Program

• Program Overview
• Detailed Program
  
• Programme booklet

Keynote speakers

We are pleased to announce the following keynote speakers for APMOD 2014:

	Clearing the Jungle of Stochastic Optimization Professor Warren B. Powell, Princeton University Mathematical programming has given us a widely used canonical framework for modeling deterministic optimization problems. However, if we introduce a random variable, the field fragments into a number of competing communities with names such as stochastic programming, dynamic programming, stochastic search, simulation optimization, and optimal control. Differences between these fields are magnified by notation and terminology, which hide subtle but more important differences in problem characteristics. Complicating matters further is a misunderstanding of basic terms such as dynamic program, policy, and state variable. While deterministic optimization problems are defined by solving for decisions, sequential stochastic optimization problems are characterized by finding functions known as policies. I will identify four fundamental classes of policies which unify the competing communities into a common framework I call computational stochastic optimization. These ideas will be illustrated using applications drawn from energy, health, freight transportation, and vehicle navigation.
	Robust Optimisation and its Guarantees Professor Berç Rustem, Imperial College Robust optimization provides solutions to decision making under uncertainty. We introduce the problem by considering the formulation of robust policy in the presence of rival models purporting to represent the underlying economic system. We then discuss a collection of models for robust decision making in finance, starting with the robust portfolio optimization problem with uncertain return mean and second moments. We review the robust equal risk contribution and the robust omega ratio problems. Further performance guarantees can be injected by integrating options within a robust framework. This can also be applied to currency portfolios, employing additional constraints to safeguard against arbitrage. We also discuss extensions to multi-stage decision models. The basic paradigm that ensures robustness is minimax: the determination of the optimal decision in view of the worst-case scenario. Most models can be solved as straightforward mathematical programming problems and by dualising the inner optimization problem. Finally, we consider robust option portfolios, for minimizing the maximum hedge error, that require specialized minimax algorithms.
	Structured Stochastic Integer Programs: Understanding the Effects of Uncertainty Professor Stein W. Wallace, Norwegian School of Economics Stochastic integer programs of industrial size are almost always unsolvable to optimality. One possible path to better understand and solve such models is to investigate how uncertainty affects optimal solutions to such programs; What characterizes optimal solutions – are there structures that emerge? This can be done by studying small scale problems that are solvable by off-the-shelf software. Inevitably we are led into the area of options theory to understand what is happening. This talk will report on such structural analyses for several classes of network design problems, and also show how the structures can be used to develop good (constructive) heuristics.