pith. sign in

arxiv: 2112.03153 · v3 · submitted 2021-12-06 · 🧮 math.OC · cs.NA· math.NA

On Discrete-Time Approximations to Infinite Horizon Differential Games

Javier de Frutos , V\'ictor Gat\'on , Julia Novo This is my paper

Pith reviewed 2026-05-24 13:05 UTC · model grok-4.3

classification 🧮 math.OC cs.NAmath.NA
keywords differential gamesdiscrete-time approximationvalue function convergenceNash equilibriuminfinite horizonsemidiscretizationfully discrete approximationepsilon-Nash equilibrium
0
0 comments X

The pith

Discrete-time approximations converge to the value functions of infinite-horizon N-player differential games as the time step vanishes.

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

The paper proves convergence of two types of discretization for noncooperative infinite-horizon differential games. Both the semidiscrete (discrete time only) and fully discrete (time and state) versions produce value functions that approach the continuous-time value functions when the time step, or both time step and mesh size, go to zero. The same limits show that any discrete Nash equilibrium is an ε-Nash equilibrium for the original continuous game. This supplies a rigorous justification for replacing the continuous problem with computable discrete models that can be solved by standard dynamic programming or other numerical methods.

Core claim

As either the discretization time step or both time step and mesh size parameters approach zero the discrete value function approximates the value function of the differential game. Furthermore, the discrete Nash equilibrium is an ε-Nash equilibrium for the continuous-time differential game both in the discrete-time and fully discrete cases.

What carries the argument

The discrete-time semidiscretization and the fully discrete (time-and-state) approximation, which are shown to converge in value and in equilibrium sense to the continuous differential game.

If this is right

  • Discrete dynamic programming can be used to compute arbitrarily accurate approximations to continuous value functions.
  • Any computed discrete Nash equilibrium supplies an explicit ε-Nash strategy profile for the original continuous game.
  • The same convergence holds when only time is discretized and when both time and state space are discretized.

Where Pith is reading between the lines

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

  • The results open the door to systematic numerical study of games whose continuous formulations are analytically intractable.
  • Error bounds derived from the convergence proofs could guide the choice of time step needed to achieve a target ε.
  • The framework may extend directly to other approximation schemes such as policy iteration on the discrete models.

Load-bearing premise

The continuous differential game is assumed to have well-defined value functions and Nash equilibria under suitable regularity conditions on the dynamics, costs, and discount factors.

What would settle it

A sequence of decreasing time steps for which the difference between the discrete value function and the continuous value function stays bounded away from zero for some admissible dynamics and costs.

read the original abstract

In this paper we study a discrete-time semidiscretization and a fully discretization (discrete-time, discrete-state) of an infinite time horizon noncooperative $N$-player differential game. We prove that as either the discretization time step or both time step and mesh size parameters approach zero the discrete value function approximates the value function of the differential game. Furthermore, the discrete Nash equilibrium is an $\epsilon$-Nash equilibrium for the continuous-time differential game both in the discrete-time and fully discrete cases.

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.

Referee Report

1 major / 1 minor

Summary. The paper examines semidiscrete (discrete-time) and fully discrete (discrete-time, discrete-state) approximations to infinite-horizon noncooperative N-player differential games. It asserts that, under suitable regularity conditions on the dynamics, costs, and discount factors, the discrete value functions converge to the continuous-time value functions as the time step (and mesh size) tends to zero, and that discrete Nash equilibria are ε-Nash equilibria for the original continuous-time game.

