EU Regional School - Neumaier Seminar
Prof. Dr. Neumaier - Global Optimiztion: Examples, Methods and Software
University of Vienna
Global optimization is the quest for finding the absolutely best solution for a problem where both the goal and the constraints are specified in mathematical terms. In the most general case, the solution may involve discrete variables, continuous variables, functions to be determined, or combinations of these. In some cases, the structure is very favorable for efficient solution techniques, other cases may be extremely hard. Knowing what one can expect from the state of the art is important to make good choices, both in the formulation of an optimization problem and in its solution.
The seminar will
- discuss the typical forms goals and constraints may have, and which kind of information about these help in finding a solution,
- introduce the basic techniques (local search, complete search, stochastic search, heuristics, relaxations) that are currently used in attacking such problems,
- discuss optimality conditions for checking whether a putative solution is globally optimal,
- give an overview over current global optimization software and recommendations for selecting a method of choice,
- illustrate everything with examples from science and engineering.