Global Optimization Algorithms in Naval Hydrodynamics

Real-life optimization problems are much more complicated than the analytical test functions usually adopted for estimating the performances of new optimization algorithms. This simple statement may sound like an obvious objection to most of ship designers as well as an outrageous criticism to the mathematical community. Anyway it is evident that for a numerical optimization procedure to be useful in the design spiral it is fundamental to understand the peculiar features of the design problem and choose the more suitable numerical technique among a bunch of methodologies. Unfortunately, although the number of paper related to numerical ship design is rapidly growing, there is still the need of testing well assessed algorithms in specific ship design applications, especially if talk about Global Optimization techniques. In this paper a certain number of Global Optimization algorithms are presented and discussed, showing results on a real ship design application. The focus is on the development of deterministic methods, which are preferred for their good theoretical properties and for the reduced number of objective function evaluations required, a fundamental property any method should have to deal with numerical optimization of ships.