“…Intuitively, every prefix of a greedy permutation is as informative as possible about the whole set, so greedy permutations form a natural ordering in which to stream large data sets. Because of these properties, greedy permutations have many additional applications, including color quantization [Xia97], progressive image sampling [ELPZ97], selecting landmarks of probabilistic roadmaps for motion planning [MAB98], point cloud simplification [MD03], halftone mask generation [SMR04], hierarchical clustering [DL05], detecting isometries between surface meshes [LF09], novelty detection and time management for autonomous robot exploration [GGD12], industrial fault detection [AYE12], and range queries seeking diverse sets of points in query regions [AAYI + 13].…”