Abstract

Let G be a weighted, directed, acyclic graph in which each edge weight is not a static quantity, but can be reduced for a certain cost. In this paper we consider the problem of determining which edges to reduce so that the length of the longest paths is minimized and the total cost associated with the reductions does not exceed a given cost. We consider two types of edge reductions, linear reductions and 0/1 reductions, which model different applications. We present efficient algorithms for different classes of graphs, including trees, series-parallel graphs, and directed acyclic graphs, and we show other edge reduction problems to be NP-hard.

Citations