Solving multimodal problems via multiobjective techniques with Application to phase equilibrium detection

Abstract

Abstract-For solving multimodal problems by means of evolutionary algorithms, one often resorts to multistarts or niching methods. The latter approach the question: 'What is elsewhere?' by an implicit second criterion in order to keep populations distributed over the search space. Induced by a practical problem that appears to be simple but is not easily solved, a multiobjective algorithm is proposed for solving multimodal problems. It employs an explicit diversity criterion as second objective. Experimental comparison with standard methods suggests that the multiobjective algorithm is fast and reliable and that coupling it with a local search technique is straightforward and leads to enormous quality gain. The combined algorithm is still fast and may be especially valuable for practical problems with costly target function evaluations.

A multiobjective approach for finding equivalent inverse images of Pareto-optimal objective vectors

“…We argue that helper functions are useful if they encode additional problem knowledge that can be exploited by by the optimization method. In case of a real-world problem in the field of thermodynamics [15] we were able to support this point of view: our helper function enlarged the basins of the local optima and it encoded implicitly which area of the search space should be avoided. This course of action has facilitated the detection of the desired optima significantly.…”

Optimization of biogas production with computational intelligence a comparative study

“…SPOT was successfully applied to numerous optimization algorithms, especially in the field of evolutionary computation, e.g., evolution strategies, particle swarm optimization or genetic programming. It was applied in various domains, e.g., machine engineering, the aerospace industry, bioinformatics, CI and games as well as in fundamental research (Beume et al, 2008;Henrich et al, 2008;Lucas and Roosen, 2009;Preuss et al, 2007;Fialho et al, 2009;Fober et al, 2009;Stoean et al, 2009;Hutter et al, 2010).…”

SMS-EMOA – Effektive evolutionäre Mehrzieloptimierung (SMS-EMOA – Effective Evolutionary Multiobjective Optimization)

“…Deshalb haben wir problemspezifisches Wissen bei der Problemformulierung geeignet eingearbeitet [22]: Wir wissen bereits, dass die nicht gewünschten, trivialen Optima auf der Raumdiagonalen liegen. Also könnte man mit Hilfe von Straffunktionen die Nähe einer Lösung zur Raumdiagonalen bestrafen, um den Algorithmus von den trivialen Lö-sungen fortzuleiten.…”