Time-Local Conditional Equilibria for $N$-Player Stochastic Differential Games with Exploratory Policies

Abstract We study entropy-regularized responses in $N$-player stochastic differential games under a prediction rule that holds the opponents' currently observed actions fixed over the continuation horizon. Each player's control problem produces a Gibbs response conditional on that action profile. We ask when these responses are the full conditionals of one joint action law, called a time-local conditional equilibrium. For smooth positive responses on a simply connected action domain, compatibility is characterized by symmetry of the entropy-scaled Hamiltonian cross-partials and normalizability of the associated potential. A decomposition identifies how continuation-value sensitivity enters this condition. On compact action intervals, a coordinate-path construction yields approximate joint laws whose conditional Kullback--Leibler errors are quadratic in a uniform compatibility gap. Under admissibility and verification conditions, these errors also bound the value loss in each frozen control problem. We obtain stationary discounted counterparts and specialize policy-improvement and martingale methods to estimate the conditional response maps. A two-player linear-quadratic example provides explicit compatible Gaussian laws and near-compatible error calculations, followed by coefficient regression with the value known. MSC Classification: 91A15 , 93E20 , 60H10 , 49L20 , 91A10

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-13
DOI
https://doi.org/10.21203/rs.3.rs-11534523/v1
Primary Topic
Game Theory and Applications
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Time-Local Conditional Equilibria for $N$-Player Stochastic Differential Games with Exploratory Policies

Jisheng Liu
Research Square
Game Theory and Applications
preprint

Time-Local Conditional Equilibria for $N$-Player Stochastic Differential Games with Exploratory Policies

Jisheng Liu
preprint en

Abstract

Abstract We study entropy-regularized responses in $N$-player stochastic differential games under a prediction rule that holds the opponents' currently observed actions fixed over the continuation horizon. Each player's control problem produces a Gibbs response conditional on that action profile. We ask when these responses are the full conditionals of one joint action law, called a time-local conditional equilibrium. For smooth positive responses on a simply connected action domain, compatibility is characterized by symmetry of the entropy-scaled Hamiltonian cross-partials and normalizability of the associated potential. A decomposition identifies how continuation-value sensitivity enters this condition. On compact action intervals, a coordinate-path construction yields approximate joint laws whose conditional Kullback--Leibler errors are quadratic in a uniform compatibility gap. Under admissibility and verification conditions, these errors also bound the value loss in each frozen control problem. We obtain stationary discounted counterparts and specialize policy-improvement and martingale methods to estimate the conditional response maps. A two-player linear-quadratic example provides explicit compatible Gaussian laws and near-compatible error calculations, followed by coefficient regression with the value known. MSC Classification: 91A15 , 93E20 , 60H10 , 49L20 , 91A10

Research Square
Fudan University (CN)
Game Theory and Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon