid Collaboration, “Euclid: I. Overview of the Euclid Mission”, Astronomy and Astrophysics, vol. 697, art. A1, 2025, DOI: 10.1051/0004-6361/202450810.
R. Laureijs et al., “Euclid Definition Study Report”, arXiv:1110.3193, 2011, DOI: 10.48550/arXiv.1110.3193.
D. Enard, “The European Southern Observatory Very Large Telescope”, Journal of Optics, vol. 22, núm. 2, pp. 33-50, 1991, DOI: https://doi.org10.1088/0150-536X/22/2/001.
M. W. McElwain et al., “The James Webb Space Telescope Mission: Optical Telescope Element Design, Development, and Performance”, Publications of the Astronomical Society of the Pacific, vol. 135, art. 058001, 2023, DOI: https://doi.org/10.1088/1538-3873/acada0.
J. P. Gardner et al., “The James Webb Space Telescope”, Space Science Reviews, vol. 123, pp. 485-606, 2006, DOI: 10.1007/s11214-006-8315-7.
F. B. Bianco et al., “Optimization of the Observing Cadence for the Rubin Observatory Legacy Survey of Space and Time: A Pioneering Process of Community-focused Experimental Design”, The Astrophysical Journal Supplement Series, vol. 258, núm. 1, 2021, DOI: https://doi.org/10.3847/1538-4365/ac3e72.
I. Hook, “The Science Case for the European ELT”, en Proceedings of the Science with the VLT in the ELT Era, A. Moorwood, Ed., 2009, pp. 225-232, DOI: 10.1007/978-1-4020-9190-2_38.
E. Palle et al., “Ground-Breaking Exoplanet Science with the ANDES Spectrograph at the ELT”, Experimental Astronomy, vol. 59, núm. 29, 2025, DOI: 10.1007/s10686-025-10000-4.
Y. Gong et al., “Introduction to the Chinese Space Station Survey Telescope (CSST)”, Science China Physics, Mechanics & Astronomy, vol. 69, núm. 3, art. 239501, 2026, DOI: 10.1007/s11433-025-2809-00.
T. E. Collett, “The Population of Galaxy-Galaxy Strong Lenses in Forthcoming Optical Imaging Surveys”, The Astrophysical Journal, vol. 811, núm. 1, 2015, DOI: 10.1088/0004-637X/811/1/20.
S. T. Myers et al., “The Cosmic Lens All-Sky Survey - I. Source Selection and Observations”, Monthly Notices of the Royal Astronomical Society, vol. 341, núm. 1, pp. 1-12, 2003, DOI: 10.1046/j.1365-8711.2003.06256.x.
I. W. A. Browne et al., “The Cosmic Lens All-Sky Survey - II. Gravitational Lens Candidate Selection and Follow-Up”, Monthly Notices of the Royal Astronomical Society, vol. 341, núm. 1, pp. 13-32, 2003, DOI: 10.1046/j.1365-8711.2003.06257.x.
N. Scoville et al., “The Cosmic Evolution Survey (COSMOS): Overview”, The Astrophysical Journal Supplement Series, vol. 172, núm. 1, pp. 1-8, 2007.
A. S. Bolton et al., “The Sloan Lens ACS Survey. I. A Large Spectroscopically Selected Sample of Massive Early-Type Lens Galaxies”, The Astrophysical Journal, vol. 638, núm. 2, pp. 703-724, 2006, DOI: 10.1086/498884.
A. S. Bolton et al., “The Sloan Lens ACS Survey. V. The Full ACS Strong-Lens Sample”, The Astrophysical Journal, vol. 682, núm. 2, pp. 964-984, 2008, DOI: 10.1086/589327.
F. Sciortino et al., “Inference of Experimental Radial Impurity Transport on Alcator C-Mod: Bayesian Parameter Estimation and Model Selection”, Nuclear Fusion, vol. 60, núm. 12, 2020 DOI: 10.1088/1741-4326/abae85.
A. Fowlie, W. Handley, y L. Su, “Nested Sampling Cross-Checks Using Order Statistics”, Monthly Notices of the Royal Astronomical Society, vol. 497, núm. 4, pp. 5256-5263, 2020, DOI: 10.1093/mnras/staa2345.
W. Zhang et al., “Shift-Invariant Neural Network for Image Processing: Learning and Generalization”, Applications of Artificial Neural Networks III, vol. 1709, pp. 257-268, 1992, DOI: 10.1117/12.140004.
P. Hála, “Spectral classification using convolutional neural networks”, arXiv:1412.8341, 2014, DOI: 10.48550/arXiv.1412.8341.
C. E. Petrillo et al., “Finding Strong Gravitational Lenses in the Kilo Degree Survey with Convolutional Neural Networks”, Monthly Notices of the Royal Astronomical Society, vol. 472, núm. 1, pp. 1129-1150, 2017, DOI: 10.1093/mnras/stx2052.
Y. D. Hezaveh, L. P. Levasseur, y P. J. Marshall, “Fast Automated Analysis of Strong Gravitational Lenses with Convolutional Neural Networks”, Nature, vol. 548, pp. 555-557, 2017, DOI: 10.1038/nature23463.
S. Schuldt et al., “HOLISMOKES - IV. Efficient Mass Modeling of Strong Lenses through Deep Learning”, Astronomy & Astrophysics, vol. 646, art. A126, 2021, DOI: 10.1051/0004-6361/202039574.
