Towards the development of unconditionally stable time-domain meshless numerical methods

Meshless methods have recently emerged as robust numerical techniques for electromagnetic modeling in time domain. In those methods, a problem domain is represented by scattered spatial nodes instead of numerical meshes, thus the conformal modeling of boundaries and solution refinements can be conveniently achieved. However, the CFL-like numerical stability condition still exists with these meshless methods, which prevents the methods being efficiently applied for general electromagnetic simulations. To overcome the problem, in this paper, we propose the unconditionally stable meshless methods by incorporating two efficiency-improved implicit schemes, namely the leapfrog alternating-direction-implicit (ADI) and the locally one-dimensional scheme (LOD) schemes, into the radial point interpolation meshless method (RPIM). The proposed methods are numerically verified for their unconditional stability, and are assessed for their numerical accuracy and efficiency. In comparisons with the conventional RPIM, computational cost can be saved by up to 80% with little sacrifice of accuracy.