An Iterative Computational Framework for Infinite-Horizon Mean-Field Linear-Quadratic Zero-Sum Stochastic Differential Games

Yiyuan Wang

arxiv: 2606.08145 · v1 · pith:DGKAUMG2new · submitted 2026-06-06 · 🧮 math.OC

An Iterative Computational Framework for Infinite-Horizon Mean-Field Linear-Quadratic Zero-Sum Stochastic Differential Games

Yiyuan Wang This is my paper

Pith reviewed 2026-06-27 19:31 UTC · model grok-4.3

classification 🧮 math.OC

keywords mean-field gamesstochastic differential gameslinear-quadratic gamesRiccati equationsiterative methodszero-sum gamessaddle point solutionsdecoupling

0 comments

The pith

A monotonically increasing matrix sequence decouples the Riccati equations in mean-field stochastic differential games to compute saddle-point solutions iteratively.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops an iterative computational framework for obtaining saddle-point solutions in infinite-horizon two-person mean-field linear-quadratic zero-sum stochastic differential games. It constructs a monotonically increasing matrix sequence that decouples the strongly coupled original problem into tractable subproblems. Stabilizing solutions to the coupled algebraic Riccati equations of these subproblems are computed sequentially to obtain the solution to the original problem, with a rigorous convergence analysis provided. This approach matters because existing methods are limited to specific cases, while this offers a general method for a broad class of such game problems.

Core claim

By generalizing classical iterative frameworks, a monotonically increasing matrix sequence is constructed to decouple the strongly coupled, analytically intractable original problem into a set of tractable subproblems. Sequentially computing the stabilizing solutions of the coupled algebraic Riccati equations for these subproblems yields the solution to the original problem, and rigorous convergence analysis validates the iterative algorithm. This framework is the first universal computational paradigm applicable to a broad class of game-driven Riccati equations.

What carries the argument

The monotonically increasing matrix sequence that decouples the strongly coupled problem into subproblems each admitting stabilizing solutions of coupled algebraic Riccati equations.

If this is right

The framework applies to a broad class of game-driven Riccati equations rather than only specific simplified setups.
The original problem is solved by sequentially addressing the subproblems instead of directly tackling the coupled system.
Rigorous convergence analysis ensures the validity of the iterative algorithm.
Stabilizing solutions of the subproblem Riccati equations lead to the saddle-point solution of the full game.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This method could potentially be extended to other types of mean-field games or finite-horizon settings.
Implementation in numerical software would allow testing on concrete examples from applications like finance or epidemiology.
Similar decoupling techniques might apply to non-zero-sum or nonlinear variants of these games.

Load-bearing premise

The construction of a monotonically increasing matrix sequence that successfully decouples the original strongly coupled problem into subproblems each admitting stabilizing solutions of their coupled algebraic Riccati equations.

What would settle it

A concrete counterexample game for which no such monotonically increasing matrix sequence exists or for which the derived solution from subproblems fails to satisfy the saddle-point condition of the original game.

read the original abstract

This work develops an iterative computational framework to obtain saddle-point solutions for infinite-horizon two-person mean-field linear-quadratic zero-sum stochastic differential games. By generalizing classical iterative framework, we construct a monotonically increasing matrix sequence to decouple the strongly coupled, analytically intractable original problem into a set of tractable subproblems. By sequentially computing the stabilizing solutions of the coupled algebraic Riccati equations for these subproblems, we can further derive the solution to the original problem. Rigorous convergence analysis is established to validate the proposed iterative algorithm. Different from existing algorithms limited to specific simplified setups, this framework proposes the first universal computational paradigm applicable to a broad class of game-driven Riccati equations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The claimed monotone matrix sequence for decoupling mean-field zero-sum Riccati equations is the central new piece, but the abstract and stress-test leave its validity for nonzero interaction terms unverified.

read the letter

The paper extends classical iterative Riccati methods to infinite-horizon mean-field LQ zero-sum stochastic differential games. It builds a monotonically increasing matrix sequence that splits the original coupled system into tractable subproblems, solves their algebraic Riccati equations in sequence, and recovers the saddle-point solution, with a claimed convergence proof.

