On the unimportance of distant players in sparse stochastic differential network games

Davide Francesco Redaelli; Marco Cirant

arxiv: 2604.27157 · v1 · submitted 2026-04-29 · 🧮 math.AP · math.OC· math.PR

On the unimportance of distant players in sparse stochastic differential network games

Marco Cirant , Davide Francesco Redaelli This is my paper

Pith reviewed 2026-05-07 08:15 UTC · model grok-4.3

classification 🧮 math.AP math.OCmath.PR

keywords stochastic differential gamessparse networksgraph distanceopen-loop strategiesapproximation errormonotonicityconvexitynetwork games

0 comments

The pith

In sparse stochastic differential network games, a player's optimal open-loop strategy can be approximated with arbitrary accuracy by solving a reduced game that includes only players up to a sufficient graph distance.

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

The paper establishes that in large-population stochastic differential games where interactions occur along sparse graphs with bounded degree, such as chains or lattices, the influence of distant players on a given agent's optimal trajectory becomes negligible. Specifically, for any desired accuracy, there exists a finite radius R such that considering only the subgame within distance R and assigning arbitrary paths to outsiders yields a strategy within the error bound. This approximation holds with an explicit estimate depending on R but independent of the time horizon, relying on convexity and monotonicity of the running and terminal costs. The same truncation principle applies to distributed strategies, suggesting that global network effects are localized in these settings. This reduction is particularly useful for analyzing or computing equilibria in very large or infinite networks without solving the full system.

Core claim

We show that, under suitable convexity and monotonicity assumptions on the costs, for any ε > 0 there is a distance R such that the optimal open-loop trajectory of a player in the N-player game on a sparse graph differs by at most ε (in appropriate norms) from the optimal trajectory in the reduced game consisting of players within graph distance R, where players beyond R are assigned arbitrary trajectories. The estimate is non-asymptotic and uniform with respect to the time horizon T and the total number of players N (provided N is large enough relative to R). Analogous results hold when restricting to distributed strategies.

What carries the argument

The graph-distance truncation operator on the interaction network, which isolates a local finite-player subgame whose optimality conditions are shown to be close to those of the full game via stability estimates derived from the monotonicity and convexity assumptions.

If this is right

The approximation error decreases as the truncation radius increases, with an explicit rate.
The result is valid for arbitrarily long time horizons, unlike many asymptotic analyses.
Reduced games remain of bounded size in sparse graphs, facilitating numerical solutions as population size grows.
The unimportance extends to distributed (closed-loop) strategies under similar conditions.
In the limit of infinite graphs, local solutions can be computed without reference to the global structure.

Where Pith is reading between the lines

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

This locality principle could be tested in concrete models like traffic flow or opinion dynamics on networks by comparing full and truncated simulations.
The approach may generalize to other dynamics, such as mean-field games with local interactions, if monotonicity is preserved.
For control design in large-scale systems, this implies that decentralized controllers need only local information up to a quantifiable range.
One could derive quantitative bounds on the required communication radius in networked control systems from similar arguments.

Load-bearing premise

The cost functions are convex in the state-control variables and monotone with respect to the empirical measure of other players' states, ensuring that perturbations from distant agents do not amplify through the optimality conditions.

What would settle it

For a fixed chain graph with N=20 players and quadratic costs, solve the full open-loop Nash equilibrium numerically, then solve the truncated game with R=2 and R=5 assigning random trajectories to outsiders; if the difference in the central player's trajectory does not decrease significantly with larger R as predicted by the bound, the main result is falsified.

Figures

Figures reproduced from arXiv: 2604.27157 by Davide Francesco Redaelli, Marco Cirant.

