Approximation Algorithms for Flexible Graph Connectivity

TL;DR

This paper develops improved approximation algorithms for the Flexible Graph Connectivity problem, achieving better ratios for various parameters, and relates these to capacitated network design problems.

Contribution

It introduces new approximation algorithms with improved ratios for the $(p,q)$-Flexible Graph Connectivity problem, extending previous results and connecting to capacitated network design.

Findings

A 2-approximation for the (1,1)-FGC problem via reduction to rooted 2-arborescence.

Extension of the 2-approximation to (1,k)-FGC problems.

A 16/11-approximation for the unweighted (1,1)-FGC problem.

Abstract

We present approximation algorithms for several network design problems in the model of Flexible Graph Connectivity (Adjiashvili, Hommelsheim and M\"uhlenthaler, "Flexible Graph Connectivity", Math. Program. pp. 1-33 (2021), and IPCO 2020: pp. 13-26). Let $k\geq 1$, $p\geq 1$ and $q\geq 0$ be integers. In an instance of the $(p,q)$-Flexible Graph Connectivity problem, denoted $(p,q)$-FGC, we have an undirected connected graph $G = (V,E)$, a partition of $E$ into a set of safe edges $S$ and a set of unsafe edges $U$, and nonnegative costs $c: E\to\Re$ on the edges. A subset $F \subseteq E$ of edges is feasible for the $(p,q)$-FGC problem if for any subset $F'$ of unsafe edges with $|F'|\leq q$, the subgraph $(V, F \setminus F')$ is $p$-edge connected. The algorithmic goal is to find a feasible solution $F$ that minimizes $c(F) = \sum_{e \in F} c_e$. We present a simple $2$-approximation algorithm for the $(1,1)$-FGC problem via a reduction to the minimum-cost rooted $2$-arborescence problem. This improves on the $2.527$-approximation algorithm of Adjiashvili et al. Our $2$-approximation algorithm for the $(1,1)$-FGC problem extends to a $(k+1)$-approximation algorithm for the $(1,k)$-FGC problem. We present a $4$-approximation algorithm for the $(p,1)$-FGC problem, and an $O(q\log|V|)$-approximation algorithm for the $(p,q)$-FGC problem. Finally, we improve on the result of Adjiashvili et al. for the unweighted $(1,1)$-FGC problem by presenting a $16/11$-approximation algorithm. The $(p,q)$-FGC problem is related to the well-known Capacitated $k$-Connected Subgraph problem (denoted Cap-k-ECSS) that arises in the area of Capacitated Network Design. We give a $\min(k,2 u_{max})$-approximation algorithm for the Cap-k-ECSS problem, where $u_{max}$ denotes the maximum capacity of an edge.