“…In such instances the I/O communication, rather than the CPU computation time, is the bottleneck. Most work on planar point location, especially if we allow the edges and vertices of to be changed dynamically, has focused on minimizing the CPU computation time under the assumption that the subdivision fits in main memory (e.g., [3,10,12,13,18,21,23]). Only a few results are known for I/O-efficient dynamic point location when the subdivision is stored in external memory [1,7].…”