Title:A locally-conservative proximal Galerkin method for pointwise bound constraints

Abstract:We introduce the first-order system proximal Galerkin (FOSPG) method, a locally mass-conserving, hybridizable finite element method for solving heterogeneous anisotropic diffusion and obstacle problems. Like other proximal Galerkin methods, FOSPG finds solutions by solving a recursive sequence of smooth, discretized, nonlinear subproblems. We establish the well-posedness and convergence of these nonlinear subproblems along with stability and error estimates under low regularity assumptions for the linearized equations obtained by solving each subproblem using Newton's method. The FOSPG method exhibits several advantages, including high-order accuracy, discrete maximum principle or bound-preserving discrete solutions, and local mass conservation. It also achieves prescribed solution accuracy within asymptotically mesh-independent numbers of subproblems and linear solves per subproblem iteration. Numerical experiments on benchmarks for anisotropic diffusion and obstacle problems confirm these attributes. Furthermore, an open-source implementation of the method is provided to facilitate broader adoption and reproducibility.