From: Hybrid modeling: towards the next level of scientific computing in engineering
Sect. | Physics | Data | Approach |
---|---|---|---|
3D magnetostatics: BEM discretization | Fields measured by Hall probe or moving coil | Synthesis of physics and data by Bayesian update | |
2D/3D magnetostatics: Ampère’s & Gauss’s law | Measured material data points \((\boldsymbol{B}^{*}_{i},\boldsymbol{H}^{*}_{i} )\) | Projection of data into admissible physics manifold | |
Electromagnetic Darwin model | Data sampling from physics model | Bayesian Optimization: GP machine learning |