Clustering with Local Restrictions

TL;DR

This paper introduces a family of graph clustering problems with local constraints, providing polynomial-time algorithms for fixed parameters and fixed-parameter tractability results for specific functions.

Contribution

It establishes polynomial-time solvability for the general problem with fixed q and demonstrates fixed-parameter tractability for three specific functions.

Findings

Polynomial-time solvability for fixed q with arbitrary monotone functions.

Fixed-parameter tractability for specific functions like cluster size and nonedges.

Efficient algorithms for natural clustering problems with local restrictions.

Abstract

We study a family of graph clustering problems where each cluster has to satisfy a certain local requirement. Formally, let $\mu$ be a function on the subsets of vertices of a graph $G$. In the $(\mu,p,q)$-PARTITION problem, the task is to find a partition of the vertices into clusters where each cluster $C$ satisfies the requirements that (1) at most $q$ edges leave $C$ and (2) $\mu(C)\le p$. Our first result shows that if $\mu$ is an {\em arbitrary} polynomial-time computable monotone function, then $(\mu,p,q)$-PARTITION can be solved in time $n^{O(q)}$, i.e., it is polynomial-time solvable {\em for every fixed $q$}. We study in detail three concrete functions $\mu$ (the number of vertices in the cluster, number of nonedges in the cluster, maximum number of non-neighbors a vertex has in the cluster), which correspond to natural clustering problems. For these functions, we show that $(\mu,p,q)$-PARTITION can be solved in time $2^{O(p)}\cdot n^{O(1)}$ and in time $2^{O(q)}\cdot n^{O(1)}$ on $n$-vertex graphs, i.e., the problem is fixed-parameter tractable parameterized by $p$ or by $q$.