An approximation algorithm for the k-prize-collecting multicut on a tree problem
Introduction
The main focus of this paper is the k-prize-collecting multicut on a tree (k-PCM(T)) problem. In the k-PCM(T) problem, we are given an undirected tree with non-negative costs for the edges , a set of m distinct pairs of vertices , with nonnegative penalty costs for the pairs , and a parameter k at most m. For , the pair is said to be separated by the edge set if it is not contained in a single connected component of , in other words, the removal of M disconnects and . Our goal is to find a multicut M that separates at least k pairs of vertices such that the total cost, including the edge cost of the multicut M and the penalty cost of the pairs not separated by M, is minimized.
The k-PCM(T) problem contains as special cases the multicut on a tree, the prize-collecting multicut on a tree and the partial multicut on a tree problems. For the multicut on a tree problem, Garg et al. [3] presented a prime-dual algorithm that constructs a feasible solution whose cost is at most twice the optimum. For the prize-collecting multicut on a tree problem, Levin and Segev [7] proved that it is equivalent to the multicut on a tree problem by modifying the original tree and set of pairs, and gave a 2-approximation algorithm. For the partial multicut on a tree problem, Levin and Segev [7] presented an -approximation algorithm by using the Lagrangian relaxation technique, where ε is any fixed positive number.
An algorithm for a prize-collecting problem is called Lagrangian-multiplier preserving (LMP) [9], if it satisfies the so-called Lagrangian multiplier preserving property, that is it satisfies the initial objective cost plus r times the penalty cost is at most r times the dual objective where r is some desired approximation ratio. LMP algorithms have been designed for the k-median problem [5], k-MST problem [2], k-Steiner tree [1], k-prize-collecting Steiner tree [4] and partial covering [6] problems.
Liu et al. [8] introduced the multicut on a tree problem with submodule penalties, which is a generalization of the prize-collecting multicut on a tree problem, and presented a 3-approximation algorithm based on the primal-dual method. Motivated by the above work, in this paper, we first restate the algorithm as an algorithm for the prize-collecting multicut on a tree problem and show that it is LMP, and then present an approximation algorithm with an approximate ratio of for the k-PCM(T) problem via the methods of primal-dual and Lagrangean relaxation, where ε is any fixed positive number.
This paper is organized as follows: In Section 2, we recall the prize-collecting multicut on a tree problem. In Section 3, we present the LP relaxation and dual program of the k-PCM(T) problem. In Section 4, we describe our algorithm for the k-PCM(T) problem. In Section 5, we analyze the approximate ratio of our algorithm.
Section snippets
Preliminaries
In this section we first recall the prize-collecting multicut on a tree problem. Then we introduce the algorithm given by Liu et al. [8] and prove that it is LMP.
In the prize-collecting multicut on a tree problem, we are given an undirected tree , a set of m distinct pairs of vertices . Every edge in E has a nonnegative cost . Every pair in P has a nonnegative penalty cost . Our goal is to find a set of edges that minimizes the cost of M plus the
The LP relaxation and its dual program
Recall that for each , is the unique path from to in the tree. Then the k-PCM(T) problem can be formulated as the following integer program:
The variable indicates that the edge e is included in the solution. The variable indicates that the pair is penalized. Thus the objective function composes of the edge cost and the unseparated pairs penalty cost. The
An algorithm for the k-PCM(T) problem
In this section, we describe our algorithm for the k-PCM(T) problem. Let OPT denote the cost of an optimal solution to the k-PCM(T) problem and let be an accuracy parameter. Set and .
Assumption 4.1 The cost of each edge is at most .
Levin and Segev [7] established an -approximation algorithm for the partial multcut on a tree problem under Assumption 4.1. Our paper employs Assumption 4.1 when we discuss the k-PCM(T) problem.
Based on the comments mentioned at
Theoretical analysis
In this section, we analyze the approximate ratio of Algorithm 2. There are four termination possibilities of Algorithm 2. We consider the cases respectively.
Recall that OPT denotes the cost of an optimal solution to the k-PCM(T) problem. Assume that . Otherwise, and the optimal solution is a set of some zero-cost edges. We can easy check such a solution exists before we run the algorithm .
Lemma 5.1 Suppose that Algorithm 2 is terminated at Line 4 with . Then the multicut is a
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgement
The authors would like to thank Professor DingZhu Du for his many valuable advices during their study of approximation algorithm. This work was supported by the NSF of China (No. 11971146), the NSF of Hebei Province of China (No. A2019205089, No. A2019205092), Hebei Province Foundation for Returnees (CL201714) and Overseas Expertise Introduction Program of Hebei Auspices (25305008).
References (9)
- et al.
Partial multicuts in trees
Theor. Comput. Sci.
(2006) - et al.
Approximate k-MSTs and k-Steiner trees via the primal-dual method and Lagrangean relaxation
A 3-approximation for the minimum tree spanning k vertices
- et al.
Primal-dual approximation algorithms for integral flow and multicut in trees
Algorithmica
(1997)
Cited by (5)
Frameworks for Nonclairvoyant Network Design with Deadlines or Delay
2023, Leibniz International Proceedings in Informatics, LIPIcsAn Approximation Algorithm for P-prize-collecting Set Cover Problem
2023, Journal of the Operations Research Society of ChinaThe B -Prize-Collecting Multicut Problem in Paths, Spider Graphs and Rings
2023, International Journal of Foundations of Computer ScienceA primal-dual approximation algorithm for the k-prize-collecting minimum power cover problem
2022, Optimization LettersAn Approximation Algorithm for the B-prize-collecting Multicut Problem in Trees
2022, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)