Neuromorphic Realization of Best Response in Finite-Action Games

Himani Sinhmar; Naomi Ehrich Leonard; Vaibhav Srivastava

arxiv: 2604.03222 · v3 · submitted 2026-04-03 · 🧮 math.DS · math.OC

Neuromorphic Realization of Best Response in Finite-Action Games

Himani Sinhmar , Vaibhav Srivastava , Naomi Ehrich Leonard This is my paper

Pith reviewed 2026-05-13 18:26 UTC · model grok-4.3

classification 🧮 math.DS math.OC

keywords neuromorphic dynamicsbest responsefinite-action gamescirculant couplingLyapunov-Schmidt reductiongeometry-aware utilitypotential gameshysteresis

0 comments

The pith

For circulant action couplings, neuromorphic dynamics implicitly compute a geometry-aware utility and converge exponentially to best response at a rate independent of action count.

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

The paper proposes neuromorphic decision dynamics that realize best response in finite-action games as the stable outcome of an internal state-space process instead of an imposed rule. This formulation explains commitment formation, symmetry breaking through attraction basins, and hysteresis under perturbations. When the action coupling is circulant, Lyapunov-Schmidt reduction proves that the dynamics compute a geometry-aware utility, converge exponentially to the best response at a rate independent of how many actions exist, and only switch on strong evidence. In a coverage game example the resulting game is exactly potential and the dynamics reach its Nash equilibria, while logit dynamics with the same utility fail to show these traits.

Core claim

What carries the argument

Neuromorphic decision dynamics that embed the action-coupling operator into a state-space process, with Lyapunov-Schmidt reduction used to isolate governing evidence components when the operator is circulant.

Load-bearing premise

The action-coupling operator must be circulant or admit a Lyapunov-Schmidt reduction that isolates the governing evidence components.

What would settle it

A simulation or proof showing that convergence rate to best response varies with the number of actions under circulant coupling, or that the dynamics switch without sufficiently strong evidence, would disprove the claims.

Figures

Figures reproduced from arXiv: 2604.03222 by Himani Sinhmar, Naomi Ehrich Leonard, Vaibhav Srivastava.

**Figure 1.** Figure 1: Bifurcation geometry and action-selection in the critical eigenspace Ec = span{ϕk⋆ , ψk⋆ }. (A) At the onset α = αc, a supercritical pitchfork generates a continuum of equilibria parameterized by (r ⋆, θ⋆), forming a ring in Ec (blue) of phase-indeterminate committed states along the stable branch r = r ⋆. (B) In the presence of an input b (purple), its projection ΨA(b) onto Ec (magenta) selects a unique e… view at source ↗

**Figure 2.** Figure 2: (A) Raw event density V (k, 0) on the action ring. (B) Mexican-hat action-coupling matrix A with dominant eigenmode k⋆ = 1 [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Figure 3: Neuromorphic decision dynamics with subcritical NOD vs. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

read the original abstract

We develop a mechanistic dynamical-systems formulation of best response in finite-action games with relational structure on the action set. The proposed neuromorphic decision dynamics realize best response as the stable outcome of an internal state-space process, rather than as an externally imposed choice rule. This provides a deterministic account of commitment formation, symmetry resolution through basins of attraction, and hysteresis and decision persistence under perturbations. For action spaces with circulant coupling, we prove using Lyapunov-Schmidt reduction that the action-coupling operator determines which components of evidence govern decision formation. We further show that the dynamics implicitly compute a geometry-aware utility, converge exponentially to the corresponding best response with rate independent of the number of actions, and switch only when evidence is sufficiently strong. In contrast, supplying the same geometry-aware utility directly to logit dynamics does not recover these properties, showing that relational structure must be embedded in the decision mechanism itself. We illustrate the framework in a repeated coverage game, prove that the induced game is an exact potential game, and show that its Nash equilibria are reached by the neuromorphic dynamics.

