A Tight Linearization Strategy for Zero-One Quadratic Programming Problems

In this paper, we present a new approach to linearizing zero-one quadratic minimization problem which has many applications in computer science and communications. Our algorithm is based on the observation that the quadratic term of zero-one variables has two equivalent piece-wise formulations, convex and concave cases. The convex piece-wise objective function and/or constraints play a great role in deducing small linearization. Further tight strategies are also discussed.