Preserving average proximity in arrays

Programmers and data structure designers are often forced to choose between alternative structures. In storing these structures, preserving logical adjacencies or “proximity” is usually an important consideration. The combinatorial problem of storing arrays as various kinds of list structures is examined. Embeddings of graphs are used to model the loss of proximity involved in such storage schemes, and an elementary proof that arrays cannot be stored as linear lists with bounded loss of proximity is presented. Average loss of proximity is then considered, and it is shown that arrays cannot be stored as linear lists with only bounded loss of average proximity, but can be so stored in binary trees. The former result implies, for instance, that row major order is an asymptotically optimal storage strategy for arrays.

View this article in the ACM Digital Library.