**Figure 3.1.** Figure 3.1: The two forms of directed lattice considered in Example 3.3—the former corresponds to the choice (3.3) and the latter is obtained after the change (3.4) with h = 2. Proof of Theorem 2.3. Let ν ∈ C 0 ([0, T];P2((R d ) N × (R d ) I ) solve ∂tν − 1 2 X j∈[[N]] tr(ΣjDxjxj ν) − 1 2 X k∈I tr(ΣkDxˆ kxˆ k ν) − X k∈I tr(ΣkDxkxˆ k ν) − X j∈[[N]] divxj view at source ↗

read the original abstract

We study stochastic differential games with $N$ players, where interactions are determined by sequences of graphs in which the number of neighbours of each node remains bounded as $N$ grows, such as chain graphs or lattices. Our main goal is to quantify the phenomenon of the "unimportance of distant players" in such a large population, sparse regime: we show that, in order to determine the optimal trajectory in open-loop strategies of a given player with an arbitrarily small error, it suffices to consider a reduced game involving only the players at a certain distance in the graph, assigning arbitrary trajectories to the farther ones. Our main result provides an explicit non-asymptotic estimate in terms of the graph distance, valid independently of the time horizon $T$, under suitable convexity and monotonicity assumptions on the costs. Similar results are obtained for games in distributed strategies.

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

This paper delivers a T-independent explicit bound on the unimportance of distant players in sparse stochastic differential games.

read the letter

The paper gives an explicit bound showing that distant players on a sparse graph have limited impact on a given player's optimal open-loop strategy in a stochastic differential game, and this bound holds uniformly in the time horizon. What stands out is the non-asymptotic character of the result. The authors derive an error estimate between the full-game optimum and the reduced-game optimum (where far players get arbitrary controls) that decays with graph distance d, typically exponentially, without depending on T or total population size N. This comes from a local energy estimate on the difference of trajectories, combined with a recursion over graph distance. Monotonicity of the costs supplies a dissipative effect that counters the usual Gronwall inequality growth, keeping everything T-independent. The bounded-neighbor condition keeps the local games finite and uniform. The argument checks out under the stated assumptions. No circular reasoning appears; the bound follows directly from the cost structure and graph properties. The main limitation is the reliance on convexity and monotonicity, which are standard but exclude some interesting non-monotone interactions. It would strengthen the paper to include a simple example, such as a chain of players with quadratic costs, to illustrate how the constant in the bound behaves. The section on distributed strategies follows the same lines but could use more explicit statements on the information constraints. This work is aimed at researchers in stochastic games and mean-field theory who deal with large but sparse networks. Readers interested in practical approximations for control problems on graphs will find the explicit estimate useful. It is worth sending to peer review, as the explicit bound represents a clear step beyond typical asymptotic analyses in the field.

Referee Report

0 major / 3 minor

Summary. The manuscript studies stochastic differential games with N players whose interactions are governed by sequences of graphs with uniformly bounded degree (e.g., chains or lattices). The central claim is that, under convexity and monotonicity assumptions on the running and terminal costs, the open-loop optimal trajectory of any fixed player can be approximated to arbitrary accuracy by the optimum of a reduced game that retains only the players lying within graph distance d of the given player and assigns arbitrary admissible controls to all farther players. An explicit, non-asymptotic error bound of the form C ρ^d (with ρ < 1 independent of both T and N) is derived; an analogous statement is proved for distributed (closed-loop) strategies.

Significance. If the stated error bound holds, the result supplies a practical, finite-N truncation principle for large sparse network games whose computational cost remains uniform in the population size and does not deteriorate with the time horizon. The proof technique—local energy estimates combined with a graph-distance recursion in which monotonicity supplies a dissipative term that absorbs the usual Gronwall growth—is noteworthy and yields a bound that is genuinely parameter-free with respect to T. This advances the literature on network mean-field games by providing explicit locality estimates without passage to the continuum limit.

minor comments (3)

[§2.1] §2.1, Definition 2.3: the precise meaning of “bounded-neighbor graph condition” (uniform bound on degree independent of N) should be stated as a numbered assumption rather than left implicit in the text preceding the main theorem.
[Theorem 3.1] Theorem 3.1: the constant C in the error bound is asserted to depend only on the monotonicity and convexity moduli; an explicit expression (or at least a clear list of the quantities on which C depends) would make the non-asymptotic character of the result easier to verify.
[§4.3] §4.3, proof of the distributed-strategy case: the argument invokes the same energy estimate as the open-loop case but with an additional Itô correction term; a short remark clarifying why this term is controlled by the same monotonicity constant would improve readability.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive and accurate summary of our work on sparse stochastic differential network games, as well as for highlighting the significance of the explicit non-asymptotic error bounds independent of T and the local energy estimate technique. We appreciate the recommendation for minor revision and will address any editorial or presentational suggestions in the revised manuscript.

