Nonlinear biobjective optimization: improvements to interval branch & bound algorithms

Abstract

Interval based solvers are commonly used for solving single-objective nonlinear optimization problems. Their reliability and increasing performance make them useful when proofs of infeasibility and/or certification of solutions are a must. On the other hand, there exist only a few approaches dealing with nonlinear optimization problems, when they consider multiple objectives. In this paper, we propose a new interval branch & bound algorithm for solving nonlinear constrained biobjective optimization problems. Although the general strategy is based on other works, we propose some improvements related to the termination criteria, node selection, upperbounding and discarding boxes using the non-dominated set. Most of these techniques use and/or adapt components of IbexOpt, a state-of-the-art interval-based single-objective optimization algorithm. The code of our plugin can be found in our git repository (https://github.com/INFPUCV/ibex-lib/tree/master/plugins/optim-mop).

This is a preview of subscription content, access via your institution.

Notes

1. An interval $${\varvec{x}}_i=[\underline{x_i},\overline{x_i}]$$ defines the set of reals $$x_i$$, such that $$\underline{x_i} \le x_i \le \overline{x_i}$$. A box $${\varvec{x}}$$ is a Cartesian product of intervals $${\varvec{x}}_1 \times \cdots \times {\varvec{x}}_i \times \cdots \times {\varvec{x}}_n$$.

2. We implemented, however, the other three simple discarding tests also used in [11], i.e., the dominance test and the simple and generalized monotonicity tests proposed in [8].

3. For the CTP instances we consider a function $$\varphi =1+\sum \nolimits _{i=2}^n x_i$$, otherwise we obtain only one non-dominated solution: $$x=(0,0,...)$$, $$y=(0,1)$$. For the CF3 instances we consider a size $$n=5$$ instead of $$n=2$$, because with $$n=2$$, we obtain the undetermined term 2 / 0 in the first objective.

4. The hypervolume [27] is a quality indicator corresponding to the area of the region dominated by the set of vectors returned by the solver, $$\mathcal {Y}'$$, and limited by a reference vector. The optimal hypervolume value corresponds thus to the area of the region dominated by the set of non-dominated vectors $$\mathcal {Y}^*$$.

5. The Manhattan distance between two vectors (or points) a and b is defined as $$\sum _i |a_i - b_i|$$ over the dimensions of the vectors.

6. The larger value obtained for $$N_{max}$$ when $$c_y$$ is used, can be explained by the fact that, strategies using $$c_y$$ generally process a smaller number of boxes (in Table 3, compare the number of processed boxes reported by each strategy), thus it is convenient to generate more feasible solutions in each box.

References

1. Marler, R.T., Arora, J.S.: Survey of multi-objective optimization methods for engineering. Struct. Multidiscipl. Optim. 26(6), 369–395 (2004)

2. Deb, K.: Multi-objective optimization. In: Search Methodologies. Springer, pp. 403–449 (2014)

3. Przybylski, A., Gandibleux, X.: Multi-objective branch and bound. Eur. J. Oper. Res. 260(3), 856–872 (2017)

4. Redondo, J.L., Fernández, J., Ortigosa, P.M.: FEMOEA: a fast and efficient multi-objective evolutionary algorithm. Math. Methods Oper. Res. 85(1), 113–135 (2017)

5. Coello, C.A.C., Lamont, G.B., Van Veldhuizen, D.A., et al.: Evolutionary Algorithms for Solving Multi-objective Problems, vol. 5. Springer, Berlin (2007)

6. Ruetsch, G.: An interval algorithm for multi-objective optimization. Struct. Multidiscip. Optim. 30(1), 27–37 (2005)

7. Fernández, J., Tóth, B.: Obtaining an outer approximation of the efficient set of nonlinear biobjective problems. J. Glob. Optim. 38(2), 315–331 (2007)

8. Fernández, J., Tóth, B.: Obtaining the efficient set of nonlinear biobjective optimization problems via interval branch-and-bound methods. Comput. Optim. Appl. 42(3), 393–419 (2009)

9. Kubica, B. J., Woźniak, A.: Tuning the interval algorithm for seeking pareto sets of multi-criteria problems. In: International Workshop on Applied Parallel Computing. Springer, pp. 504–517 (2012)

