Holant* Dichotomy on Domain Size 3: A Geometric Perspective

Abstract

Holant problems are a general framework to study the computational complexity of counting problems. It is a more expressive framework than counting constraint satisfaction problems (CSP) which are in turn more expressive than counting graph homomorphisms (GH). In this paper, we prove the first complexity dichotomy of $\mathrm{Holant}_3(\mathcal{F})$ where $\mathcal{F}$ is an arbitrary set of symmetric, real valued constraint functions on domain size $3$. We give an explicit tractability criterion and prove that, if $\mathcal{F}$ satisfies this criterion then $\mathrm{Holant}_3(\mathcal{F})$ is polynomial time computable, and otherwise it is \#P-hard, with no intermediate cases. We show that the geometry of the tensor decomposition of the constraint functions plays a central role in the formulation as well as the structural internal logic of the dichotomy.