Convex hull property and maximum principle for finite element minimisers of general convex functionals (Q2391813): Difference between revisions

For some classes of minimisation problems with energy functionals, the vector-valued solutions satisfy a convex hull property, see [\textit{M. Bildhauer} and \textit{M. Fuchs}, Asymptotic Anal. 32, No. 3--4, 293--315 (2002; Zbl 1076.49018)], for example. This convex hull property represents a generalisation of the maximum principle from scalar to vector-valued functions. In the current paper, the authors consider minimisation problems with a functional of the form \(J(v) = \int_\Omega f(x,\| \nabla v (x) \|) dx\) for vector-valued functions \(v\) with predetermined boundary conditions, where the Frobenius matrix norm is used. The domain \(\Omega\) is allowed to have an arbitrary dimension larger than one. A finite element method with a simplicial triangulation and piecewise linear polynomials yields a discretisation of the problem. The main result of the paper consists in the proof of the convex hull property for the solution of the discrete problem under certain assumptions. For example, the triangulation is required to be non-obtuse, i.e., a restriction on the angles between the sides of each simplex separately appears. Furthermore, the authors apply the theoretical results to show maximum principles for scalar quasilinear partial differential equations of elliptic type like the \(p\)-Laplace problem and the mean curvature problem. A strong convex hull property is proved for nonlinear elliptic equations with vector-valued solutions under stronger assumptions on the triangulation. Results from numerical simulations of test examples are not within the scope of the paper.