We show, through an example in surface-plasmons assisted nano-lithography, the great influence of the definition of the objective function on the quality of the solutions obtained after optimization. We define the visibility and the contrast of a surface-plasmons interference pattern as possible objective functions that will serve to characterize the geometry of a nano-structure. We optimize them with an Elitist Evolution Strategy and compare, by means of some numerical experiments, their effects on the geometrical parameters found. The maximization of the contrast seems to provide solutions more stable than those obtained when the visibility is maximized. Also, it seems to avoid the lack-of-uniqueness problems resulting from the optimization of the visibility.
展开▼
