Decomposições Lagrangeanas para o problema de programação quadrática binária irrestrita

Abstract

The Unconstrained Binary Quadratic Programming Problem - PQ is a classical non-linear problem of optimizing a quadratic objective by suitable choice of binary decisions variables. This paper proposes new alternatives of Lagrangean decomposition to find bounds for PQ. The presented methods treat a mixed binary linear version (PQL) of PQ with constraints represented by a graph. This graph is partitioned in clusters of vertices forming a dual problem that is solved by a subgradient algorithm. The subproblems formed by the generated subgraphs are solved by the CPLEX. Computational experiments consider with a data set formed by several difficult instances with different characteristics. The results show the efficiency of the proposed methods over traditional Lagrangean relaxations and other methods found in literature.