Exterior Nonlocal Variational Inequalities
Harbir Antil, Madeline O. Horton, Mahamadi Warma
TL;DR
This work extends obstacle problems to the exterior of the observation domain by formulating a nonlocal variational inequality driven by the fractional Laplacian. It develops a rigorous well-posedness framework, establishing equivalence among a minimization problem, a variational inequality, a slack-variable (Lagrange multiplier) system, and a distributional Euler–Lagrange formulation, including a nonnegative multiplier and complementarity. Two penalty strategies, an $L^2$ Moreau–Yosida penalty and a Sobolev-norm penalty, are analyzed via Mosco convergence, yielding explicit rates: linear convergence of the solution in $W^{s,2}(\,\\Omega,\\ olinebreak[0] \\Sigma_1)$ and quadratic decay of constraint violation with respect to the penalty parameter, and linear convergence in the Sobolev penalty parameter for the corresponding multiplier. These contributions enable exterior obstacle modeling in nonlocal contexts and provide a foundation for numerical schemes and finite element approximations.
Abstract
This paper introduces a new class of variational inequalities where the obstacle is placed in the exterior domain that is disjoint from the observation domain. This is carried out with the help of nonlocal fractional operators. The need for such novel variational inequalities stems from the fact that the classical approach only allows placing the obstacle either inside the observation domain or on the boundary. A complete analysis of the continuous problem is provided. Additionally, perturbation arguments to approximate the problem are discussed.
