arXiv++ Combinatorics

Given an undirected graph $G=(V,E)$ with a nonnegative edge length function and an integer $p$, $0 < p < |V|$, the $p$-centdian problem is to find $p$ vertices (called the {\it centdian set}) of $V$ such that the {\it eccentricity} plus {\it median-distance} is minimized, in which the {\it eccentricity} is the maximum (length) distance of all vertices to their nearest {\it centdian set} and the {\it median-distance} is the total (length) distance of all vertices to their nearest {\it centdian set}. The {\it eccentricity} plus {\it median-distance} is called the {\it centdian-distance}. The purpose of the $p$-centdian problem is to find $p$ open facilities (servers) which satisfy the quality-of-service of the minimum total distance ({\it median-distance}) and the maximum distance ({\it eccentricity}) to their service customers, simultaneously. If we converse the two criteria, that is given the bound of the {\it centdian-distance} and the objective function is to minimize the cardinality of the {\it centdian set}, this problem is called the converse centdian problem. In this paper, we prove the $p$-centdian problem is NP-Complete. Then we design the first non-trivial brute force exact algorithms for the $p$-centdian problem and the converse centdian problem, respectively. Finally, we design two approximation algorithms for both problems.

Given a complete graph $G=(V,E)$, with nonnegative edge costs, two subsets $R \subset V$ and $R^{\prime} \subset R$, a partition $\mathcal{R}=\{R_1,R_2,\ldots,R_k\}$ of $R$, $R_i \cap R_j=φ$, $i \neq j$ and $\mathcal{R}^{\prime}=\{R^{\prime}_1,R^{\prime}_2,\ldots,R^{\prime}_k\}$ of $R^{\prime}$, $R^{\prime}_i \subset R_i$, a clustered Steiner tree is a tree $T$ of $G$ that spans all vertices in $R$ such that $T$ can be cut into $k$ subtrees $T_i$ by removing $k-1$ edges and each subtree $T_i$ spanning all vertices in $R_i$, $1 \leq i \leq k$. The cost of a clustered Steiner tree is defined to be the sum of the costs of all its edges. A clustered selected-internal Steiner tree of $G$ is a clustered Steiner tree for $R$ if all vertices in $R^{\prime}_i$ are internal vertices of $T_i$, $1 \leq i \leq k$. The clustered selected-internal Steiner tree problem is concerned with the determination of a clustered selected-internal Steiner tree $T$ for $R$ and $R^{\prime}$ in $G$ with minimum cost. In this paper, we present the first known approximation algorithm with performance ratio $(ρ+4)$ for the clustered selected-internal Steiner tree problem, where $ρ$ is the best-known performance ratio for the Steiner tree problem.