Physical Systems Modeled Without Physical Laws
This work addresses the challenge of reducing computational costs for physics simulations, but it is incremental as it applies existing tree-based methods to new simulation data.
The paper tackled the problem of emulating physics-based simulations (Navier-Stokes, stress analysis, electromagnetic fields) using tree-based machine learning methods without relying on physical laws, achieving generalization to finer spatial grids without computational costs.
Physics-based simulations typically operate with a combination of complex differentiable equations and many scientific and geometric inputs. Our work involves gathering data from those simulations and seeing how well tree-based machine learning methods can emulate desired outputs without "knowing" the complex backing involved in the simulations. The selected physics-based simulations included Navier-Stokes, stress analysis, and electromagnetic field lines to benchmark performance as numerical and statistical algorithms. We specifically focus on predicting specific spatial-temporal data between two simulation outputs and increasing spatial resolution to generalize the physics predictions to finer test grids without the computational costs of repeating the numerical calculation.