Modeling the Propagation of Acoustic Waves in an Elastic Medium Using Physics-Informed Neural Networks

Accurate modeling of wave propagation in elastic media plays a central role in a variety of geophysical and engineering problems, including both acoustic and seismic phenomena. While acoustic waves are a specific subset of seismic waves (typically involving compressional modes in fluids or gases), seismic waves more generally encompass both acoustic and elastic wave types in solids. Classical numerical techniques, such as finite difference and finite element methods, can effectively simulate these dynamics but often require fine grids and significant computational overhead. In this work, we propose a novel framework based on Physics-Informed Neural Networks (PINNs) to enable efficient modeling of both acoustic and general seismic wave propagation in large or semi-infinite elastic domains. Our approach incorporates the full elastodynamic governing equations—including initial and boundary conditions—into the neural network’s loss function, ensuring solutions remain physically consistent without reliance on large labeled datasets. Qualitative comparison with conventional methods is discussed. Experimental results demonstrate that PINNs accurately capture complex wave phenomena and offer a potentially efficient, extensible alternative for computational geophysics.