Abstract:We consider two classic problems: maximum coverage and monotone submodular maximization subject to a cardinality constraint. [Nemhauser--Wolsey--Fisher '78] proved that the greedy algorithm provides an approximation of $1-1/e$ for both problems, and it is known that this guarantee is tight ([Nemhauser--Wolsey '78; Feige '98]). Thus, one would naturally assume that everything is resolved when considering the approximation guarantees of these two problems, as both exhibit the same tight approximation and hardness. In this work we show that this is not the case, and study both problems when the cardinality constraint is a constant fraction $c \in (0,1]$ of the ground set. We prove that monotone submodular maximization subject to a cardinality constraint admits an approximation of $1-(1-c)^{1/c}$; This approximation equals $1$ when $c=1$ and it gracefully degrades to $1-1/e$ when $c$ approaches $0$. Moreover, for every $c=1/s$ (for any integer $s \in \mathbb{N}$) we present a matching hardness. Surprisingly, for $c=1/2$ we prove that Maximum Coverage admits an approximation of $0.7533$, thus separating the two problems. To the best of our knowledge, this is the first known example of a well-studied maximization problem for which coverage and monotone submodular objectives exhibit a different best possible approximation.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is about the approximation algorithms of the Maximum Coverage (MC) problem and the Monotone Submodular Maximization (SM) problem under cardinality constraints and their performance differences. Specifically: 1. **Approximability of classical problems**: Nemhauser, Wolsey and Fisher proved in their 1978 work that the greedy algorithm can provide an approximation guarantee of $1-\frac{1}{e}$ for both problems, and this result is tight. Therefore, it is natural for people to think that these two problems have been completely solved in terms of approximation guarantees. 2. **New research findings**: The research in this paper shows that when the cardinality constraint is a constant proportion $c\in(0, 1]$ of the ground set, the performance of these two problems is not the same. By studying the approximation ratio in this specific case, the author discovers the differences between the two problems. 3. **Main contributions**: - For the Monotone Submodular Maximization (SM) problem, the author proves that its approximation ratio is $1-(1 - c)^{1/c}$, and for some specific values of $c$ (such as $c = \frac{1}{s}$), gives the matching hardness results. - For the Maximum Coverage (MC) problem, the author proves that when $c=\frac{1}{2}$, an approximation ratio of at least $0.7533$ can be achieved, which is strictly better than the hardness result of SM (i.e., $\rho(1/2)=\frac{3}{4}$). This is the first time that a difference has been found in the best possible approximation ratios between the maximization of the coverage function and the maximization of the general monotone submodular function. In summary, the main goal of this paper is to reveal and prove that there are significant differences in the approximation performance of the Maximum Coverage problem and the Monotone Submodular Maximization problem under specific cardinality constraints, thus challenging the previous assumption that the two are the same in terms of approximation guarantees. ### Formula organization - **Approximation ratio of SM**: \[ 1-(1 - c)^{1/c} \] - **Approximation ratio of MC** (when $c = \frac{1}{2}$): \[ 0.7533 \] - **Definition of $\rho(c)$**: \[ \rho(c)= \begin{cases} 1-(1 - c)^{1/c}&\text{if }c = \frac{1}{s}\text{ for some integer }s\in\mathbb{N}\\ 1-\sigma(\alpha^*, s)&\text{otherwise} \end{cases} \] where, \[ \sigma(\alpha, m)=\left(1-\alpha c-\frac{1-\alpha}{m}\right)^m \] $\alpha^*$ is the unique solution of the equation $\sigma(\alpha^*, s)=\sigma(\alpha^*, s + 1)$. These formulas show how the approximation ratios of the two problems change under different conditions and reveal the differences between them.

Separating Coverage and Submodular: Maximization Subject to a Cardinality Constraint

Set multi-covering via inclusion-exclusion

Non-submodular maximization with a decomposable objective function

Practical $0.385$-Approximation for Submodular Maximization Subject to a Cardinality Constraint

The Cost of Consistency: Submodular Maximization with Constant Recourse

Maximization of nonsubmodular functions under multiple constraints with applications

Submodular Maximization Beyond Non-negativity: Guarantees, Fast Algorithms, and Applications

A Unified Continuous Greedy Algorithm for Submodular Maximization

Fair Submodular Cover

Nonmonontone submodular maximization under routing constraints

Constrained Non-Monotone Submodular Maximization: Offline and Secretary Algorithms

Discretely Beyond $1/e$: Guided Combinatorial Algorithms for Submodular Maximization

Approximating Low-Dimensional Coverage Problems

Maximizing a Nonnegative, Monotone, Submodular Function Constrained to Matchings

Monotone Submodular Multiway Partition

Budget-Smoothed Analysis for Submodular Maximization

Practical and Parallelizable Algorithms for Non-Monotone Submodular Maximization with Size Constraint

Comparing Apples and Oranges: Query Tradeoff in Submodular Maximization

Submodular Maximization in Exactly $n$ Queries

Approximation Algorithms for Maximization of -Submodular Function Under a Matroid Constraint