4.3. The number of constraints
Except the above two factors, the number of constraints in ILP problems is also an important index to sure the complexity of computation. The numerical results are shown in Table 4. The results indicate that the cases with many constraints could reduce lots of constraints. Therefore, the necessary computation and stor age space to find out solutions is decreased. However, for those cases with few constraints, the proposed method gives little help.
5. Conclusions In this paper, we propose an improved B&B method to solve ILP problems efficiently. For this goal, the solution area is divided into smaller subspaces, and then the B&B procedure is used to search the optimum solution within these subspaces. Compared with the original B&B method, we find that it works efficiently while handling ILP problems with many constraints. Although the branching number in our method is almost the same as that in B&B method. We unearth that the proposed method could reduce the number of constraint for cases with many more constraints. Besides, the optimum solution can be found in the first subspace except for rare cases. As demonstrated by numerical examples, 93% of the cases are of this type. While handling the ILP problems with fewer constraints but many variables, it is suggested to use duality concept to change vari- ables by constraints. Then our method can be used to reduce the number of constraints. Therefore, the pro- posed method brings up another thought for solving ILP problems. Although it does not make an enormous breakthrough in this field, it would be a great opportunity to reopen the door and evoke more in further researches in the future
Acknowledgement This research was supported by the National Science Council in Taiwan, ROC, under grants NSC93-2416- H-194-003 and NSC 94-2416-H-194-007
สิ้นสุดการสนทนา
พิมพ์ข้อความ...