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained from assumptions

full rationale

The central result is an explicit non-asymptotic error bound between the full-game open-loop optimum and a reduced-game optimum truncated at graph distance d. The proof proceeds by constructing a local energy estimate on trajectory differences, recursing along graph distance, and using monotonicity to produce a dissipative term that absorbs Gronwall growth, yielding a factor C ρ^d independent of T and N. The bounded-neighbor condition ensures only finitely many interaction edges enter the cost, controlled directly by the same monotonicity. No parameter is fitted to data and then renamed a prediction, no self-citation supplies a load-bearing uniqueness theorem or ansatz, and the derivation does not reduce to its inputs by construction. The argument relies only on the stated convexity/monotonicity and graph sparsity; it is externally falsifiable via the explicit bound and does not invoke prior work by the same authors in a circular manner.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Result depends on these standard assumptions from stochastic game theory; no additional free parameters or new entities.

axioms (2)

domain assumption Costs are convex and monotone
Necessary for the error bound to hold independently of T.
domain assumption Interaction graphs have uniformly bounded degree
Ensures sparsity so distant players can be ignored.

pith-pipeline@v0.9.0 · 10720 in / 1136 out tokens · 107904 ms · 2026-05-07T08:15:35.948653+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

[1]

Aurell, R

A. Aurell, R. Carmona, and M. Lauri\` e re , Stochastic graphon games: II. The linear-quadratic case , Appl. Math. Optim., 85 (2022), Article No. 39

work page 2022
[2]

Baldi , Stochastic calculus

P. Baldi , Stochastic calculus. An introduction through theory and exercises , Universitext, Springer, Cham, 2017

work page 2017
[3]

Bayraktar, R

E. Bayraktar, R. Wu, and X. Zhang , Propagation of chaos of forward-backward stochastic differential equations with graphon interactions , Appl. Math. Optim., 88 (2023), Article No. 25

work page 2023
[4]

P. E. Caines , Embedded vertexon-graphons and embedded GMFG systems , in 2022 IEEE 61st Conference on Decision and Control, pp. 5550--5557

work page 2022
[5]

P. E. Caines and M. Huang , Graphon mean field games and their equations , SIAM J. Control Optim., 59 (2021), pp. 4373--4399

work page 2021
[6]

P. E. Caines and M. Huang , Mean field games on dense and sparse networks: The graphexon MFG equations , in 2024 American Control Conference, pp. 4230--4235

work page 2024
[7]

P. E. Caines and M. Huang , Sparse network mean field games: Ring structures and related topologies , in 2024 IEEE 63rd Conference on Decision and Control, pp. 2584--2590

work page 2024
[8]

Cardaliaguet, F

P. Cardaliaguet, F. Delarue, J.-M. Lasry, and P.-L. Lions , The master equation and the convergence problem in mean field games , Ann. of Math. Stud., 201, Princeton University Press, Princeton, NJ, 2019

work page 2019
[9]

Cardaliaguet and A

P. Cardaliaguet and A. Porretta , An introduction to mean field game theory , Lecture Notes in Math., 2281, Springer, Cham, 2020

work page 2020
[10]

Carmona and F

R. Carmona and F. Delarue , Probabilistic theory of mean field games with applications. I: Mean field FBSDEs, control, and games , Probab. Theory Stoch. Model., 83, Springer, Cham, 2018

work page 2018
[11]

A non-asymptotic approach to stochastic differential games with many players under semi-monotonicity

M. Cirant, J. Jackson, and D. F. Redaelli , A non-asymptotic approach to stochastic differential games with many players under semi-monotonicity , arXiv:2505.01526

work page arXiv
[12]

Cirant and D

M. Cirant and D. F. Redaelli , Some remarks on Linear-Quadratic closed-loop games with many players , Dyn. Games Appl., 15 (2025), pp. 558--591

work page 2025
[13]

Cirant and D

M. Cirant and D. F. Redaelli , A priori estimates and large population limits for some nonsymmetric Nash systems with semimonotonicity , Comm. Pure Appl. Math. 79 (2026), pp. 3--88

work page 2026
[14]

