Multi-agent learning dynamics

Within the field of artificial intelligence, multi-agent systems are used to model and solve complex problems in today's society. The complexity of such systems requires that individual agents have the ability to learn to optimise their own behaviour. An important challenge is to gain qualitative and theoretical insight into the dynamics of these learning multi-agent systems, as their results are often hard to predict in advance. This dissertation describes how methods from the field of evolutionary game theory can be used for this. These methods are then applied to systems such as social networks and the stock market.