B.Koc from University of Seville had a secondment with Virginia Tech for 1,5 months. The purpose of the secondment was to build a consistent data-driven approach, residual-based data-driven, reduced-order model (R-D2-VMS-ROM) for the parameter-dependent convection-diffusion problem as well as to to show that the R-D2-VMS-ROM is more consistent and accurate than the standard data-driven, coefficient-based, one (D2-VMS-ROM).
The team run all the models for the parameter–dependent convection–diffusion problem with a predictive regime. Based on how the subscales are modeled, we implement two different R–D2–VMS–ROMs, they numerically compare the R–D2–VMS–ROM and the D2–VMS–ROM in terms of accuracy and consistency. They also compare the R–D2–VMS–ROM and D2–VMS–ROM results with the streamline upwind Petrov–Galerkin ROM (SUPG–ROM) with optimal stabilized coefficient which is found by trial and error.