Significance. If the convergence and ε-Nash results hold under the stated conditions, the work supplies a rigorous justification for using discrete approximations to analyze or compute solutions for continuous differential games. This is relevant for numerical methods in dynamic game theory. The manuscript supplies proofs of the approximation statements (as asserted in the abstract), which is a positive feature for a contribution in this area.

major comments (1)
  1. [Abstract and main theorems (presumably §3–§5)] The central claims (convergence of value functions and ε-Nash property) are meaningful only if the continuous-time game possesses well-defined value functions and Nash equilibria. The abstract invokes “suitable regularity conditions” for this background existence but the manuscript does not contain a self-contained verification or a precise citation to a theorem that matches the exact setting (N players, infinite horizon, specific dynamics and costs) used in the approximation theorems. This assumption is load-bearing for the main results.
minor comments (1)
  1. [Notation and preliminaries] Notation for the discretization parameters (time step h and mesh size) should be introduced once with a clear table or list of symbols to avoid ambiguity when both semidiscrete and fully discrete cases are discussed.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading and for highlighting the need to make the foundational existence assumptions fully explicit. We address the single major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract and main theorems (presumably §3–§5)] The central claims (convergence of value functions and ε-Nash property) are meaningful only if the continuous-time game possesses well-defined value functions and Nash equilibria. The abstract invokes “suitable regularity conditions” for this background existence but the manuscript does not contain a self-contained verification or a precise citation to a theorem that matches the exact setting (N players, infinite horizon, specific dynamics and costs) used in the approximation theorems. This assumption is load-bearing for the main results.

    Authors: We agree that the existence of value functions and Nash equilibria for the continuous-time N-player infinite-horizon game is a load-bearing assumption. Under the paper’s standing hypotheses (Lipschitz-continuous dynamics and running costs, bounded controls, positive discount factors, and the Isaacs condition), existence follows from standard results in the differential-games literature. In the revised manuscript we will insert a short paragraph (new §2.3) that states the precise conditions and cites a matching theorem (e.g., Theorem 4.2 of Bardi & Capuzzo-Dolcetta, Optimal Control and Viscosity Solutions, or the infinite-horizon extension in Başar & Olsder, Dynamic Noncooperative Game Theory, 2nd ed., §6.5) that applies verbatim to our N-player setting. If the editor prefers, we can also sketch the fixed-point argument in an appendix. This change clarifies the background without affecting the approximation proofs. revision: yes

Circularity Check

0 steps flagged

No significant circularity; convergence proofs are independent limit statements.

full rationale

The paper establishes convergence of discrete value functions and ε-Nash equilibria to their continuous counterparts as the time step (and mesh size) approach zero. These are standard approximation theorems under regularity assumptions on the data; the background existence of continuous value functions and Nash equilibria is invoked via external regularity conditions rather than derived internally or via self-citation. No step reduces a claimed result to a fitted parameter, self-definition, or load-bearing self-citation chain. The derivation chain is therefore self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on standard existence results for value functions and Nash equilibria in continuous differential games; these are domain assumptions typical of the field but not detailed in the abstract. No free parameters or invented entities are mentioned.

axioms (1)
  • domain assumption The continuous-time N-player differential game admits value functions and Nash equilibria under appropriate regularity conditions on the dynamics and payoffs.
    The approximation theorems presuppose the existence of the continuous objects whose discrete counterparts are shown to converge.

pith-pipeline@v0.9.0 · 5610 in / 1358 out tokens · 31242 ms · 2026-05-24T13:05:41.795969+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages

  1. [1]

    Akian, M., Gaubert, S

    M. Akian, M., Gaubert, S. & Lakhoua, A., The max-plus finite element method for solving deterministic optimal control problems: basic properties and conver- gence analysis, SIAM J. Control Optim. 47, 2008, pp. 817–848

    work page 2008

  2. [2]

    & Capuzzo-Dolcetta, I., Optimal Control and Viscosity Solutions of Hamilton-Jacobi-Belmann Equations

    Bardi, M. & Capuzzo-Dolcetta, I., Optimal Control and Viscosity Solutions of Hamilton-Jacobi-Belmann Equations. Springer Science+Business Media, LLC, New York, 1997

    work page 1997

  3. [3]

    & Olsder G.J., Dynamic noncooperative game theory

    Ba¸ sar T. & Olsder G.J., Dynamic noncooperative game theory. SIAM, Philadel- phia, (1999)

    work page 1999

  4. [4]