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 paper gives a clean dynamical realization of best response that builds in relational structure and delivers rate-independent exponential convergence for circulant games, but the circulant restriction is load-bearing and the generality claim is thin.

read the letter

The main contribution is a state-space model where best response emerges from internal dynamics rather than being imposed as a rule. For circulant action coupling the authors use Lyapunov-Schmidt reduction to show that the system implicitly computes a geometry-aware utility, converges exponentially to the best response at a rate independent of the number of actions, and exhibits hysteresis. They also show that feeding the same utility into standard logit dynamics loses these features, which is a useful contrast. The coverage-game example is worked out cleanly and shown to be an exact potential game whose equilibria are reached by the dynamics. That part is solid and worth knowing if you work on multi-agent systems or neuromorphic hardware implementations of games. The proofs appear to rest on standard reduction techniques applied to a circulant operator, and the abstract states the rate-independence result explicitly. The stress-test worry about N-dependent corrections in the nonlinear terms is reasonable in principle, but the paper claims to have ruled it out via the reduction, so the central claim holds under the stated assumptions. The soft spot is that everything after the circulant case is left as future work; without that assumption the reduction does not isolate the evidence components in the same way and the rate-independence is not guaranteed. The paper is therefore mainly for people already working at the intersection of dynamical systems and game theory who want a mechanistic model rather than a black-box choice rule. It is not broad enough or general enough to be a must-read outside that niche, but the formal results are sharp enough that a serious editor should send it to referees rather than desk-reject it.

Referee Report

1 major / 1 minor

Summary. The manuscript develops a dynamical-systems model for realizing best response in finite-action games with relational structure on the action set. For circulant coupling, it claims to prove via Lyapunov-Schmidt reduction that the dynamics compute a geometry-aware utility, converge exponentially to best response at a rate independent of the number of actions, and switch only under sufficiently strong evidence. It contrasts this with logit dynamics and illustrates with a repeated coverage game that is shown to be an exact potential game whose Nash equilibria are reached by the dynamics.

Significance. If the central claims hold, particularly the N-independent convergence rate and the necessity of embedding relational structure in the mechanism rather than the utility, this work offers a novel mechanistic account of decision formation in structured games. It could bridge dynamical systems, game theory, and neuromorphic computing. The potential-game result strengthens the applicability.

major comments (1)

[Abstract] Abstract: The claim that the dynamics 'converge exponentially to the corresponding best response with rate independent of the number of actions' relies on the Lyapunov-Schmidt reduction yielding a reduced flow whose decay rate is uniform in N. However, for circulant operators, the quadratic and higher-order terms in the vector field may introduce resonant interactions whose coefficients scale with N through mode normalizations or the number of participating modes. The manuscript does not appear to provide explicit spectral gap estimates or remainder bounds that rule out N-dependent prefactors in the effective rate on the center manifold.

minor comments (1)

[Abstract] The abstract introduces 'geometry-aware utility' without an explicit equation or definition in the provided summary; adding a precise characterization early in the manuscript would improve readability.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful reading and constructive comments, which highlight an important technical point regarding the uniformity of the convergence rate. We address the concern below and will strengthen the manuscript accordingly.

read point-by-point responses

Referee: The claim that the dynamics 'converge exponentially to the corresponding best response with rate independent of the number of actions' relies on the Lyapunov-Schmidt reduction yielding a reduced flow whose decay rate is uniform in N. However, for circulant operators, the quadratic and higher-order terms in the vector field may introduce resonant interactions whose coefficients scale with N through mode normalizations or the number of participating modes. The manuscript does not appear to provide explicit spectral gap estimates or remainder bounds that rule out N-dependent prefactors in the effective rate on the center manifold.

