Paper detail

An LP-Rounding $2\sqrt{2}$ Approximation for Restricted Maximum Acyclic Subgraph

In the classical Maximum Acyclic Subgraph problem (MAS), given a directed-edge weighted graph, we are required to find an ordering of the nodes that maximizes the total weight of forward-directed edges. MAS admits a 2 approximation, and this approximation is optimal under the Unique Game Conjecture. In this paper we consider a generalization of MAS, the Restricted Maximum Acyclic Subgraph problem (RMAS), where each node is associated with a list of integer labels, and we have to find a labeling of the nodes so as to maximize the weight of edges whose head label is larger than the tail label. The best known (almost trivial) approximation for RMAS is 4. The interest of RMAS is mostly due to its connections with the Vertex Pricing problem (VP). In VP we are given an undirected graph with positive edge budgets. A feasible solution consists of an assignment of non-negative prices to the nodes. The profit for each edge $e$ is the sum of its endpoints prices if that sum is at most the budget of $e$, and zero otherwise. Our goal is to maximize the total profit. The best known approximation for VP, which works analogously to the mentioned approximation algorithm for RMAS, is 4. Improving on that is a challenging open problem. On the other hand, the best known 2 inapproximability result is due to a reduction from a special case of RMAS. In this paper we present an improved LP-rounding $2\sqrt{2}$ approximation for RMAS. Our result shows that, in order to prove a 4 hardness of approximation result for VP (if possible), one should consider reductions from harder problems. Alternatively, our approach might suggest a different way to design approximation algorithms for VP.