Low space data structures for geometric range mode query

Abstract

Let S be a set of n points in an [n] d grid, such that each point is assigned a color. Given a query range, the geometric range mode query problem asks to report the most frequent color (i.e., a mode) of the multiset of colors corresponding to points in S ∩ Q. When d = 1, Chan et al. (STACS 2012 [2]) gave a data structure that requires O(n + (n/∆) 2 /w) words of space and supports range mode queries in O(∆) time for any ∆ ≥ 1, where w = Ω(log n) is the word size. Chan et al. also proposed a data structures for higher dimensions (i.e., d ≥ 2) with O(s n + (n/∆) 2d ) space and O(∆ · t n ) query time, where s n and t n denote the space and query time of a data structure that supports orthogonal range counting queries on the set S. In this paper we show that the space can be improved without any increase to the query time, by presenting an O(s n + (n/∆) 2d /w)-space data structure that supports orthogonal range mode queries on a set of n points in d dimensions in O(∆ · t n ) time, for any ∆ ≥ 1. When d = 1, these space and query time costs match those achieved by the current best known one-dimensional data structure.

Linear-Space Data Structures for Range Frequency Queries on Arrays and Trees

“…Chan et al [8,9] achieved o( √ n) query time with an O(n)-space data structure that supports queries in O( √ n/w) ⊆ O( √ n/ log n) time on arrays (also see [17]). In this paper we present a new O(n)-space data structure that improves the path mode query time on trees to O( √ n/w log log n).…”