This is a direct generalization of earlier iterative schemes that were restricted to simpler cases without mean-field terms or with additional structure. The abstract positions the construction as the first universal framework for this class, which is the main concrete advance.

The soft spot is exactly where the stress-test flags it: the abstract gives no explicit recurrence for the matrix sequence and no check that monotonicity survives when the mean-field coefficients are nonzero. If the sequence stops increasing or the subproblem Riccati equations lose stabilizability once the mean-field measure enters, the decoupling fails and the broad applicability claim does not hold. The paper also needs to confirm that the saddle-point property is preserved at each step rather than assumed.

The rest of the setup follows standard LQ stochastic control assumptions, so the technical foundation is familiar. No obvious circularity or invented entities appear in the description.

Control theorists who need numerical methods for mean-field games would find this relevant. A reader already working on Riccati iterations or stochastic differential games gets the most value. The work is coherent enough on its own terms to merit referee time; the proofs of monotonicity and decoupling are the items that need external scrutiny.

Referee Report

1 major / 0 minor

Summary. The paper claims to develop an iterative computational framework for infinite-horizon mean-field linear-quadratic zero-sum stochastic differential games. By generalizing classical iterative Riccati methods, it constructs a monotonically increasing matrix sequence to decouple the strongly coupled game-driven Riccati equations into tractable subproblems; sequentially computing the stabilizing solutions of the subproblems' coupled algebraic Riccati equations then recovers the original saddle-point solution, with a claimed rigorous convergence analysis. The work positions the framework as the first universal computational paradigm for a broad class of such equations, unlike prior algorithms limited to simplified setups.

Significance. If the monotonicity, decoupling, and convergence hold under general mean-field coefficients while preserving the saddle-point property, the result would supply a practical computational method for an analytically intractable class of mean-field games, extending classical iterative Riccati techniques in a verifiable way.

major comments (1)

[Abstract] Abstract: the central claim that a monotonically increasing matrix sequence successfully decouples the original strongly coupled problem into subproblems each admitting stabilizing solutions rests on an unstated recurrence and unverified monotonicity when mean-field interaction terms are nonzero; without these details the iteration may fail to recover the original saddle point or introduce hidden coupling through the mean-field measure.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the thoughtful review and for highlighting the need for clarity regarding the iterative construction in the abstract. We address the major comment below.

read point-by-point responses

Referee: [Abstract] Abstract: the central claim that a monotonically increasing matrix sequence successfully decouples the original strongly coupled problem into subproblems each admitting stabilizing solutions rests on an unstated recurrence and unverified monotonicity when mean-field interaction terms are nonzero; without these details the iteration may fail to recover the original saddle point or introduce hidden coupling through the mean-field measure.

Authors: The recurrence relation for the matrix sequence is explicitly stated in Equation (3.5) of the manuscript, which generalizes the classical iterative Riccati scheme by incorporating the mean-field interaction terms directly into the update. Theorem 4.2 establishes monotonicity of the sequence for general (including nonzero) mean-field coefficients under the standard stabilizability and detectability assumptions, with the proof relying on an inductive argument that preserves the positive-semidefinite ordering. Section 4 further shows that the limit satisfies the original coupled AREs and recovers the saddle-point strategies without residual coupling through the mean-field measure, as the decoupling is achieved by construction at each iteration step. revision: partial

Circularity Check

0 steps flagged

No circularity: generalization of classical Riccati iteration presented as independent construction

full rationale

The provided abstract and context describe an iterative framework obtained by generalizing classical iterative Riccati methods to the mean-field zero-sum setting, with a claimed monotonically increasing matrix sequence that decouples the problem and a separate rigorous convergence analysis. No equations, definitions, or steps are exhibited that reduce the saddle-point solution, the matrix sequence, or the convergence claim to a fitted input, self-definition, or self-citation chain by construction. The derivation is therefore self-contained relative to external classical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review performed on abstract only; full derivation details, explicit assumptions, and any fitted quantities are unavailable.

axioms (1)

domain assumption Existence of stabilizing solutions to the coupled algebraic Riccati equations arising in each decoupled subproblem
The sequential computation step relies on this standard LQ-control assumption being satisfied for the constructed subproblems.

pith-pipeline@v0.9.1-grok · 5642 in / 1289 out tokens · 25558 ms · 2026-06-27T19:31:47.398098+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

