— Tuning a system to an operating environment calls for experimentation, and a question that arises naturally is: how many experiments are needed to come up with a system meeting certain performance specifications? This paper is an attempt to answer this fundamental question at a somehow general level. We shall here refer to a set-up where adaptation is done according to a worst-case approach, but many facts established are central to adaptation in general.
Sergio Bittanti, Marco C. Campi, Maria Prandini