Implementace Shorova r-algoritmu pro nehladkou optimalizaci

Abstract

By nonsmooth optimization we mean finding global extremes (minimum or maximum) of a continuous function, that is not differentiable at some points in its domain. This problem often appears in mechanics, economics and game theory. This Bachelor thesis focuses on implementation of Shor's r-algorithm and its further optimization to achieve reliable, fast and accurate solution.

The Bachelor thesis consists of the following parts:

1. Convex analysis, 2. Clarke's calculus – generalization of the differential calculus for a nonsmooth function, 3. Nonsmooth optimization, 4. Shor's r-algorithm – introducing, implementation and examples of use, 5. Numerical experiments – dependence of the calculation on the input data.