23 extracted references · 11 canonical work pages

[1]

A Deterministic Setting for the Numerical Computation of the Stabilizing Solutions to Stochastic Game-Theoretic Riccati Equations

Samir Aberkane and Vasile Dragan. “A Deterministic Setting for the Numerical Computation of the Stabilizing Solutions to Stochastic Game-Theoretic Riccati Equations”. In:Mathematics11.9 (2023). DOI:10.3390/math11092068

work page doi:10.3390/math11092068 2023
[2]

Mean-field stochastic differential equations and associated PDEs

Rainer Buckdahn, Juan Li, Shige Peng, and Catherine Rainer. “Mean-field stochastic differential equations and associated PDEs”. In: (2017)

2017
[3]

Iterative algorithms for stabilizing solutions of game theoretic Riccati equations of stochastic control

V Dragan, G Freiling, T Morozan, and AM Stoica. “Iterative algorithms for stabilizing solutions of game theoretic Riccati equations of stochastic control”. In:Proceedings of the 18th International Symposium on Mathematical Theory of Networks & Systems. 2008, pp. 1–11

2008
[4]

Stabilizing solution of periodic game-theoretic Riccati differential equation of stochas- tic control

Vasile Dragan. “Stabilizing solution of periodic game-theoretic Riccati differential equation of stochas- tic control”. In:IMA Journal of Mathematical Control and Information32.4 (Dec. 2015), pp. 839– 865.DOI:10.1093/imamci/dnu026

work page doi:10.1093/imamci/dnu026 2015
[5]

Computing The Stabilizing Solution of a Large Class of Stochas- tic Game Theoretic Riccati Differential Equations: A Deterministic Approximation

Vasile Dragan and Samir Aberkane. “Computing The Stabilizing Solution of a Large Class of Stochas- tic Game Theoretic Riccati Differential Equations: A Deterministic Approximation”. In:SIAM Jour- nal on Control and Optimization55.2 (2017), pp. 650–670.DOI:10.1137/15M1049038

work page doi:10.1137/15m1049038 2017
[6]

On computing the stabilizing solution of a class of discrete-time periodic Riccati equations

Vasile Dragan, Samir Aberkane, and Ivan G. Ivanov. “On computing the stabilizing solution of a class of discrete-time periodic Riccati equations”. In:International Journal of Robust and Nonlinear Control25.7 (2015), pp. 1066–1093.DOI:https://doi.org/10.1002/rnc.3131

work page doi:10.1002/rnc.3131 2015
[7]

Computation of the stabilizing solution of game theoretic Riccati equation arising in stochasticH ∞ control problems

Vasile Dragan and Ivan G. Ivanov. “Computation of the stabilizing solution of game theoretic Riccati equation arising in stochasticH ∞ control problems”. In:Numerical Algorithms57.3 (July 1, 2011), pp. 357–375.DOI:10.1007/s11075-010-9432-7

work page doi:10.1007/s11075-010-9432-7 2011
[8]

Mathematical Methods in Robust Control of Linear Stochastic Systems

Vasile Dragan, Morozan Toader, and Adrian-Mihail Stoica. Mathematical Methods in Robust Control of Linear Stochastic Systems. Jan. 2013.DOI:10.1007/978-1-4614-8663-3

work page doi:10.1007/978-1-4614-8663-3 2013
[9]

An iterative algorithm to solve state-perturbed stochastic algebraic Riccati equations in LQ zero-sum games

Yantao Feng and Brian D. O. Anderson. “An iterative algorithm to solve state-perturbed stochastic algebraic Riccati equations in LQ zero-sum games”. In:Systems & Control Letters59.1 (Jan. 1, 2010–), pp. 50–56.DOI:https://doi.org/10.1016/j.sysconle.2009.11.006

work page doi:10.1016/j.sysconle.2009.11.006 2010
[10]

On the iterative solution toH ∞ control prob- lems

Ivan G Ivanov, Ivelin G Ivanov, and Nikolay C Netov. “On the iterative solution toH ∞ control prob- lems”. In:Applied Mathematics6.8 (2015), pp. 1263–1270

2015
[11]

