Effective Variable Fixing and Scoring Strategies for Binary Quadratic Programming

We investigate two variable fixing strategies and two variable scoring strategies within a tabu search algorithm, using the unconstrained binary quadratic programming (UBQP) problem as a case study. In particular, we provide insights as to why one particular variable fixing and scoring strategy leads to better computational results than another one. For this purpose, we perform two investigations, the first analyzing deviations from the best known solution and the second analyzing the correlations between the fitness distances of high-quality solutions. We find that one of our strategies obtains the best solutions in the literature for all of the test problems examined.