In Table 3, in addition to the LP solution, we report the first MIP solution produced by CPLEX during its branch-and-bound (B&B) exploration, as well the best available MIP solution that the solver could obtain within a specified computational limit of 300 CPU seconds. We compare these two heuristic approaches to solving the problem to optimality by reporting the percentage deviation (% Dev.) between the heuristic solution value from the optimal solution value.
However, it is worthwhile mentioning that by considering the first MIP solution obtained by the solver, or by imposing a time-bound (of 300 CPU seconds) on the computational effort, the resulting heuristic solutions respectively sacrificed only 2% and 0.18% of optimality on average for n = 30, while achieving an average computational savings of 98.6% and 93.8%, respectively, as compared with determining optimal solutions.