Competitive difference analysis of the cash management problem with uncertain demands
Résumé
The cash management problem with uncertain demands belongs to the class of online problems which are free of any distribution assumptions. The demands can either be globally bounded or interrelated. We consider the performance measure competitive difference. This measure can be interpreted as maximum regret for online problems. The minimization of maximum regret is a major point of interest in the area of combinatorial optimization. We derive new algorithms which aim at minimizing the maximum regret incurred. Their experimental performance is compared to already established solutions, which consider the performance measure competitive ratio. From this comparison we confirm the practicability of our solutions for real-case scenarios. Hence, our algorithms are particularly relevant for risk averse cash managers, who are interested in minimizing their maximum regret while ensuring a solid performance in non-worst case scenarios.
Origine | Fichiers produits par l'(les) auteur(s) |
---|