COURSE OBJECTIVES

This course will include stochastic optimization problem modeling and processing with dynamic programming and derivative techniques. 

TARGET AUDIENCE

Anyone aiming to discover stochastic optimization through dynamic programming. 

Deterministic Dynamic Programming 
• Introduction.
• Deterministic dynamic programming: principles. Transition equation, state, Bellman values. Shortest path problems. Management of production unit start-ups.
• Bellman values and dual variables. Economical interpretation of Bellman values. Inventory management case. Link with dual variables. 

Stochastic dynamic programming 
• From deterministic to stochastic. Modeling a dynamic stochastic optimization problem. Non-Anticipativity constraints. Dynamic programming on scenario trees. Application to option pricing.
• Dynamic stochastic programming. State definition and probabilistic structure. Examples of modeling. Bellman values economic interpretation. Valuation of future contracts. 

Stochastic dynamic programming (continued) 
• Inventory management and dynamic programming: some examples. Modeling and effects on Bellman functions. 
• Large dynamic problems. Limits of dynamic programming for large problems. 
• Decomposition methods: dual dynamic programming, scenario decomposition, scenario tree method.
• Processing large dynamic problems: application to annual electricity production planning. Dynamic management modeling of interconnected systems in the energy sector. Resolution by decomposition. Solving by dual dynamic programming.
• Links with reinforcement learning. 
• Sampling and generalization techniques. 
• Dynamic learning and optimization schemes.

Training wrap-up