For problems requiring medium to long term decision making, such as in environmental management, industrial planning (for example production rates for processing plants, determining the best location of processing plants) and transport logistics (for example determining optimal road designs), the future uncertainty of key underlying variables such as yields and demand must be taken into account. However, adding stochasticity into a potentially very large number of variables can create insurmountable computational complexity.
Currently, there are multiple approaches one can take to find optimal solutions in the presence of stochasticity, such as stochastic linear programming, Real Options methodology, optimal stochastic control and adaptive management. Each of these different approaches have their own merits and drawbacks. Recent developments in solving medium to long term decision making problems where stochasticity is present have focussed on combining different methodologies to amplify the relative merits while limiting drawbacks, to develop more efficient algorithms that allow more complex problems to be solved.
This session aims to highlight these recent developments in combining different approaches to solving stochastic optimisation problems, and bring together researchers and practitioners with experience in different stochastic optimisation methodologies to discuss their relative merits and catalyse further collaboration to continue the development of potentially more efficient algorithms for decision making under uncertainty.
Key topics: Stochastic optimisation, Real options, Stochastic control, Stochastic linear programming