Year: 2023
Type: dissertation
Abstract: The equilibria of auction markets with incomplete information are poorly understood, and their computation is believed to be generally intractable. In this dissertation, we study a machine learning approach to find such Bayes-Nash equilibria based on modified multi-agent gradient dynamics. We find that in small and medium-sized markets, such dynamics often indeed converge to equilibria, and provably so for symmetric markets. We further study the underlying theory, as well as applications.
Authors: Heidekrüger, Stefan, Bichler, Martin 1971-, Technische Universität München
Institution: Technical University of Munich
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
Funder: Deutsche Forschungsgemeinschaft
OpenAlex ID: W7029850927
