An Optimized, Parallel Computation of the Ghost Layer for Adaptive Hybrid Forest Meshes

Abstract

We discuss parallel algorithms to compute the ghost layer in computational, distributed memory, recursively adapted meshes. Its creation is a fundamental, necessary task in executing most parallel, element-based computer simulations. Common methods differ in that the ghost layer may either be inherently part of the mesh data structure that is maintained and modified, or kept separate and constructed/deleted as needed. In this work, we present a design following the latter approach, which we chose for its modularity of algorithms and data structures. We target arbitrary adaptive, nonconforming forest-of-trees meshes of mixed element shapes, such as cubes, prisms, and tetrahedra, and restrict ourselves to ghost elements across mesh faces. Our algorithm has low code complexity and redundancy since we reduce it to generic codimension-1 subalgorithms that can be flexibly combined. We recover older algorithms for cubic elements as special cases and optimize further using recursive, amortized tree searches and traversals.