Optimization using metaheuristics

• Solve complex and/or large optimization problems using meta-heuristics.
• Identify what are the most appropriate meta-heuristics for a specific optimization problem.
• Device a problem representation that allows a specific problem to be optimized using a meta-heuristic.
• Specialize a meta-heuristic such that it can be applied to a specific optimization problem.
• Implement a metaheuristic such that it can be applied to a specific optimization problem.
• Test a meta-heuristic such that its efficiency can be reliably evaluated.
• Describe in written form the adaptation of a meta-heuristic to a specific problem formulation.
• Understand the difference between the diversification and intensification strategies of different meta-heuristic frameworks.

Many interesting optimization problems cannot be solved using standard solvers dues to their size and complexity.

A pragmatic approach to optimization is to use tailored computer algorithms to test a large number of solutions in order to find a good feasible solution. Such algorithms are called heuristics. Those algorithms do not guarantee to find optimal solutions but attempt to identify good solutions. There exist a number of more general algorithmic frameworks which can be applied to a wide variety of optimization problems. Those are the so-called meta-heuristics. In this course a number of these meta-heuristics will be presented:

– Simulated Annealing
– Genetic Algorithms/Evolutionary Algorithms
– TABU search
– GRASP
– ALNS
– ILS

As this field is constantly developing, the content of the course is constantly updated. The exercises in the course will be in the computing language Julia, hence prior experience with Julia is an advantage.

42101, Introduction to Operations Research and programming experience. The course heavily relies on your ability to program.

Lectures, excercises and project work.