BEA512 Modeling decision problems under uncertainty
The course will give the students in-depth knowledge of how to include uncertainty into mathematical planning models, with a focus on mathematical programming. There will be an emphasis on how to model uncertainty to obtain meaningful models for planning purposes within logistics / supply chains as well as resource planning. Basic algorithmic structures will also be presented, but the course is primarily modeling oriented, not algorithmically oriented.
The course will cover:
- The role of what-if analysis and parametric optimization in decision-making under uncertainty
- How to handle feasibility in optimization under uncertainty
- Basic algorithmic structures for stochastic programming
- Scenario generation, that is, how to represent uncertainty in mathematical programming models
- Alternative modeling concepts for handling uncertainty
- Stochastic network design
- Stochastic facility layout
- Examples from energy systems modeling.
After completing the course, students will be able to:
- handle uncertainty in the context of mathematical planning models, particularly mathematical programming
- model complex, time-dependent planning problems facing uncertainty in such as demand and prices
- critically analyze published material on uncertainty in planning
- generate scenarios for stochastic optimization, to end up with numerically efficient models.
- review, assess and utilize relevant scientific papers addressing stochastic programming
Lectures and student presentations. The course will meet three times during the term, each time lunch to lunch, covering two full days, typically lunch Wednesday to lunch Friday.
It is necessary with solid knowledge of standard optimization theory, especially linear programming. Experience with stochastic programming is not expected.
Basic knowledge of mathematical programming modeling
Requirements for course approval
Presence is required.
Each student must hand in a term paper after the course is finished, where the theories of the course are used on a mathematical programming model, preferably one the student is already working with in his / her thesis. Deadline will be end of January.
It is an advantage to know advanced modeling languages, such as AMPL, MPL or GAMS, but not required. But this will not be taught. A good basis would be ENE420 or something equivalent.
Alan King and Stein W. Wallace, Modeling with stochastic programming, Springer, 2012.
Peter Kall and Stein W. Wallace, Stochastic Programming, Wiley, Chichester, 1994. Available for free.
Articles published in scientific journals.
Stein W. Wallace, Department of Business and Management Science