A Common Mesh Infrastructure for Parallel Adaptive Multi-Scale and Multi-Physics Integrated Simulations in Complex Geometries

The goal of this proposal is to develop a common mesh infrastructure (CMI) and algorithms for efficient adaptive simulations of transient multi-scale and multi-physics processes in complex geometries based on a new, fully threaded tree (FTT) abstract data type. An outcome of the proposed research will be a universal, platform-independent, stand-alone software library for managing a geometrically and topologically complex adaptive computational mesh shared by multiple algorithms during integrated numerical simulations of multi-scale and multi-physics phenomena. The CMI will define both global topology of a computational domain and local connectivity information on the mesh through a map of a multi-dimensional computational space onto a one-dimensional abstract index space. A critical property of the proposed CMI is that it will enable adaptive numerical algorithms to operate on an index space instead of a computational space, and will free the algorithms from topological and geometrical constraints either explicitly defined or implicitly imposed by application-specific and computer language-specific layouts of computational data in a computer memory, thus making the CMI-based adaptive algorithms flexible, portable, and universal.