Approximating the Stable Model Semantics is Hard

In this paper we investigate the complexity of problems concerned with approximating the stable model semantics. We show that under rather weak assumptions it is NP-hard to decide whether the size of a polynomially computable approximation is within a constant factor from the size of the intersection (union) of stable models of a program. We also show that unless P=NP, no approximation exists that uniformly bounds the intersection (union) of stable models.