Preconditioned descent algorithms for \(p\)-Laplacian (Q2385516)

The authors deal with some computational issues on a finite element discretization of the \(p\)-Laplacian with Dirichlet data. The authors propose a class of descent methods with multi-grid finite element preconditioners, and carry out a convergence analysis. A mesh-independent convergence rate is proved. The authors study the behavior of the algorithms with large \(p\). An essential part of the paper contains numerical experiments, which show that these algorithms are able to solve a large-scale \(p\)-Laplacian with very large \(p\). The algorithms are then used to solve a variational inequality.