Online Decision-Making and Market Design

Catalog description

This course focuses on studying optimization problems faced with two sources of uncertainty: Online decision-making: Settings involving multiple decisions over time, with uncertainty about future events, and where actions affect current and future outcomes. Market design: Settings involving multiple decision-making agents, with uncertainty about agents’ types, and where agents’ actions affect each other as well as overall outcomes. Our aim is to develop mathematical foundations for these problems – how to formulate them (and why they are related); common algorithmic tools for solving them; analyzing the performance and limits of these algorithms; comparing different models of knowledge in terms of their effect on decision-making; and understanding how to handle different objectives in these settings. Our approach will be primarily theoretical, focusing on mathematical techniques for designing and analyzing algorithms with formal guarantees. However, the problems we consider have significant practical motivation, and we will see lots of examples throughout the course.