The iterative solution to LQ zero-sum stochastic differential games

Ivan G. Ivanov and Ivelin G. Ivanov. “The iterative solution to LQ zero-sum stochastic differential games”. In:Journal of Applied Mathematics and Computing56.1 (Feb. 1, 2018), pp. 547–559.DOI: 10.1007/s12190-017-1086-3

work page doi:10.1007/s12190-017-1086-3 2018
[12]

Foundations of kinetic theory

M Kac. “Foundations of kinetic theory”. In:Proc. Third Berkely Symp. on Math. Stat. and Prob. V ol. 3. 1956, pp. 171–197

1956
[13]

IEEE Transactions on Automatic Control70(4), 2690–2697 (2024)

Na Li, Xun Li, and Zuo Quan Xu. “Policy Iteration Reinforcement Learning Method for Continuous- Time Linear–Quadratic Mean-Field Control Problems”. In:IEEE Transactions on Automatic Control 70.4 (Apr. 2025), pp. 2690–2697.DOI:10.1109/TAC.2024.3494656

work page doi:10.1109/tac.2024.3494656 2025
[14]

Indefinite mean-field type linear–quadratic stochastic optimal con- trol problems

Na Li, Xun Li, and Zhiyong Yu. “Indefinite mean-field type linear–quadratic stochastic optimal con- trol problems”. In:Automatica122 (2020), p. 109267. 28 REFERENCES

2020
[15]

Mean-field linear-quadratic stochastic differential games in an infinite horizon*

Li, Xun, Shi, Jingtao, and Yong, Jiongmin. “Mean-field linear-quadratic stochastic differential games in an infinite horizon*”. In:ESAIM: COCV27 (2021), p. 81.DOI:10.1051/cocv/2021078

work page doi:10.1051/cocv/2021078 2021
[16]

A class of Markov processes associated with nonlinear parabolic equations

Henry P McKean Jr. “A class of Markov processes associated with nonlinear parabolic equations”. In:Proceedings of the National Academy of Sciences56.6 (1966), pp. 1907–1911

1966
[17]

Mean-field linear-quadratic stochastic differential games

Jingrui Sun, Hanxiao Wang, and Zhen Wu. “Mean-field linear-quadratic stochastic differential games”. In:Journal of Differential Equations296 (2021), pp. 299–334

2021
[18]

Stochastic linear-quadratic optimal control theory: differential games and mean-field problems

Jingrui Sun and Jiongmin Yong. Stochastic linear-quadratic optimal control theory: differential games and mean-field problems. Springer Nature, 2020

2020
[19]

Stochastic linear-quadratic optimal control theory: Open-loop and closed-loop solutions

Jingrui Sun and Jiongmin Yong. Stochastic linear-quadratic optimal control theory: Open-loop and closed-loop solutions. Springer Nature, 2020

2020
[20]

Reinforcement learning for exploratory linear-quadratic two-person zero-sum stochastic differential games

Zhongshi Sun and Guangyan Jia. “Reinforcement learning for exploratory linear-quadratic two-person zero-sum stochastic differential games”. In:Applied Mathematics and Computation442 (2023), p. 127763.DOI:https://doi.org/10.1016/j.amc.2022.127763

work page doi:10.1016/j.amc.2022.127763 2023
[21]

A closed-loop saddle point for zero-sum linear-quadratic stochastic differential games with mean-field type

Ran Tian, Zhiyong Yu, and Rucheng Zhang. “A closed-loop saddle point for zero-sum linear-quadratic stochastic differential games with mean-field type”. In:Systems & Control Letters136 (2020), p. 104624

2020
[22]

A Unified Computational Approach for Zero-Sum Linear-Quadratic Stochastic Differ- ential Games in Infinite Horizons

Yiyuan Wang. A Unified Computational Approach for Zero-Sum Linear-Quadratic Stochastic Differ- ential Games in Infinite Horizons. 2025

2025
[23]

Reinforcement Learning Method for Zero-Sum Linear-Quadratic Stochastic Differen- tial Games in Infinite Horizons

Yiyuan Wang. Reinforcement Learning Method for Zero-Sum Linear-Quadratic Stochastic Differen- tial Games in Infinite Horizons. 2026

2026

[1] [1]

