A New Bound for 3-Satisfiable Maxsat and Its Algorithmic Application

Abstract

Let F be a CNF formula with n variables and m clauses. F is t-satisfiable if for any t clauses in F, there is a truth assignment which satisfies all of them. Lieberherr and Specker (1982) and, later, Yannakakis (1994) proved that in each 3-satisfiable CNF formula at least $\frac{2}{3}$ of its clauses can be satisfied by a truth assignment. Yannakakis’s proof utilizes the fact that $\frac{2}{3}m$ is a lower bound on the expected number of clauses satisfied by a random truth assignment over a certain distribution. A CNF formula F is called expanding if for every subset X of the variables of F, the number of clauses containing variables of X is not smaller than |X|. In this paper we strengthen the $\frac{2}{3}m$ bound for expanding 3-satisfiable CNF formulas by showing that for every such formula F at least $\frac{2}{3}m + \rho n$ clauses of F can be satisfied by a truth assignment, where ρ( > 0.0019) is a constant. Our proof uses a probabilistic method with a sophisticated distribution for truth values. We use the bound $\frac{2}{3}m + \rho n$ and results on matching autarkies to obtain a new lower bound on the maximum number of clauses that can be satisfied by a truth assignment in any 3-satisfiable CNF formula.

We use our results above to show that the following parameterized problem is fixed-parameter tractable and, moreover, has a kernel with a linear number of variables. In 3-S-MaxSat-AE, we are given a 3-satisfiable CNF formula F with m clauses and asked to determine whether there is an assignment which satisfies at least $\frac{2}{3}m + k$ clauses, where k is the parameter. Note that Mahajan and Raman (1999) asked whether 2-S-MaxSat-AE, the corresponding problem for 2-satisfiable formulas, is fixed-parameter tractable. Crowston and the authors of this paper proved in [9] that 2-S-MaxSat-AE is fixed-parameter tractable and, moreover, has a kernel with a linear number of variables. 2-S-MaxSat-AE appears to be easier than 3-S-MaxSat-AE and, unlike this paper, [9] uses only deterministic combinatorial arguments.