A Note on Submodular Maximization over Independence Systems

TL;DR

This paper investigates the complexity of maximizing submodular functions over independence systems, establishing hardness results, analyzing greedy algorithms, and introducing a nearly linear-time algorithm for specific cases.

Contribution

It provides the first nearly linear-time algorithm for non-monotone submodular maximization over p-extendible independence systems and characterizes approximation hardness.

Findings

Hardness of approximation within (2/n)^{1-ε} for general independence systems.

Greedy algorithm achieves a 2/n ratio under mild assumptions.

First nearly linear-time algorithm for p-extendible independence systems.

Abstract

In this work, we consider the maximization of submodular functions constrained by independence systems. Because of the wide applicability of submodular functions, this problem has been extensively studied in the literature, on specialized independence systems. For general independence systems, even when all of the bases of the independence system have the same size, we show that for any $\epsilon > 0$, the problem is hard to approximate within $(2/n)^{1-\epsilon}$, where $n$ is the size of the ground set. In the same context, we show the greedy algorithm does obtain a ratio of $2/n$ under an additional mild additional assumption. Finally, we provide the first nearly linear-time algorithm for maximization of non-monotone submodular functions over $p$-extendible independence systems.