Workshop by the Corvinus Institute of Advanced Studies
The Corvinus Center for Operation Research (CCOR) invites you to the workshop: Dynamics and Decisions
Keynote speaker: William Sudderth (University of Minnesota)
Speakers: János Flesch (Maastricht University) and Arkadi Predtetchinski (Maastricht University)
Date: October 27, 2022
Time: 13:00-16:00
Venue: Corvinus University, Building C, Room: C510
Program:
- 13:00-14:00: William Sudderth: Remarks on Finitely Additive Probability
- 14:00-14:30: Coffee break
- 14:30-15:10: Arkadi Predtetchinski: Bounded foresight in random decision problems
- 15:15-15:55: Janos Flesch: Existence of equilibria in repeated games with long-run payoffs
Talks:
William Sudderth:
Title: Remarks on Finitely Additive Probability
Abstract: Although finitely additive probability theory is less widely studied than the more conventional countably additive theory, it has a number of interesting features and open questions. I will discuss some of these and give several examples.
Arkadi Predtetchinski:
Title: Bounded foresight in random decision problems
Abstract: A decision-maker (DM) has to make an infinite sequence of decisions to reach a certain objective. She is restricted in her choices by the decision tree. The decision tree, however, is chosen randomly by Nature and is only been gradually revealed to the DM as she proceeds with making decisions. In this framework, called a random decision problem, we examine several scenarios of bounded foresight, differing in the depth of the decision tree revealed to the DM at each stage. Our focus is on the value, i.e. the probability for the DM to produce a sequence of decisions that would meet her objective. Our benchmark is the scenario of an omniscient DM, under which the entire decision tree is revealed to the DM before she makes any decisions.
Janos Flesch:
Title: Existence of equilibria in repeated games with long-run payoffs
Abstract: We consider repeated games with tail-measurable payoffs, i.e., when the payoffs depend only on what happens in the long run. We show that every repeated game with tail-measurable payoffs admits an ε-equilibrium, for every ε>0, provided that the set of players is finite or countably infinite and the action sets are finite. The proof relies on techniques from stochastic games and from alternating-move games with Borel-measurable payoffs.