A Classification of Connected f -factor Problems inside NP

TL;DR

This paper introduces an algorithm for the connected f-factor problem in graphs, showing its complexity bounds and implications for NP-Intermediate problems under the ETH, with extensions to weighted graphs.

Contribution

The authors develop a new algorithm for connected f-factor existence and minimum weight solutions, establishing complexity bounds related to the ETH and classifying the problem within NP-Intermediate.

Findings

Algorithm runs in polynomial time for constant g(n)

Connected f-factor is unlikely NP-Complete under ETH for certain f

Problem is NP-Intermediate for specific growth rates of f

Abstract

Given an undirected graph G = (V, E) with n vertices, and a function f : V -> N, we consider the problem of finding a connected f -factor in G. In this work we design an algorithm to check for the existence of a connected f -factor, for the case where f (v) >= n/g(n), for all v in V and g(n) is polylogarithmic in n. The running time of our algorithm is O(n^{2g(n)}. As a consequence of this algorithm we conclude that the complexity of connected f -factor for the case we consider is unlikely to be NP-Complete unless the Exponential Time Hypothesis (ETH) is false. Secondly, under the assumption of the ETH, we show that it is also unlikely to be in P for g(n) in O((log n)^{1+eps} ) for any eps> 0. Therefore, our results show that for all eps> 0, connected f -factor for f (v) >= n/O(log n)^{1+eps}) is in NP-Intermediate unless the ETH is false. Further, for any constant c > 0, when g(n) = c, our algorithm for connected f -factor runs in polynomial time. Finally, we extend our algorithm to compute a minimum weight connected f -factor in edge weighted graphs in the same asymptotic time bounds.