J. W. Park et al., “Large-Scale Gravitational Lens Modeling with Bayesian Neural Networks for Accurate and Precise Inference of the Hubble Constant”, The Astrophysical Journal, vol. 910, núm. 1, 2021, DOI: 10.3847/1538-4357/abdfc4.
J. Pearson, J. Maresca, N. Li, y S. Dye, “Strong Lens Modelling: Comparing and Combining Bayesian Neural Networks and Parametric Profile Fitting”, Monthly Notices of the Royal Astronomical Society, vol. 505, núm. 1, pp. 4362-4382, 2021, DOI: 10.1093/mnras/stab1547.
V. Busillo et al., “Euclid Quick Data Release (Q1). LEMON - LEns Modelling with Neural networks”, arXiv:2503.15329, 2026, DOI: 10.48550/arXiv.2503.15329.
R. Parlange et al., “GraViT: Transfer Learning with Vision Transformers and MLP-Mixer for Strong Gravitational Lens Discovery”, Monthly Notices of the Royal Astronomical Society, vol. 545, núm. 2, 2025, DOI: https://doi.org/10.1093/mnras/staf1747.
R. Cañameras et al., “HOLISMOKES - VI. New galaxy-scale strong lens candidates from the HSC-SSP imaging Survey”, Astronomy & Astrophysics, vol. 653, art. L6, 2021, DOI: 10.1051/0004-6361/202141758.
A. T. Jaelani et al., “Survey of gravitationally lensed objects in HSC imaging (SuGOHI) – X. Strong lens finding in the HSC-SSP using convolutional neural networks”, Monthly Notices of the Royal Astronomical Society, vol. 535, núm. 2, pp. 1625-1639, 2024, DOI: 10.1093/mnras/stae2442.
L. P. Garate Nuñez, A. S. G. Robotham, S. Bellstedt, L. J. M. Davies, C. Martínez Lombilla, “The Hyper Suprime-Cam extended point spread functions and applications”, Monthly Notices of the Royal Astronomical Society, vol. 531, núm. 2, pp. 2517-2530, 2024, DOI: 10.1093/mnras/stae1292.
G. E. Hinton, A. Krizhevsky, I. Sutskever, y Y. E. Rachmad, “ImageNet Classification with Deep Convolutional Neural Networks”, Communications of the ACM, vol. 60, núm. 6, pp. 84-90, 2012, DOI: 10.1145/3065386.
J. J. Ancona-Flores, A. Hernández-Almada, and V. Motta, “Enhancing gravitational lens study with deep learning: A study on effects of dropout regularization”, Galaxies, vol. 14, núm. 2, 2026, DOI: 10.3390/galaxies14020018.
R. Kormann, P. Schneider, y M. Bartelmann, “Isothermal Elliptical Gravitational Lens Models”, Astronomy and Astrophysics, vol. 284, pp. 285-299, 1994, https://www.semanticscholar.org/paper/Isothermal-elliptical-gravitational-lens-models.-Kormann-Schneider/ed1caf1f6fdf1866fc6eac34eef6a15e780037b2.
A. Etherington et al., “Automated Galaxy-Galaxy Strong Lens Modelling: No Lens Left Behind”, Monthly Notices of the Royal Astronomical Society, vol. 517, núm. 3, pp. 3275-3302, 2022, DOI: 10.1093/mnras/stac2639.
Y. LeCun et al., “Backpropagation applied to handwritten zip code recognition”, Neural computation, vol. 1, núm. 4, pp. 541-551, 1989, DOI: 10.1162/neco.1989.1.4.541.
P. Baldi y P. J. Sadowski, “Understanding Dropout”, Advances in Neural Information Processing Systems, vol. 26, 2013, https://papers.nips.cc/paper_files/paper/2013/hash/71f6278d140af599e06ad9bf1ba03cb0-Abstract.html.
G. E. Hinton, N. Srivastava, A. Krizhevsky, I. Sutskever, y R. R. Salakhutdinov, “Improving Neural Networks by Preventing Co-Adaptation of Feature Detectors”, arXiv:1207.0580, 2012, DOI: 10.48550/arXiv.1207.0580.
I. Salehin y D. K. Kang, “A Review on Dropout Regularization Approaches for Deep Neural Networks within the Scholarly Domain”, Electronics, vol. 12, núm. 14, 2023, DOI: 10.3390/electronics12143106.
J. L. Sérsic, Atlas de galaxias australes. Observatorio Astronómico, Argentina: Universidad Nacional de Córdoba, 1968.
S. Mukherjee et al.; “SEAGLE—I. A Pipeline for Simulating and Modelling Strong Lenses from Cosmological Hydrodynamic Simulations”, Monthly Notices of the Royal Astronomical Society, vol. 479, núm. 3, pp. 4108-4125, 2018, DOI: 10.1093/mnras/sty1741.
T. Treu y L. V. E. Koopmans, “Massive Dark Matter Halos and Evolution of Early-Type Galaxies to z ≈ 1”, The Astrophysical Journal, vol. 611, núm. 2, pp. 739-760, Aug, 2004, DOI: 10.1086/422245.
R. Caruana, “Multitask Learning”, Machine Learning, vol. 28, pp. 41-75, 1997, DOI: 10.1023/A:1007379606734.
Z. Wang, A. C. Bovik, H. R. Sheikh, y E. P.Simoncelli, “Image quality assessment: from error visibility to structural similarity”, IEEE Transactions on Image Processing, vol. 13, núm. 4, pp. 600-12, 2004, DOI: 10.1109/tip.2003.819861.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Copyright (c) 2026 Perspectivas de la Ciencia y la Tecnología