AI and the World's Hard Problems · Physics · arXiv · 2026-10-19
Gabor-Enhanced Physics-Informed Neural Networks for Fast Simulations of Acoustic Wavefields
Physics-informed neural networks (PINNs) for solving the scattered Helmholtz equation tend to fit low frequencies first, which slows convergence and reduces accuracy for high-frequency wavefields. The authors add trainable Gabor basis functions and learn a nonlinear mapping from input coordinates to a custom Gabor coordinate system. Numerical experiments on various velocity models reportedly show faster convergence, higher accuracy, and more robustness than traditional PINNs and earlier Gabor-based methods.