Surrogate parametric metamodel based on Optimal Transport

The description of a physical problem through model necessarily involves the introduction parameters. Hence, one wishes to have solution that is function all these parameters: parametric solution. However, construction such solutions exhibiting localization in space only ensured by costly and time-consuming tests, which can be both numerical or experimental. Numerical methodologies used classically imply enormous computational efforts for exploring design space. Therefore, obtained using advanced nonlinear regressions are an essential tool address this challenge. classical regression techniques, even most ones, lead non interpolation some fields as fluid dynamics, where localizes different regions depending on parameters choice. In context, Optimal Transport (OT) offers mathematical approach measure distances interpolate between general objects a, sometimes, more way than approach. Thus, OT has become fundamental statistics computer vision, it being increasingly mechanics. usually computationally solve not adapted accessed online manner. aim paper combining with order implement real-time based OT. To purpose, built offline relying Model Order Reduction OT, leading following theory. Such major interest processes, but also within digital twin rationale.