A Deterministic Setting for the Numerical Computation of the Stabilizing Solutions to Stochastic Game-Theoretic Riccati Equations

Samir Aberkane and Vasile Dragan. “A Deterministic Setting for the Numerical Computation of the Stabilizing Solutions to Stochastic Game-Theoretic Riccati Equations”. In:Mathematics11.9 (2023). DOI:10.3390/math11092068

work page doi:10.3390/math11092068 2023

[2] [2]

Mean-field stochastic differential equations and associated PDEs

Rainer Buckdahn, Juan Li, Shige Peng, and Catherine Rainer. “Mean-field stochastic differential equations and associated PDEs”. In: (2017)

2017

[3] [3]

Iterative algorithms for stabilizing solutions of game theoretic Riccati equations of stochastic control

V Dragan, G Freiling, T Morozan, and AM Stoica. “Iterative algorithms for stabilizing solutions of game theoretic Riccati equations of stochastic control”. In:Proceedings of the 18th International Symposium on Mathematical Theory of Networks & Systems. 2008, pp. 1–11

2008

[4] [4]

Stabilizing solution of periodic game-theoretic Riccati differential equation of stochas- tic control

Vasile Dragan. “Stabilizing solution of periodic game-theoretic Riccati differential equation of stochas- tic control”. In:IMA Journal of Mathematical Control and Information32.4 (Dec. 2015), pp. 839– 865.DOI:10.1093/imamci/dnu026

work page doi:10.1093/imamci/dnu026 2015

[5] [5]

Computing The Stabilizing Solution of a Large Class of Stochas- tic Game Theoretic Riccati Differential Equations: A Deterministic Approximation

Vasile Dragan and Samir Aberkane. “Computing The Stabilizing Solution of a Large Class of Stochas- tic Game Theoretic Riccati Differential Equations: A Deterministic Approximation”. In:SIAM Jour- nal on Control and Optimization55.2 (2017), pp. 650–670.DOI:10.1137/15M1049038

work page doi:10.1137/15m1049038 2017

[6] [6]

On computing the stabilizing solution of a class of discrete-time periodic Riccati equations

Vasile Dragan, Samir Aberkane, and Ivan G. Ivanov. “On computing the stabilizing solution of a class of discrete-time periodic Riccati equations”. In:International Journal of Robust and Nonlinear Control25.7 (2015), pp. 1066–1093.DOI:https://doi.org/10.1002/rnc.3131

work page doi:10.1002/rnc.3131 2015

[7] [7]

Computation of the stabilizing solution of game theoretic Riccati equation arising in stochasticH ∞ control problems

Vasile Dragan and Ivan G. Ivanov. “Computation of the stabilizing solution of game theoretic Riccati equation arising in stochasticH ∞ control problems”. In:Numerical Algorithms57.3 (July 1, 2011), pp. 357–375.DOI:10.1007/s11075-010-9432-7

work page doi:10.1007/s11075-010-9432-7 2011

[8] [8]

Mathematical Methods in Robust Control of Linear Stochastic Systems

Vasile Dragan, Morozan Toader, and Adrian-Mihail Stoica. Mathematical Methods in Robust Control of Linear Stochastic Systems. Jan. 2013.DOI:10.1007/978-1-4614-8663-3

work page doi:10.1007/978-1-4614-8663-3 2013

[9] [9]

An iterative algorithm to solve state-perturbed stochastic algebraic Riccati equations in LQ zero-sum games

Yantao Feng and Brian D. O. Anderson. “An iterative algorithm to solve state-perturbed stochastic algebraic Riccati equations in LQ zero-sum games”. In:Systems & Control Letters59.1 (Jan. 1, 2010–), pp. 50–56.DOI:https://doi.org/10.1016/j.sysconle.2009.11.006

work page doi:10.1016/j.sysconle.2009.11.006 2010

[10] [10]

On the iterative solution toH ∞ control prob- lems

Ivan G Ivanov, Ivelin G Ivanov, and Nikolay C Netov. “On the iterative solution toH ∞ control prob- lems”. In:Applied Mathematics6.8 (2015), pp. 1263–1270

2015

[11] [11]

