Searcharxiv⌕ Search

arXiv subjects

Anouar El Harrak

Publications and source records attributed to Anouar El Harrak.

2 recordsLinked to original sources

Exterior-Row Replacement for Grid-Based Multinode Shepard Collocation on Implicit Planar Domains

Grid-based multinode Shepard collocation combines tensor-product Lagrange interpolants on overlapping Cartesian subgrids. On an implicitly defined domain, retained subgrids may cross the boundary and contain exterior source nodes. Removing or moving these nodes changes the local interpolants. We introduce exterior-row replacement (ERR) to preserve the source nodes and coefficients of every retained subgrid. We assemble PDE rows at strictly interior sources and supply the remaining equations through boundary evaluations. We consider two square completion methods based on projection and redistribution, an oversampled weighted least-squares one (ERR-OS), and a completion method that selects boundary rows through the interior PDE nullspace (ERR-QR). Full column rank is characterized by the boundary action on this nullspace. We also give completion conditions for prescribed fitted rows and an error estimate in terms of the residual. Global-full (GF) support uses all retained subgrids, whereas nearest-active (NA) support uses only those containing the active source nearest to the evaluation point. NA keeps off-source boundary rows sparse. The numerical study covers a disk, an annulus, a smooth star-shaped domain, a square hole with fitted and implicit boundary components, and a fitted L-shaped domain. With degree-$4$ local tensor polynomials, ERR-OS and ERR-QR recover the expected fifth-order accuracy in the reported refinement study. Further experiments examine the effects of boundary oversampling and trimming on algebraic admissibility and boundary observability.

math.NA↗

Efficient and Fast Tensor-Product Multinode Shepard Collocation for Elliptic PDEs on Cartesian Grids

We introduce a Grid-Based Multinode Shepard Collocation Method (GBMSC) for two-dimensional elliptic boundary value problems on rectangular domains. The method combines Shepard-type partition functions with tensor-product Lagrange interpolation on overlapping local Cartesian subgrids. The Cartesian structure avoids the local unisolvency search required by multinode Shepard constructions on scattered data, while the local character of the approximation yields sparse collocation matrices and moderate conditioning. We establish the main structural properties of the method, including the cardinal property, tensor-product polynomial reproduction, and local interpolation estimates. Numerical experiments for Poisson problems with Dirichlet and mixed boundary conditions confirm the polynomial reproduction property up to round-off accuracy and show regular convergence for smooth non-polynomial solutions. Comparisons with Multinode Shepard Collocation Method (MSC) and Kansa's collocation methods indicate that the proposed discretization preserves high accuracy with significantly better conditioning than the RBF-based schemes considered.

math.NA↗