Unsymmetric meshless methods for operator equations (Q2267768)

The author proposes a general approach for approximatively solving the well-posed linear operator equations in normed linear spaces. The operators need not be elliptic and the problems can be posed in weak or strong form without changing the theory. The paper seems to be the first to provide a rigid mathematical foundation of certain variations of the ``meshless local Petrov-Galerkin'' technique of \textit{S. N. Atluri} and \textit{S. Shen} [The meshless local Petrov-Galerkin (MLPG) method. Encino, CA: Tech Science Press (2002; Zbl 1012.65116)]. The author is focused on unsymmetric methods and the abstract framework will not distinguish between strong and weak testing. Error bounds and convergence of a large class of unsymmetric meshless methods are discussed. Under a stability condition, optimal convergence rates a obtained. The abstract framework presented in this paper allows generalizations to other forms of testing and to other types of operator equations. As an example, the Poisson problem on a bounded domain with piecewise smooth Lipschitz boundary is discussed with the new approach in detail.