Information complexity of functional optimization problems and their approximation schemes
AbstractFunctional optimization is investigated using tools from information-based complexity. In such optimization problems, a functional has to be minimized with respect to admissible solutions belonging to an infinite-dimensional space of functions. This context models tasks arising in optimal control, systems identification, machine learning, time-series analysis, etc. The solution via variable-basis approximation schemes, which provide a sequence of nonlinear programming problems approximating the original functional one, is considered. Also for such problems, the information complexity is estimated.