We study mathematical models of the economy, e.g. the Keen model and its

extension by Grasselli from the perspective of the state. Here we are

interested in the question whether economic breakdown and crises can be

prevented by an optimal state intervention with low costs. The

calculation of an optimal policy is a classical problem of machine

learning, especially reinforcement learning, which had quite a success

in providing good policies in a variety of games (e.g. Alpha Zero in

chess) and complex environments (for more information, see e.g. here.)

Principle Investigators: Pavel Gurevich, Hannes Stuke

Members: Johannes Buchner