An acceleration approach for rendering 3D FDTD grids - automatically generated from faceted models of complex geometries

As the raw power of and access to high performance computers (HPC) continues to grow, finite-difference methods are becoming more widely used in several areas of research. A particular favorite method for studying electromagnetic phenomena is the finite-difference time-domain (FDTD) method. Common to all research areas is the requirement to work with regular grids. While these grids are simple to work with, there do not appear to be good tools available to manipulate structured grids. The vast majority of geometry libraries are formatted for codes that require unstructured faceted grids. Automatic grid generation tools have been developed to convert files from faceted formats into structured grids. Thus, all the power and resources available to the finite element community can now be shared and exploited by the finite-difference community. The time required to render a FDTD grid can now be accelerated so that it is comparable to that for a faceted grid of the same size. As HPC resources continue to mature, the efficiency and simplicity of the FDTD method will cause it to continue to become a more widely implemented analytical tool of choice.