Implementation and numerical analysis of the nonlinear Galerkin methods with finite elements discretization (Q1339325)

A recent numerical algorithm called nonlinear Galerkin method (NLG) is analysed mostly in the context of finite element approximations. The paper first describes theoretical and numerical backgrounds of NLG methods which are based on the computation of approximate inertial manifolds. The authors suggest a finite element analogue to NLG, using the hierarchical bases (coarse grid corresponds to large wavelengths, finer grid to small ones). Several numerical experiments are then reported for the 2D Burgers and Navier-Stokes problems. The results are incomparably better than with classical Galerkin methods. Though the computations were done on a Cray 2 computer, with very fine grids, the authors themselves express some objections to unphysical solutions and suggest various future ways.