On Density, Threshold and Emptiness Queries for Intervals in the Streaming Model

In this paper, we study the maximum density, threshold and emptiness queries for intervals in the streaming model. The input is a stream S of n points in the real line R and a floating closed interval W of width α. The specific problems we consider in this paper are as follows. Maximum density: find a placement of W in R containing the maximum number of points of S. Threshold query: find a placement of W in R, if it exists, that contains at least ∆ elements of S. Emptiness query: find, if possible, a placement of W within the extent of S so that the interior of W does not contain any element of S. The stream S, being huge, does not fit into main memory and can be read sequentially at most a constant number of times, usually once. The problems studied here in the geometric setting have relations to frequency estimation and heavy hitter identification in a stream of data. We provide lower bounds and results on trade-off between extra space and quality of solution. We also discuss generalizations for the higher dimensional variants for a few cases. 1998 ACM Subject Classification F.2.2 Nonnumerical Algorithms and Problems