Authors: We appreciate the referee's observation on potential N-dependent prefactors arising from nonlinear terms. The linear part of the circulant operator is diagonalized by the discrete Fourier basis, with eigenvalues determined solely by the fixed coupling kernel; the gap separating the center manifold from the stable subspace is therefore uniform in N. For the quadratic and higher-order terms, the circulant symmetry and Fourier orthogonality ensure that resonant interactions projecting onto the center manifold do not introduce N-dependent scaling in their coefficients. Nevertheless, we acknowledge that the original manuscript does not supply explicit remainder estimates or a uniform spectral-gap lemma. In the revised version we will add a dedicated lemma (in the section on Lyapunov-Schmidt reduction) that furnishes (i) an N-independent lower bound on the linear decay rate and (ii) explicit bounds on the projected nonlinear remainder, confirming that the effective exponential rate on the center manifold remains bounded away from zero independently of N. These additions will be purely technical and will not alter the main claims or results. revision: yes

Circularity Check

0 steps flagged

Standard Lyapunov-Schmidt reduction on circulant operators with no load-bearing self-definition or fitted predictions

full rationale

The derivation begins from an explicitly stated dynamical system and applies the standard Lyapunov-Schmidt procedure to isolate center-manifold dynamics for circulant coupling. No equation in the abstract or described chain defines a target quantity (geometry-aware utility or N-independent rate) in terms of itself or a fitted parameter that is then re-derived from the same system. The central claims rest on the circulant assumption and the reduction theorem rather than on a self-referential loop. A minor self-citation to prior dynamical-systems work by the authors is present but is not invoked as the sole justification for the uniqueness or rate-independence result; the reduction itself supplies the independent content. Hence only a low-level score is warranted.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on the existence of a circulant action-coupling operator that permits Lyapunov-Schmidt reduction, the definition of the neuromorphic vector field, and the assumption that the resulting system is an exact potential game in the coverage example. No free parameters are explicitly fitted in the abstract; the geometry-aware utility emerges from the dynamics rather than being postulated separately.

axioms (2)

domain assumption The action-coupling operator is circulant, allowing Lyapunov-Schmidt reduction to isolate governing evidence components.
Invoked to obtain the rate-independence and geometry-aware utility results.
domain assumption The induced coverage game is an exact potential game.
Used to guarantee that Nash equilibria are reached by the dynamics.

pith-pipeline@v0.9.0 · 5485 in / 1439 out tokens · 23967 ms · 2026-05-13T18:26:44.401233+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

For action spaces with circulant coupling, we prove using Lyapunov–Schmidt reduction that the action-coupling operator determines which components of evidence govern decision formation... converge exponentially to the corresponding best response with rate independent of the number of actions
IndisputableMonolith/Foundation/BranchSelection.lean branch_selection unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

the reduced equilibrium conditions g_r=0 and g_θ=0... Γ=s3/4 α_c³ λ_k⋆³

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

17 extracted references · 17 canonical work pages

[1]

Utility and mechanism design in multi-agent systems: An overview,

D. Paccagnan, R. Chandan, and J. R. Marden, “Utility and mechanism design in multi-agent systems: An overview,”Annual Reviews in Control, vol. 53, pp. 315–328, 2022

work page 2022
[2]

On best-response dynamics in potential games,

B. Swenson, R. Murray, and S. Kar, “On best-response dynamics in potential games,”SIAM Journal on Control and Optimization, vol. 56, no. 4, pp. 2734–2767, 2018

work page 2018
[3]

Quantal response equilibria for normal form games,

R. D. McKelvey and T. R. Palfrey, “Quantal response equilibria for normal form games,”Games and Economic Behavior, vol. 10, no. 1, pp. 6–38, 1995

work page 1995
[4]

The statistical mechanics of strategic interaction,

L. E. Blume, “The statistical mechanics of strategic interaction,” Games and Economic Behavior, vol. 5, no. 3, pp. 387–424, 1993

work page 1993
[5]

