Dresden 2020 – wissenschaftliches Programm
Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 16: Evolutionary Game Theory and Networks (joint SOE/DY/BP)
SOE 16.4: Vortrag
Donnerstag, 19. März 2020, 15:45–16:00, GÖR 226
Degenerated mirror strategies extort extortioners — Haowei Shi1,2, Sergey Sosnovskiy1, Florian Ellsässer1, Gregory Wheeler3, and •Jan Nagler1 — 1Deep Dynamics Group and Centre for Human and Machine Intelligence, Frankfurt School of Finance and Management, Frankfurt, Germany — 2School of International Economics, China Foreign Affairs University, Beijing, China — 3Centre for Human and Machine Intelligence, Frankfurt School of Finance and Management, Frankfurt, Germany
In iterated games the payoff a player receives in a given round depends on the player's own action and the action of his opponent. Thus, it came as a surprise when Press and Dyson in [PNAS 109:10409 (2012)] introduced so-called extortion zero-determinant (extZD) strategies that - independently of the opponent's strategy - ensure an equal or higher expected payoff. Here, we introduce degenerated mirror strategies that, with a trembling hand that may accidentally take unintended actions, extort any extZD strategy, thereby persistently receiving a higher expected payoff than extortioners. We also show that degenerated mirror strategies outperform the most successful traditional strategies and do well against adaptive strategies. In particular, we demonstrate that they perform equal to or better than a memory-n player that may use the past n actions to determine her next move, where n may be arbitrarily large. Nevertheless, degenerated mirror strategies may be generous when mirroring cooperative strategies.