The complexity of finding and enumerating optimal subgraphs to represent spatial correlation

TL;DR

This paper investigates the computational complexity of optimizing subgraphs to model spatial correlation, proving NP-hardness, and introduces two parameterized algorithms for exact enumeration under specific constraints.

Contribution

It establishes the NP-hardness of the subgraph optimization problem and presents two efficient parameterized algorithms for exact enumeration in restricted cases.

Findings

The problem is NP-hard unless P=NP.

Two parameterized algorithms solve the problem exactly.

Algorithms enable efficient enumeration considering spatial correlation uncertainty.

Abstract

Understanding spatial correlation is vital in many fields including epidemiology and social science. Lee, Meeks and Pettersson (Stat. Comput. 2021) recently demonstrated that improved inference for areal unit count data can be achieved by carrying out modifications to a graph representing spatial correlations; specifically, they delete edges of the planar graph derived from border-sharing between geographic regions in order to maximise a specific objective function. In this paper we address the computational complexity of the associated graph optimisation problem. We demonstrate that this problem cannot be solved in polynomial time unless P = NP; we further show intractability for two simpler variants of the problem. We follow these results with two parameterised algorithms that exactly solve the problem. Both of these solve not only the decision problem, but also enumerate all solutions with polynomial time precalculation, delay, and postcalculation time in respective restricted settings. For this problem, efficient enumeration allows the uncertainty in the spatial correlation to be utilised in the modelling. The first enumeration algorithm utilises dynamic programming on a tree decomposition, and has polynomial time precalculation and linear delay if both the treewidth and maximum degree are bounded. The second algorithm is restricted to problem instances with maximum degree three, as may arise from triangulations of planar surfaces, but can output all solutions with FPT precalculation time and linear delay when the maximum number of edges that can be removed is taken as the parameter.