2022/05/16 by Serafim Grubas, Serafim I. Grubas, Anton Duchkov +3 · 29 citations
Earth and Planetary Sciences · Engineering · Mathematics · Physics and Astronomy · #Algorithm #Applied mathematics #Artificial intelligence #Artificial neural network #Computer science #Differential equation #Eikonal equation #Fast marching method #Mathematical analysis #Mathematical optimization #Mathematics #Model Reduction and Neural Networks #Ordinary differential equation #Physics #Reservoir Engineering and Simulation Methods #Seismic Imaging and Inversion Techniques #Solver #physics.comp-ph #physics.geo-ph
paper · pdf · doi:10.1016/j.jcp.2022.111789
published in Journal of Computational Physics 474, 111789 (Elsevier BV) · The paper has 14 pages and 6 figures. Source code is available at https://github.com/sgrubas/NES
arxiv created 2022/05/16 · openalex publication_date 2022/11/21 · arxiv updated 2022/12/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/27
The concept of physics-informed neural networks has become a useful tool for solving differential equations due to its flexibility. There are a few approaches using this concept to solve the eikonal equation which describes the first-arrival traveltimes of acoustic and elastic waves in smooth heterogeneous velocity models. However, the challenge of the eikonal is exacerbated by the velocity models producing caustics, resulting in instabilities and deterioration of accuracy due to the non-smooth solution behaviour. In this paper, we revisit the problem of solving the eikonal equation using neural networks to tackle the caustic pathologies. We introduce the novel Neural Eikonal Solver (NES) for solving the isotropic eikonal equation in two formulations: the one-point problem is for a fixed source location; the two-point problem is for an arbitrary source-receiver pair. We present several techniques which provide stability in velocity models producing caustics: improved factorization; non-symmetric loss function based on Hamiltonian; gaussian activation; symmetrization. In our tests, NES showed the relative-mean-absolute error of about 0.2-0.4% from the second-order factored Fast Marching Method, and outperformed existing neural-network solvers giving 10-60 times lower errors and 2-30 times faster training. The inference time of NES is comparable with the Fast Marching. The one-point NES provides the most accurate solution, whereas the two-point NES provides slightly lower accuracy but gives an extremely compact representation. It can be useful in various seismic applications where massive computations are required (millions of source-receiver pairs): ray modeling, traveltime tomography, hypocenter localization, and Kirchhoff migration.