Dynamic Programming for Pure-Strategy Subgame Perfection in an Arbitrary Game
원문 초록
This paper uses value functions to characterize the pure-strategy subgame-perfect equilibria of an arbitrary, possibly infinite-horizon game. It specifies the game's extensive form as a pentaform (Streufert 2023p, arXiv:2107.10801v4), which is a set of quintuples formalizing the abstract relationships between nodes, actions, players, and situations (situations generalize information sets). Because a pentaform is a set, this paper can explicitly partition the game form into piece forms, each of which starts at a (Selten) subroot and contains all subsequent nodes except those that follow a subsequent subroot. Then the set of subroots becomes the domain of a value function, and the piece-form partition becomes the framework for a value recursion which generalizes the Bellman equation from dynamic programming. The main results connect the value recursion with the subgame-perfect equilibria of the original game, under the assumptions of upper- and lower-convergence. Finally, a corollary characterizes subgame perfection as the absence of an improving one-piece deviation.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Writing-RL: Advancing Long-form Writing via Adaptive Curriculum Reinforcement Learning
SuperWriter: Reflection-Driven Long-Form Generation with Large Language Models
How Does Response Length Affect Long-Form Factuality
MedScore: Factuality Evaluation of Free-Form Medical Answers
Frankentext: Stitching random text fragments into long-form narratives
Teaching Large Language Models to Maintain Contextual Faithfulness via Synthetic Tasks and Reinforcement Learning
토론
이 논문과 연결된 포럼 글이 아직 없습니다.