A Novel Double Link Structure (DLS) with Applications to Computational Engineering and Design

摘要

We present a new data storage structure that eliminates the defrag-mentation and scalability issues of using conventional hashing based associative containers and dynamic arrays in mesh-dependent computational algorithms. The efficiency stems from the allocation of memory in data containers that are always a pre-calculated fixed amount and contagious for fast random data access. We have demonstrated the use of DLS in particular operations such as mesh adjacencies, and the set-like operations on the associativities (classifications and features) where commonly used unstructured mesh databases fail to provide scalable memory allocations. Technical design details of DLS, its performance evaluation and storage requirements are investigated and compared against known unstructured mesh databases in this study.