On the relationship between semi-Lagrangian and Lagrange-Galerkin schemes (Q1950443): Difference between revisions

The author compares semi-Lagrangian (SL) and Lagrange-Galerkin (LG) schemes for variable coefficient advection equations in view of removing the assumption of constant advection speed from this theory. For simplicity, the one-dimensional case is considered. In the LG form, large time-step schemes are implemented in a finite element framework and inherit a number of related features, mainly the possibility of working on unstructured grids. The main ideas of this paper are sketched on the following model problem of the evolutive advection equation \(v_t (x, t) + f (x, t) \cdot \nabla v(x, t) = 0,\; (x, t) \in {\mathbb R}^N \times (0, T ]\), with initial conditions \(v(x, 0) = v_0(x), \; x \in {\mathbb R}^N\), mostly with \(N=1\). The construction of large time-step schemes for the above equation stems from the application of the method of characteristics. Under a first set of basic assumptions, Theorem 1 proves the equivalence of the SL and LG schemes. Under an additional set of basic assumptions, Theorem 4 establishes the stability of the area weighted LG scheme, which is the main result of the paper.