Dynamic stabbing queries with sub-logarithmic local updates for overlapping intervals

Abstract

We present a data structure to maintain a set of intervals on the real line subject to fast insertions and deletions of the intervals, stabbing queries, and local updates. Intuitively, a local update replaces an interval by another one of roughly the same size and location. We inves-tigate whether local updates can be implemented faster than a deletion followed by an insertion. We present the first results for this problem for sets of possibly over-lapping intervals. If the maximum depth of the overlap (a.k.a. ply) is bounded by a constant, our data structure performs insertions, dele-tions and stabbing queries in time O(log n), and local updates in time O(log n/ log log n), where n is the number of intervals. We also analyze the dependence on the ply when it is not constant. Our results are adap-tive: the times depend on the current ply at the time of each operation.