A Goal-Oriented and Model-Constrained Optimization (GOMCO) approach was proposed as a discriminant technique for the Variational Geomano Method (VGM) in the coefficient determinations of variational multiscale Unresolved-Scale (URS) model in steady Stokes equations. Numerical implementations using both linear and nonlinear models were performed with both the GOMCO and VGM. Numerical results show that the coefficients determined by the GOMCO are scale-invariant, while they are scale-variant by the VGM. The GOMCO technique is found to be more appropriate for coefficient determinations in steady Stokes equations, as the VGM is sensitive to the computing procedures. Moreover, the GOMCO could provide reliable coefficients for the URS model.
- goal-oriented optimization
- stokes equations
- variational Germano identity
- Variational multiscale method