Informatica

This is a survey of the main achievements in the methodology and theory of stochastic global optimization. It comprises two complimentary directions: global random search and the methodology based on the use of stochastic models about the objective function. The main attention is paid to theoretically substantiated methods and mathematical results proven in the last 25 years.

In this paper, we consider the so-called structured low rank approximation (SLRA) problem as a problem of optimization on the set of either matrices or vectors. Briefly, SLRA is defined as follows. Given an initial matrix with a certain structure (for example, Hankel), the aim is to find a matrix of specified lower rank that approximates this initial matrix, whilst maintaining the initial structure. We demonstrate that the optimization problem arising is typically very difficult; in particular, the objective function is multiextremal even in simple cases. We also look at different methods of solving the SLRA problem. We show that some traditional methods do not even converge to a locally optimal matrix.

The maximization problem for an objective function f given on a feasible region X is considered, where X is a compact subset of Rⁿ and f belongs to a set of continuous multiextremal functions on X can be evaluated at any point x in X without error, and its maximum M=max x∈Xf(x) together with a maximizer x^*(a point x^* in X such that M=f(x^*)) are to be approximated. We consider a class of the global random search methods, underlying an apparatus of the mathematical statistics and generalizing the so-called branch and bound methods.