Conver- gence analysis of the best response algorithm for time-varying games,

Z. Wang, Y . Shen, M. M. Zavlanos, and K. H. Johansson, “Conver- gence analysis of the best response algorithm for time-varying games,” inIEEE Conference on Decision and Control, 2023, pp. 1144–1149

work page 2023
[6]

Convergence to equilibrium of logit dynamics for strategic games,

V . Auletta, D. Ferraioli, F. Pasquale, P. Penna, and G. Persiano, “Convergence to equilibrium of logit dynamics for strategic games,” inACM Symp. Parallelism in Alg. and Arch., 2011, pp. 197–206

work page 2011
[7]

A class of games possessing pure-strategy nash equilibria,

R. W. Rosenthal, “A class of games possessing pure-strategy nash equilibria,”Int. J. Game Theory, vol. 2, no. 1, pp. 65–67, 1973

work page 1973
[8]

Potential games,

D. Monderer and L. S. Shapley, “Potential games,”Games and Economic Behavior, vol. 14, no. 1, pp. 124–143, 1996

work page 1996
[9]

Nonlinear opinion dynamics with tunable sensitivity,

A. Bizyaeva, A. Franci, and N. E. Leonard, “Nonlinear opinion dynamics with tunable sensitivity,”IEEE Transactions on Automatic Control, vol. 68, no. 3, pp. 1415–1430, 2023

work page 2023
[10]

Tuning cooperative behavior in games with nonlinear opinion dynam- ics,

S. Park, A. Bizyaeva, M. Kawakatsu, A. Franci, and N. E. Leonard, “Tuning cooperative behavior in games with nonlinear opinion dynam- ics,”IEEE Control Systems Letters, vol. 6, pp. 2030–2035, 2022

work page 2030
[11]

Fast- and-flexible decision-making with modulatory interactions,

R. Moreno-Morton, A. Bizyaeva, N. E. Leonard, and A. Franci, “Fast- and-flexible decision-making with modulatory interactions,”IEEE Control Systems Letters, vol. 8, pp. 3333–3338, 2024

work page 2024
[12]

Spatially- invariant opinion dynamics on the circle,

G. Amorim, A. Bizyaeva, A. Franci, and N. E. Leonard, “Spatially- invariant opinion dynamics on the circle,”IEEE Control Systems Letters, vol. 8, pp. 3231–3236, 2024

work page 2024
[13]

Neural dynamics for landmark orientation and angular path integration,

J. D. Seelig and V . Jayaraman, “Neural dynamics for landmark orientation and angular path integration,”Nature, vol. 521, no. 7551, pp. 186–191, 2015

work page 2015
[14]

Fast and flexible multiagent decision-making,

N. E. Leonard, A. Bizyaeva, and A. Franci, “Fast and flexible multiagent decision-making,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 7, 2024

work page 2024
[15]

Toeplitz and circulant matrices: A review,

R. M. Gray, “Toeplitz and circulant matrices: A review,” 2006

work page 2006
[16]

Golubitsky, I

M. Golubitsky, I. Stewart, and D. G. Schaeffer,Singularities and groups in bifurcation theory. Springer Science & Business Media, 2012, vol. 2

work page 2012
[17]

Cooperative control and potential games,

J. R. Marden, G. Arslan, and J. S. Shamma, “Cooperative control and potential games,”IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 39, no. 6, pp. 1393–1407, 2009

work page 2009

[1] [1]

Utility and mechanism design in multi-agent systems: An overview,

D. Paccagnan, R. Chandan, and J. R. Marden, “Utility and mechanism design in multi-agent systems: An overview,”Annual Reviews in Control, vol. 53, pp. 315–328, 2022

work page 2022

[2] [2]

On best-response dynamics in potential games,

B. Swenson, R. Murray, and S. Kar, “On best-response dynamics in potential games,”SIAM Journal on Control and Optimization, vol. 56, no. 4, pp. 2734–2767, 2018

work page 2018

[3] [3]

