(1) Dipartimento di Informatica, Universitá di Torino, Corso Svizzera, 185 10149 Torino, Italy;(2) Department of Statistics, Macquarie University, North Ryde, 2109, NSW, Australia
Abstract:
The purposes of this discussion paper are twofold. First, features of an objective function landscape which provide barriers to rapid finding of the global optimum are described. Second, stochastic algorithms are discussed and their performance examined, both theoretically and computationally, as the features change. The paper lays a foundation for the later findings paper.