The complete edge relaxation for binary polynomial optimization

TL;DR

This paper introduces the complete edge relaxation, a stronger convex relaxation for binary polynomial optimization, which extends the multilinear polytope under alpha-acyclic hypergraphs and generalizes triangle inequalities.

Contribution

The paper defines the complete edge relaxation, proves its strength and conditions for extension, and introduces new facet-defining inequalities for alpha-cycles.

Findings

Complete edge relaxation is stronger than existing relaxations.

Extension of the multilinear polytope depends on hypergraph acyclicity.

New facet inequalities for alpha-cycles of length three.

Abstract

We consider the multilinear polytope defined as the convex hull of the feasible region of a linearized binary polynomial optimization problem. We define a relaxation in an extended space for this polytope, which we refer to as the complete edge relaxation. The complete edge relaxation is stronger than several well-known relaxations of the multilinear polytope, including the standard linearization, the flower relaxation, and the intersection of all possible recursive McCormick relaxations. We prove that the complete edge relaxation is an extension of the multilinear polytope if and only if the corresponding hypergraph is alpha-acyclic; i.e., the most general type of hypergraph acyclicity. This is in stark contrast with the widely-used standard linearization which describes the multilinear polytope if and only if the hypergraph is Berge-acyclic; i.e., the most restrictive type of hypergraph acyclicity. We then introduce a new class of facet-defining inequalities for the multilinear polytope of alpha-cycles of length three, which serve as the generalization of the well-known triangle inequalities for the Boolean quadric polytope.