Exponential convergence of the \(hp\) virtual element method in presence of corner singularities (Q2413467)

The authors continue their series of papers on the virtual element method mainly basing on their paper [Math. Models Methods Appl. Sci. 26, No. 8, 1567--1598 (2016; Zbl 1344.65109)] and on the book of \textit{C. Schwab} [\(p\)- and \(hp\)-finite element methods. Theory and applications in solid and fluid mechanics. Oxford: Clarendon Press (1998; Zbl 0910.73003)]. In this paper, the aim is to prove exponential convergence in the presence of singularities in corner points of the polygonal boundary in a homogeneous Dirichlet problem for the Poisson equation in two dimensions. The virtual element method allows polygonal nonconvex elements with hanging nodes, but has the speciality that its basic functions are defined only depending on their polynomial values on the boundaries of the elements while inside the functions are unknown (though belonging to \(H^1\)). Concerning regularity, the solution of the problem is assumed to belong to Babuška or weighted Sobolev spaces. For the approximation of this solution, the authors allow for non-uniformly distributed degree of the polynomials for the virtual elements but restrict themselves to degrees being at most 2. The discrete bilinear form is constructed elementwise including a stabilizing bilinear form containg element boundary terms, the local degree of accuracy and \(L^2\) projections. They prove stability of this approximation and its exponential convergence when the number of degrees of freedom goes to infinity and the mesh around any singularity is geometrically refined. For the proof of convergence they parallel developments in the cited book of Schwab, use own former results and show a (compared to \textit{E. H. Georgoulis} [Math. Comput. 77, No. 261, 201--219 (2008; Zbl 1130.65063)]) simplified (for their polygonal elements) proof of an inverse estimate in an appendix. In the numerical experiments of the authors they exhibit results belonging to (from the point of view of \(hp\)-FEM) rather strange grids and get comparable results to those of \(hp\)-FEM on a standard graded FEM grid.