Continuous Optimization 2024

Optimization methods or algorithms seek for a state of a system that is optimal with respect to an objective function. Depending on the properties of the objective function, different strategies may be used to find such an optimal state (or point). The fundamental knowledge of the classes of functions that can be optimized, the properties of available optimization strategies, and the properties of the optimal points are crucial for appropriately modeling practical real world problems as optimization problems. An exact model of the real world that cannot be solved is as useless as a too simplistic model that can be solved easily.

This lecture introduces the basic algorithms, and concepts and analysis tools for several fundamental classes of continuous optimization problems and algorithms. The lecture covers the basics of generic Descent Methods, Gradient Descent, Newton Method, Quasi-Newton Method, Gauss-Newton Method, Conjugate Gradient, linear programming, non-linear programming, as well as optimality conditions for unconstrained and constrained optimization problems. These may be considered as the classical topics of continuous optimization. Some of these methods will be implemented and explored for practical problems in the tutorials.

After taking this course, students will have an overview of classical optimization methods and analysis tools for continuous optimization problems, which allows them to model and solve practical problems. Moreover, in the tutorials, some experience will be gained to implement and numerically solve practical problems.