Delarue , Mean field games: a toy model on an Erd o s--Renyi graph , ESAIM Proc

F. Delarue , Mean field games: a toy model on an Erd o s--Renyi graph , ESAIM Proc. Surveys, 60 (2017), pp. 1--26

work page 2017
[15]

Feng, J.-P

Y. Feng, J.-P. Fouque, and T. Ichiba , Linear-quadratic stochastic differential games on directed chain networks , J. Math. Stat. Sci., 7 (2020), pp. 25--67

work page 2020
[16]

Feng, J.-P

Y. Feng, J.-P. Fouque, and T. Ichiba , Linear-quadratic stochastic differential games on random directed networks , J. Math. Stat. Sci., 7 (2020), 79--108

work page 2020
[17]

Huang, R

M. Huang, R. P. Malham\'e, and P. E. Caines , Large population stochastic dynamic games: closed-loop McKean-Vlasov systems and the Nash certainty equivalence principle , Commun. Inf. Syst., 6 (2006), pp. 221--251

work page 2006
[18]

Lacker, K

D. Lacker, K. Ramanan, and R. Wu , Local weak convergence for sparse networks of interacting processes , Ann. Appl. Probab., 33 (2023), pp. 843--888

work page 2023
[19]

Lacker and A

D. Lacker and A. Soret , A case study on stochastic games on large graphs in mean field and sparse regimes , Math. Oper. Res., 47 (2022), pp. 1530--1565

work page 2022
[20]

Lacker and A

D. Lacker and A. Soret , A label-state formulation of stochastic graphon games and approximate equilibria on large networks , Math. Oper. Res., 48 (2023), pp. 1987--2018

work page 2023
[21]

Lasry and P.-L

J.-M. Lasry and P.-L. Lions , Mean field games , Jpn., J., Math., 2 (2007), pp. 229--260

work page 2007
[22]

Lions , Cours du Coll\` e ge de France (2008), available at https://www.college-de-france.fr/fr/agenda/cours/jeux-champ-moyen-suite-2

P.-L. Lions , Cours du Coll\` e ge de France (2008), available at https://www.college-de-france.fr/fr/agenda/cours/jeux-champ-moyen-suite-2

work page 2008
[23]

Neuman and S

E. Neuman and S. Tuschmann , Stochastic Games on Large Sparse Graphs , arXiv:2602.15557

work page arXiv
[24]

D. F. Redaelli , On nonlinear systems of PDEs arising in the theory of differential games , Doctoral Thesis, University of Padua, 2024

work page 2024
[25]

D. F. Redaelli , Short-time existence and uniqueness for some infinite-dimensional Nash systems , SIAM J. Math. Anal., 58 (2026), pp. 1978--1999

work page 2026

[1] [1]

Aurell, R

A. Aurell, R. Carmona, and M. Lauri\` e re , Stochastic graphon games: II. The linear-quadratic case , Appl. Math. Optim., 85 (2022), Article No. 39

work page 2022

[2] [2]

Baldi , Stochastic calculus

P. Baldi , Stochastic calculus. An introduction through theory and exercises , Universitext, Springer, Cham, 2017

work page 2017

[3] [3]

Bayraktar, R

E. Bayraktar, R. Wu, and X. Zhang , Propagation of chaos of forward-backward stochastic differential equations with graphon interactions , Appl. Math. Optim., 88 (2023), Article No. 25

work page 2023

[4] [4]

P. E. Caines , Embedded vertexon-graphons and embedded GMFG systems , in 2022 IEEE 61st Conference on Decision and Control, pp. 5550--5557

work page 2022

[5] [5]

P. E. Caines and M. Huang , Graphon mean field games and their equations , SIAM J. Control Optim., 59 (2021), pp. 4373--4399

work page 2021

[6] [6]

P. E. Caines and M. Huang , Mean field games on dense and sparse networks: The graphexon MFG equations , in 2024 American Control Conference, pp. 4230--4235

work page 2024

[7] [7]

P. E. Caines and M. Huang , Sparse network mean field games: Ring structures and related topologies , in 2024 IEEE 63rd Conference on Decision and Control, pp. 2584--2590

work page 2024

[8] [8]

