This document summarizes a research article that investigates using physics-informed neural networks (PINNs) to solve the forward and inverse problems of the Euler equations, which model high-speed aerodynamic flows. For the forward problem, PINNs are used to solve 1D and 2D Euler equations with both smooth and discontinuous solutions. For the inverse problem, PINNs infer density, velocity, and pressure fields using limited data like density gradients and pressure values. PINNs are shown to be better suited than traditional methods for inverse problems that are otherwise unsolvable. The choice of data location is important, and characteristics-form Euler equations yield better results than conservative form. PINNs can also identify parameters like

Physics-Informed Neural Networks For High-Speed Ows: Computer Methods in Applied Mechanics and Engineering March 2020 PDF

The list of books you might like