The iterative solution to LQ zero-sum stochastic differential games

Ivan G. Ivanov and Ivelin G. Ivanov. “The iterative solution to LQ zero-sum stochastic differential games”. In:Journal of Applied Mathematics and Computing56.1 (Feb. 1, 2018), pp. 547–559.DOI: 10.1007/s12190-017-1086-3

work page doi:10.1007/s12190-017-1086-3 2018

[12] [12]

Foundations of kinetic theory

M Kac. “Foundations of kinetic theory”. In:Proc. Third Berkely Symp. on Math. Stat. and Prob. V ol. 3. 1956, pp. 171–197

1956

[13] [13]

IEEE Transactions on Automatic Control70(4), 2690–2697 (2024)

Na Li, Xun Li, and Zuo Quan Xu. “Policy Iteration Reinforcement Learning Method for Continuous- Time Linear–Quadratic Mean-Field Control Problems”. In:IEEE Transactions on Automatic Control 70.4 (Apr. 2025), pp. 2690–2697.DOI:10.1109/TAC.2024.3494656

work page doi:10.1109/tac.2024.3494656 2025

[14] [14]

Indefinite mean-field type linear–quadratic stochastic optimal con- trol problems

Na Li, Xun Li, and Zhiyong Yu. “Indefinite mean-field type linear–quadratic stochastic optimal con- trol problems”. In:Automatica122 (2020), p. 109267. 28 REFERENCES

2020

[15] [15]

Mean-field linear-quadratic stochastic differential games in an infinite horizon*

Li, Xun, Shi, Jingtao, and Yong, Jiongmin. “Mean-field linear-quadratic stochastic differential games in an infinite horizon*”. In:ESAIM: COCV27 (2021), p. 81.DOI:10.1051/cocv/2021078

work page doi:10.1051/cocv/2021078 2021

[16] [16]

A class of Markov processes associated with nonlinear parabolic equations

Henry P McKean Jr. “A class of Markov processes associated with nonlinear parabolic equations”. In:Proceedings of the National Academy of Sciences56.6 (1966), pp. 1907–1911

1966

[17] [17]

Mean-field linear-quadratic stochastic differential games

Jingrui Sun, Hanxiao Wang, and Zhen Wu. “Mean-field linear-quadratic stochastic differential games”. In:Journal of Differential Equations296 (2021), pp. 299–334

2021

[18] [18]

Stochastic linear-quadratic optimal control theory: differential games and mean-field problems

Jingrui Sun and Jiongmin Yong. Stochastic linear-quadratic optimal control theory: differential games and mean-field problems. Springer Nature, 2020

2020

[19] [19]

Stochastic linear-quadratic optimal control theory: Open-loop and closed-loop solutions

Jingrui Sun and Jiongmin Yong. Stochastic linear-quadratic optimal control theory: Open-loop and closed-loop solutions. Springer Nature, 2020

2020

[20] [20]

Reinforcement learning for exploratory linear-quadratic two-person zero-sum stochastic differential games

Zhongshi Sun and Guangyan Jia. “Reinforcement learning for exploratory linear-quadratic two-person zero-sum stochastic differential games”. In:Applied Mathematics and Computation442 (2023), p. 127763.DOI:https://doi.org/10.1016/j.amc.2022.127763

work page doi:10.1016/j.amc.2022.127763 2023

[21] [21]

A closed-loop saddle point for zero-sum linear-quadratic stochastic differential games with mean-field type

Ran Tian, Zhiyong Yu, and Rucheng Zhang. “A closed-loop saddle point for zero-sum linear-quadratic stochastic differential games with mean-field type”. In:Systems & Control Letters136 (2020), p. 104624

2020

[22] [22]

A Unified Computational Approach for Zero-Sum Linear-Quadratic Stochastic Differ- ential Games in Infinite Horizons

Yiyuan Wang. A Unified Computational Approach for Zero-Sum Linear-Quadratic Stochastic Differ- ential Games in Infinite Horizons. 2025

2025

[23] [23]

Reinforcement Learning Method for Zero-Sum Linear-Quadratic Stochastic Differen- tial Games in Infinite Horizons

Yiyuan Wang. Reinforcement Learning Method for Zero-Sum Linear-Quadratic Stochastic Differen- tial Games in Infinite Horizons. 2026

2026