Production and Delivery Scheduling using Stochastic Programming

The Trading Agent Competition (TAC) Supply Chain Management (SCM) game presents trading agents with several difficult scheduling decisions. Each agent must decide how to build and deliver goods across multiple time periods in the presence of uncertain information about the environment. Stochastic programming (SP) is suggested as a natural, adaptable, and robust solution to scheduling in stochastic environments, particularly TAC SCM. A greedy algorithm, an integer linear program, and several stochastic programs are tested and compared on a simplified scheduling problem that is inspired by TAC SCM, showing that an SP approach can lead to consistently higher profit with only a small increase in running time.