Cardaliaguet, F

P. Cardaliaguet, F. Delarue, J.-M. Lasry, and P.-L. Lions , The master equation and the convergence problem in mean field games , Ann. of Math. Stud., 201, Princeton University Press, Princeton, NJ, 2019

work page 2019

[9] [9]

Cardaliaguet and A

P. Cardaliaguet and A. Porretta , An introduction to mean field game theory , Lecture Notes in Math., 2281, Springer, Cham, 2020

work page 2020

[10] [10]

Carmona and F

R. Carmona and F. Delarue , Probabilistic theory of mean field games with applications. I: Mean field FBSDEs, control, and games , Probab. Theory Stoch. Model., 83, Springer, Cham, 2018

work page 2018

[11] [11]

A non-asymptotic approach to stochastic differential games with many players under semi-monotonicity

M. Cirant, J. Jackson, and D. F. Redaelli , A non-asymptotic approach to stochastic differential games with many players under semi-monotonicity , arXiv:2505.01526

work page arXiv

[12] [12]

Cirant and D

M. Cirant and D. F. Redaelli , Some remarks on Linear-Quadratic closed-loop games with many players , Dyn. Games Appl., 15 (2025), pp. 558--591

work page 2025

[13] [13]

Cirant and D

M. Cirant and D. F. Redaelli , A priori estimates and large population limits for some nonsymmetric Nash systems with semimonotonicity , Comm. Pure Appl. Math. 79 (2026), pp. 3--88

work page 2026

[14] [14]

Delarue , Mean field games: a toy model on an Erd o s--Renyi graph , ESAIM Proc

F. Delarue , Mean field games: a toy model on an Erd o s--Renyi graph , ESAIM Proc. Surveys, 60 (2017), pp. 1--26

work page 2017

[15] [15]

Feng, J.-P

Y. Feng, J.-P. Fouque, and T. Ichiba , Linear-quadratic stochastic differential games on directed chain networks , J. Math. Stat. Sci., 7 (2020), pp. 25--67

work page 2020

[16] [16]

Feng, J.-P

Y. Feng, J.-P. Fouque, and T. Ichiba , Linear-quadratic stochastic differential games on random directed networks , J. Math. Stat. Sci., 7 (2020), 79--108

work page 2020

[17] [17]

Huang, R

M. Huang, R. P. Malham\'e, and P. E. Caines , Large population stochastic dynamic games: closed-loop McKean-Vlasov systems and the Nash certainty equivalence principle , Commun. Inf. Syst., 6 (2006), pp. 221--251

work page 2006

[18] [18]

Lacker, K

D. Lacker, K. Ramanan, and R. Wu , Local weak convergence for sparse networks of interacting processes , Ann. Appl. Probab., 33 (2023), pp. 843--888

work page 2023

[19] [19]

Lacker and A

D. Lacker and A. Soret , A case study on stochastic games on large graphs in mean field and sparse regimes , Math. Oper. Res., 47 (2022), pp. 1530--1565

work page 2022

[20] [20]

Lacker and A

D. Lacker and A. Soret , A label-state formulation of stochastic graphon games and approximate equilibria on large networks , Math. Oper. Res., 48 (2023), pp. 1987--2018

work page 2023

[21] [21]

Lasry and P.-L

J.-M. Lasry and P.-L. Lions , Mean field games , Jpn., J., Math., 2 (2007), pp. 229--260

work page 2007

[22] [22]

Lions , Cours du Coll\` e ge de France (2008), available at https://www.college-de-france.fr/fr/agenda/cours/jeux-champ-moyen-suite-2

P.-L. Lions , Cours du Coll\` e ge de France (2008), available at https://www.college-de-france.fr/fr/agenda/cours/jeux-champ-moyen-suite-2

work page 2008

[23] [23]

Neuman and S

E. Neuman and S. Tuschmann , Stochastic Games on Large Sparse Graphs , arXiv:2602.15557

work page arXiv

[24] [24]

D. F. Redaelli , On nonlinear systems of PDEs arising in the theory of differential games , Doctoral Thesis, University of Padua, 2024

work page 2024

[25] [25]

D. F. Redaelli , Short-time existence and uniqueness for some infinite-dimensional Nash systems , SIAM J. Math. Anal., 58 (2026), pp. 1978--1999

work page 2026