Physics-informed neural networks with adaptive localized artificial viscosity

• Published in: Journal of computational physics 2023-09, Vol.489, p.112265, Article 112265
• Main Authors: Coutinho, Emilio Jose Rocha, Dall'Aqua, Marcelo, McClenny, Levi, Zhong, Ming, Braga-Neto, Ulisses, Gildin, Eduardo
• Format: Article
• Language: English
• Subjects: Artificial viscosity; Hyperbolic PDEs; Physics-informed neural networks
• ISSN: 0021-9991; 1090-2716
• DOI: 10.1016/j.jcp.2023.112265

Physics-informed Neural Network (PINN) is a promising tool that has been applied in a variety of physical phenomena described by partial differential equations (PDE). However, it has been observed that PINNs are difficult to train in certain “stiff” problems, which include various nonlinear hyperbolic PDEs that display shocks in their solutions. Recent studies added a diffusion term to the PDE, and an artificial viscosity (AV) value was manually tuned to allow PINNs to solve these problems. In this paper, we propose three approaches to address this problem, none of which rely on an a priori definition of the artificial viscosity value. The first method learns a global AV value, whereas the other two learn localized AV values around the shocks, by means of a parametrized AV map or a residual-based AV map. We applied the proposed methods to the inviscid Burgers equation and the Buckley-Leverett equation, the latter being a classical problem in Petroleum Engineering. The results show that the proposed methods are able to learn both a small AV value and the accurate shock location and improve the approximation error over a nonadaptive global AV alternative method. •Methods to add adaptive artificial viscosity to train PINNs to solve hyperbolic PDEs with shocks.•Adaptive methods accurately learn both the value and location of the artificial viscosity.•Successfully solved the Inviscid Burgers and Buckley-Leverett equations.•Additional computation cost over the baseline PINN is small.