Are fibonacci heaps optimal?

Abstract

In this paper we investigate the inherent complexity of the priority queue abstract data type. We show that, under reasonable assumptions, there exist sequences of n Insert, n Delete, m DecreaseKey and t FindMin operations, where 1 ≤ t ≤ n, which have ω(nlogt+ n + m) complexity. Although Fibonacci heaps do not achieve this bound, we present a modified Fibonacci heap which does, and so is optimal under our assumptions. Using our modified Fibonacci heap we are able to obtain an efficient algorithm for the shortest path problem.