Abstract (EN):
In operational problems under uncertainty, the explicit consideration of risk concerns is a fundamental but often overlooked issue. Recently, however, there has been an increasing attention to this topic, in both literature and practice. With this background as motivation, we propose in this paper an approach based on multiobjective metaheuristics and sample approximations to tackle mean-variance models of stochastic combinatorial optimisation (SCO) problems. The potential of the approach, as a generic and efficient algorithm for the identification of good quality approximations to mean-risk efficient sets in SCO problems, is validated with an application to the static stochastic knapsack problem, involving a computational study whose results are discussed. The approach is able to handle several difficulties inherent to these problems, and the computational results, obtained with a straightforward implementation, are promising.
Language:
Portuguese
Type (Professor's evaluation):
Scientific