A Global seminar on Financial Mathematics was held on May 15, at Moscow School of Economics.
Topic: «An Algorithmic Approach to Optimal Asset Liquidation Problems».
Speaker: Juri Hinz (University of Technology Sydney).
Optimal control problems of stochastic switching type appear frequently in applications but are notoriously challenging from computational viewpoint. Furthermore, given a numerical approximation of the optimal control strategy, its quality is not easy to assess. Using duality techniques of by C. Rogers, we examine an efficient recursive algorithm which yields upper and lower bounds to perform diagnostics of numerical solutions to certain stochastic switching problems. We extend this approach to control problems under partial observability and illustrate our techniques by applications to mining operations and asset liquidation.