10. Martin, B., Goldsztejn, A., Granvilliers, L., Jermann, C.: On continuation methods for non-linear bi-objective optimization: towards a certified interval-based approach. J. Glob. Optim. 64(1), 3–16 (2016)

11. Martin, B., Goldsztejn, A., Granvilliers, L., Jermann, C.: Constraint propagation using dominance in interval branch & bound for nonlinear biobjective optimization. Eur. J. Oper. Res. 260(3), 934–948 (2017)

12. Niebling, J., Eichfelder, G.: A branch-and-bound based algorithm for nonconvex multiobjective optimization. Preprint-Series of the Institute for Mathematics (2018)

13. Goldsztejn, A., Domes, F., Chevalier, B.: First order rejection tests for multiple-objective optimization. J. Glob. Optim. 58(4), 653–672 (2014)

14. Araya, I., Trombettoni, G., Neveu, B., Chabert, G.: Upper bounding in inner regions for global optimization under inequality constraints. J. Glob. Optim. 60(2), 145–164 (2014)

15. Trombettoni, G., Araya, I., Neveu, B., Chabert, G.: Inner regions and interval linearizations for global optimization. In: AAAI Conference on Artificial Intelligence, pp. 99–104 (2011)

16. Araya, I., Neveu, B.: lsmear: a variable selection strategy for interval branch and bound solvers. J. Glob. Optim. 71(3), 483–500 (2018)

17. Benhamou, F., Goualard, F., Granvilliers, L., Puget, J.-F.: Revising hull and box consistency. In: International Conference on Logic Programming, Citeseer (1999)

18. Neveu, B., Trombettoni, G., et al.: Adaptive constructive interval disjunction. In: International Conference on Tools with Artificial Intelligence (ICTAI), pp. 900–906 (2013)

19. Ninin, J., Messine, F., Hansen, P.: A reliable affine relaxation method for global optimization. 4OR 13(3), 247–277 (2015)

20. Araya, I., Trombettoni, G., Neveu, B.: A contractor based on convex interval taylor. In: Integration of AI and OR Techniques in Contraint Programming for Combinatorial Optimzation Problems. Springer, pp. 1–16 (2012)

21. Martin, B.: Rigorous algorithms for nonlinear biobjective optimization. Ph.D. dissertation, Université de Nantes (2014)

22. Tóth, B., Fernández, J.: Interval Methods for Single and Bi-objective Optimization Problems-Applied to Competitive Facility Location Problems. Lambert Academic Publishing, Saarbrücken (2010)

23. Chabert, G., Jaulin, L.: Contractor programming. Artif. Intell. 173, 1079–1100 (2009)

24. Trombettoni, G., Chabert, G.: Constructive interval disjunction. In: Principles and Practice of Constraint Programming (CP 2007). Springer, pp. 635–650 (2007)

25. Csendes, T., Ratz, D.: Subdivision direction selection in interval methods for global optimization. SIAM J. Numer. Anal. 34(3), 922–938 (1997)

26. Moore, R.E.: Interval Analysis. Prentice-Hall, Englewood Cliffs, NJ (1966)

27. Zitzler, E., Thiele, L.: Multiobjective optimization using evolutionary algorithms: a comparative case study. In: International Conference on Parallel Problem Solving from Nature. Springer, pp. 292–301 (1998)

28. Neveu, B., Trombettoni, G., Araya, I.: Node selection strategies in interval branch and bound algorithms. J. Glob. Optim. 64(2), 289–304 (2016)

Acknowledgements

This work is supported by the Fondecyt Project 1160224.

Author information

Authors

Corresponding author

Correspondence to Ignacio Araya.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

Araya, I., Campusano, J. & Aliquintui, D. Nonlinear biobjective optimization: improvements to interval branch & bound algorithms. J Glob Optim 75, 91–110 (2019). https://doi.org/10.1007/s10898-019-00768-z

• Accepted:

• Published:

• Issue Date:

• DOI: https://doi.org/10.1007/s10898-019-00768-z

Keywords

• Interval methods
• Branch & bound
• Multiobjective optimization