Scientific Discovery of Shock Dynamics via Generative Artificial Intelligence

Creating shock waves numerically is based on the behavior of partial differential equations. These equations contain parameters that can be adjusted to create a data set representing different scalar values, various surface boundary conditions, and the physics governing the domain. By creating a data set that represents different shock wave configurations, we can learn a mapping between partial differential equations and solution fields. In this way, the generative artificial intelligence algorithm can be tailored to shock waves and their physics, allowing for testing whether the artificial intelligence algorithm can create physics that physically represents shock waves by creating a large language model that can generate physics based on solution fields related to numerical solutions or advanced solution fields derived from sensors in physical experiments.