跳到主要导航 跳到搜索 跳到主要内容

Monotone submodular maximization over the bounded integer lattice with cardinality constraints

  • East China Normal University
  • Wuhan University
  • University of Texas at Dallas

科研成果: 期刊稿件文章同行评审

摘要

We consider the problem of maximizing monotone submodular function over the bounded integer lattice with a cardinality constraint. Function f: ℤ+E → R+ is submodular over integer lattice if f(x) + f(y) ≥ f(x ∨ y) + f(x ∧ y), ∀ x,y ∈ ℤ+E, where ∨ and ∧ represent elementwise maximum and minimum, respectively. Let B ∈ ℤ+E, and k ∈ ℤ+, we study the problem of maximizing submodular function f(x) with constraints 0 ≤x ≤B and x(E) ≤ k. A random greedy (1 -1/e)-approximation algorithm and a deterministic 1 e-approximation algorithm are proposed in this paper. Both algorithms work in value oracle model. In the random greedy algorithm, we assume the monotone submodular function satisfies diminishing return property, which is not an equivalent definition of submodularity on integer lattice. Additionally, our random greedy algorithm makes Ο((|E| + 1)· k) value oracle queries and deterministic algorithm makes Ο(|E|· B Ο · k3) value oracle queries.

源语言英语
文章编号1950075
期刊Discrete Mathematics, Algorithms and Applications
11
6
DOI
出版状态已出版 - 1 12月 2019

指纹

探究 'Monotone submodular maximization over the bounded integer lattice with cardinality constraints' 的科研主题。它们共同构成独一无二的指纹。

引用此