Decision-Making Under Uncertainty
Overall Course Objectives
To provide the student with the skills to tackle decision-making problems with uncertain information within different application fields (such as energy systems, finance, logistics) by making use of techniques of optimization under uncertainty.
Learning Objectives
- Explain techniques for decision-making under uncertainty (stochastic programming, robust optimization)
- Formulate decision-making problems in different applications (energy systems, electricity markets, finance, logistics) as mathematical programs.
- Apply scenario generation techniques to handle uncertain data as input to the decision-making process.
- Apply a technique of optimization under uncertainty to a new planning problem
- Solve optimization problems including uncertainty using appropriate programming languages and software
- Analyze and interpret the solution to an optimization problem in relation to the planning problem and with regards to quality.
- Debate the different techniques of optimization under uncertainty in terms of uncertainty modeling, objective function, degree of conservatism and problem structure
- Explain the concepts of decomposition techniques and heuristic solution methods as techniques to solve large-scale optimization problems
- Assess and judge the best technique of optimization under uncertainty to be applied to a specific decision-making problem given input information, modeling of uncertainty, risk criterion, sequence of decisions and computational tractability
- Document, structure and present results in a written report.
- Keep track of one’s own learning process.
Course Content
Core elements:
* Techniques of optimization under uncertainty: stochastic programming, robust optimization, decomposition techniques, heuristics, scenario generation
Key concepts: here-and-now vs. recourse decisions; 1-stage, 2-stage and multi-stage decision-making processes; robust and stochastic solutions; worst-case and expected-value optimization; risk aversion; heuristics; decomposition techniques; scenario generation; decision rules; value of stochastic solution, expected value of perfect information, etc.
Recommended prerequisites
42112/42002/42111, or similar. This course will build on modeling using mathematical programming. Knowledge of a programming language such as Python or Julia is an advantage, but is not required.
Teaching Method
This course uses lectures, exercises and group assignments.
Faculty
Limited number of seats
Minimum: 12.
Please be aware that this course will only be held if the required minimum number of participants is met. You will be informed 8 days before the start of the course, whether the course will be held.