🤖 AI Summary
This work addresses the confluence problem of node-dominance and edge-dominance rewriting rules in preprocessing for the Minimum Hitting Set problem on hypergraphs, aiming to ensure uniqueness of the reduced hypergraph up to isomorphism. We formalize a graph-isomorphism-preserving rewriting system and, for the first time, rigorously prove that the combined application of both dominance rules is confluent. Consequently, every hypergraph admits a unique isomorphism-class minimal reduced form. This result establishes a sound structural reduction theory for Minimum Hitting Set computation and provides verifiable, correctness-preserving preprocessing guarantees. By eliminating ambiguity in reduction outcomes, it enhances both the reliability and efficiency of downstream algorithms—particularly exact solvers and kernelization procedures—without compromising solution optimality.
📝 Abstract
In this note, we study two rewrite rules on hypergraphs, called edge-domination and node-domination, and show that they are confluent. These rules are rather natural and commonly used before computing the minimum hitting sets of a hypergraph. Intuitively, edge-domination allows us to remove hyperedges that are supersets of another hyperedge, and node-domination allows us to remove nodes whose incident hyperedges are a subset of that of another node. We show that these rules are confluent up to isomorphism, i.e., if we apply any sequences of edge-domination and node-domination rules, then the resulting hypergraphs can be made isomorphic via more rule applications. This in particular implies the existence of a unique minimal hypergraph, up to isomorphism.