Quantal response equilibria for normal form games,

R. D. McKelvey and T. R. Palfrey, “Quantal response equilibria for normal form games,”Games and Economic Behavior, vol. 10, no. 1, pp. 6–38, 1995

work page 1995

[4] [4]

The statistical mechanics of strategic interaction,

L. E. Blume, “The statistical mechanics of strategic interaction,” Games and Economic Behavior, vol. 5, no. 3, pp. 387–424, 1993

work page 1993

[5] [5]

Conver- gence analysis of the best response algorithm for time-varying games,

Z. Wang, Y . Shen, M. M. Zavlanos, and K. H. Johansson, “Conver- gence analysis of the best response algorithm for time-varying games,” inIEEE Conference on Decision and Control, 2023, pp. 1144–1149

work page 2023

[6] [6]

Convergence to equilibrium of logit dynamics for strategic games,

V . Auletta, D. Ferraioli, F. Pasquale, P. Penna, and G. Persiano, “Convergence to equilibrium of logit dynamics for strategic games,” inACM Symp. Parallelism in Alg. and Arch., 2011, pp. 197–206

work page 2011

[7] [7]

A class of games possessing pure-strategy nash equilibria,

R. W. Rosenthal, “A class of games possessing pure-strategy nash equilibria,”Int. J. Game Theory, vol. 2, no. 1, pp. 65–67, 1973

work page 1973

[8] [8]

Potential games,

D. Monderer and L. S. Shapley, “Potential games,”Games and Economic Behavior, vol. 14, no. 1, pp. 124–143, 1996

work page 1996

[9] [9]

Nonlinear opinion dynamics with tunable sensitivity,

A. Bizyaeva, A. Franci, and N. E. Leonard, “Nonlinear opinion dynamics with tunable sensitivity,”IEEE Transactions on Automatic Control, vol. 68, no. 3, pp. 1415–1430, 2023

work page 2023

[10] [10]

Tuning cooperative behavior in games with nonlinear opinion dynam- ics,

S. Park, A. Bizyaeva, M. Kawakatsu, A. Franci, and N. E. Leonard, “Tuning cooperative behavior in games with nonlinear opinion dynam- ics,”IEEE Control Systems Letters, vol. 6, pp. 2030–2035, 2022

work page 2030

[11] [11]

Fast- and-flexible decision-making with modulatory interactions,

R. Moreno-Morton, A. Bizyaeva, N. E. Leonard, and A. Franci, “Fast- and-flexible decision-making with modulatory interactions,”IEEE Control Systems Letters, vol. 8, pp. 3333–3338, 2024

work page 2024

[12] [12]

Spatially- invariant opinion dynamics on the circle,

G. Amorim, A. Bizyaeva, A. Franci, and N. E. Leonard, “Spatially- invariant opinion dynamics on the circle,”IEEE Control Systems Letters, vol. 8, pp. 3231–3236, 2024

work page 2024

[13] [13]

Neural dynamics for landmark orientation and angular path integration,

J. D. Seelig and V . Jayaraman, “Neural dynamics for landmark orientation and angular path integration,”Nature, vol. 521, no. 7551, pp. 186–191, 2015

work page 2015

[14] [14]

Fast and flexible multiagent decision-making,

N. E. Leonard, A. Bizyaeva, and A. Franci, “Fast and flexible multiagent decision-making,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 7, 2024

work page 2024

[15] [15]

Toeplitz and circulant matrices: A review,

R. M. Gray, “Toeplitz and circulant matrices: A review,” 2006

work page 2006

[16] [16]

Golubitsky, I

M. Golubitsky, I. Stewart, and D. G. Schaeffer,Singularities and groups in bifurcation theory. Springer Science & Business Media, 2012, vol. 2

work page 2012

[17] [17]

Cooperative control and potential games,

J. R. Marden, G. Arslan, and J. S. Shamma, “Cooperative control and potential games,”IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 39, no. 6, pp. 1393–1407, 2009

work page 2009