Mean-field Markov decision processes with common noise and open-loop controls

DOI10.1214/21-AAP1713zbMATH Open1491.90179arXiv1912.07883MaRDI QIDQ2135276FDOQ2135276

Authors: Médéric Motte, Huyên Pham 

Publication date: 6 May 2022

Published in: The Annals of Applied Probability (Search for Journal in Brave)

Abstract: We develop an exhaustive study of Markov decision process (MDP) under mean field interaction both on states and actions in the presence of common noise, and when optimization is performed over open-loop controls on infinite horizon. Such model, called CMKV-MDP for conditional McKean-Vlasov MDP, arises and is obtained here rigorously with a rate of convergence as the asymptotic problem of N-cooperative agents controlled by a social planner/influencer that observes the environment noises but not necessarily the individual states of the agents. We highlight the crucial role of relaxed controls and randomization hypothesis for this class of models with respect to classical MDP theory. We prove the correspondence between CMKV-MDP and a general lifted MDP on the space of probability measures, and establish the dynamic programming Bellman fixed point equation satisfied by the value function, as well as the existence of-optimal randomized feedback controls. The arguments of proof involve an original measurable optimal coupling for the Wasserstein distance. This provides a procedure for learning strategies in a large population of interacting collaborative agents. MSC Classification: 90C40, 49L20.

Full work available at URL: https://arxiv.org/abs/1912.07883

Recommendations

• Quantitative propagation of chaos for mean field Markov decision process with common noise
• Mean field Markov decision processes
• On mean field games with common noise and McKean-Vlasov SPDEs
• McKean-Vlasov optimal control: the dynamic programming principle
• A mean field approach for optimization in discrete time

zbMATH Keywords

Markov decision processmean-fieldrandomized controlconditional propagation of chaosmeasurable coupling

Mathematics Subject Classification ID

Dynamic programming in optimal control and differential games (49L20) Markov and semi-Markov decision processes (90C40)

Cites Work

• Weak convergence and empirical processes. With applications to statistics
• Title not available (Why is that?)
• Optimal Transport
• Mass transportation problems. Vol. 1: Theory. Vol. 2: Applications
• Mean field games and mean field type control theory
• On the rate of convergence in Wasserstein distance of the empirical measure
• On the mean speed of convergence of empirical and occupation measures in Wasserstein distance
• Reinforcement learning. An introduction
• Dynamic programming and optimal control. Vol. 2
• Discrete time McKean-Vlasov control problem: a dynamic programming approach
• Dynamic Programming for Optimal Control of Stochastic McKean--Vlasov Dynamics
• Measurability of optimal transportation and strong coupling of martingale measures
• Probabilistic theory of mean field games with applications II. Mean field games with common noise and master equations
• Dynamic programming for mean-field type control
• Randomized dynamic programming principle and Feynman-Kac representation for optimal control of McKean-Vlasov dynamics
• Mean-field optimal control as Gamma-limit of finite agent controls
• Limit theory for controlled McKean-Vlasov dynamics
• Extended mean field control problem: a propagation of chaos result

Cited In (14)

• Model-free mean-field reinforcement learning: mean-field MDP and mean-field Q-learning
• Large Sample Mean-Field Stochastic Optimization
• Extended mean field control problem: a propagation of chaos result
• Continuous-time mean field Markov decision models
• Centralized systemic risk control in the interbank system: weak formulation and gamma-convergence
• Recent developments in machine learning methods for stochastic control and games
• Unified reinforcement Q-learning for mean field game and control problems
• Infinite horizon average cost optimality criteria for mean-field control
• Mean field approximation of an optimal control problem for the continuity equation arising in smart charging
• A dynamic analytic method for risk-aware controlled martingale problems
• Equilibrium pricing of securities in the co-presence of cooperative and non-cooperative populations
• Quantitative propagation of chaos for mean field Markov decision process with common noise
• Large population games with interactions through controls and common noise: convergence results and equivalence between open-loop and closed-loop controls
• Mean field Markov decision processes

This page was built for publication: Mean-field Markov decision processes with common noise and open-loop controls

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2135276)