    & Zaccour, G., Nonzero-Sum Differential Games, in Hand- book of Differential Games, Ba¸ sar T

    Ba¸ sar T., Haurie A. & Zaccour, G., Nonzero-Sum Differential Games, in Hand- book of Differential Games, Ba¸ sar T. and Zaccour, G., eds., Sp ringer, 2018

    work page 2018

  5. [5]

    Garcke, J., Griebel, M

    Bokanowski, O.,. Garcke, J., Griebel, M. & Klompmaker, I ., An adaptive sparse grid semi-Lagrangian scheme for first order Hamilton-Jacob i Bellman equa- tions, J. Sci. Comput. 55, 2013, pp. 575–605

    work page 2013

  6. [6]

    & Ferretti, R., An efficient algorithm for Hamilton- Jacobi equations in high dimension , Comput

    Carlini, E., Falcone, M. & Ferretti, R., An efficient algorithm for Hamilton- Jacobi equations in high dimension , Comput. and Visualization in Science, 7, 2004, pp. 15–29

    work page 2004

  7. [7]

    , Van Long, N

    Dockner, E., Jørgensen, S. , Van Long, N. & Sorger, G. Differential Games in Economics and managenement Science. Cambridge University Press, Cam- bridge, 2000

    work page 2000

  8. [8]

    Ap- pendix A in M

    Falcone, M., Numerical Solution of Dynamic Programming Equations. Ap- pendix A in M. Bardi & I. Capuzzo-Dolcetta. Optimal Control a nd Viscosity Solutions of Hamilton-Jacobi-Belmann Equations. Springe r Science+Business Media, LLC, New York, 1997

    work page 1997

  9. [9]

    & Ferreti, R., Discrete time High-Order schemes for Viscosity So- lutions of Hamilton-Jacobi-Bellman Equations Numerische Mathenmatik, 67, 1994, pp

    Falcone, M. & Ferreti, R., Discrete time High-Order schemes for Viscosity So- lutions of Hamilton-Jacobi-Bellman Equations Numerische Mathenmatik, 67, 1994, pp. 315–344

    work page 1994

  10. [10]

    & Mart ´ ın-Herr´ an G., Does flexibility facilitate sustainability of cooperation over time? A case study from environmental econ omics

    De Frutos J. & Mart ´ ın-Herr´ an G., Does flexibility facilitate sustainability of cooperation over time? A case study from environmental econ omics. Journal of Optimization Theory and Applications 165, 2015, pp. 657– 677. 10

    work page 2015

  11. [11]

    & a Mart ´ ın-Herr´ an, G.,Selection of a Markov perfect Nash equi- librium in a class of differential games

    De Frutos, J. & a Mart ´ ın-Herr´ an, G.,Selection of a Markov perfect Nash equi- librium in a class of differential games . Dynamic Games and Applications, 8, 2018, pp. 620–636,

    work page 2018

  12. [12]

    Journal of Environmental Economics and Management 97, 2019, pp

    De Frutos J., Mart ´ ın-Herr´ an G.,Spatial effects and strategic behavior in a mul- tiregional transboundary pollution game . Journal of Environmental Economics and Management 97, 2019, pp. 182–207

    work page 2019

  13. [13]

    & Wu, T.T., Approximation of optimal feedback control: a dynamic programming approach, J

    Guo, B.Z. & Wu, T.T., Approximation of optimal feedback control: a dynamic programming approach, J. Global Optim. 46, 2010, pp. 395–422

    work page 2010

  14. [14]

    & Zaccour, G., Games and dynamic games

    Haurie, A., Krawczyk, J.B. & Zaccour, G., Games and dynamic games . World Scientific, Singapore, 2012

    work page 2012

  15. [15]

    Krawczyk, J.B. & V. Petkov, Multistage Games , in Handbook of Dynamic Game Theory, T. Ba¸ sar and G. Zaccour , eds., Springer Nature , 2018. 11

    work page 2018