Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.36518

Detailed balance and phase transition in Nash equilibria of mean-field Ising game

Ken Hiura·Hajime Koike·Kiyoshi Kanazawa
View original source ↗Read source document ↗Export metadata ↓

Abstract

We study a finite-state mean-field game with entropic control cost. In the large population limit, we show that its stationary Nash equilibria satisfy the detailed balance condition and their distributions take the Gibbsian form with the value function playing the role of a Hamiltonian. We apply the entropic control cost to a ferromagnetic mean-field Ising game and investigate its stationary and non-stationary Nash equilibria. We find that the stationary Nash equilibria exhibit a second-order phase transition from disordered to ordered equilibria as temperature decreases. A center-manifold reduction yields non-stationary Nash equilibrium near the critical temperature and reveals the critical slowing down in the relaxation toward the stationary Nash equilibrium.

Explore related subjects

math.OCcond-mat.stat-mechphysics.soc-ph

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Ken Hiura, Hajime Koike, Kiyoshi Kanazawa. 2026-10-27. Detailed balance and phase transition in Nash equilibria of mean-field Ising game. https://arxiv.org/abs/2610.36518

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

NAG-GS: A Semi-Implicit Stochastic Optimizer with Large-Step Stability

Stochastic first-order methods are widely used for large-scale optimization and machine learning, where their performance often depends on the learning rate. We study NAG-GS, a stochastic momentum method based on a semi-implicit Gauss--Seidel discretization of a Nesterov-type flow. For strongly convex quadratics, we determine the exact range of stable step sizes for fixed damping and the step size that minimizes the worst-case spectral radius. When the damping equals the strong-convexity parameter, the method recovers the accelerated asymptotic factor $(\sqrtκ-1)/(\sqrtκ+1)$, where $κ$ is the condition number. We also prove that the quadratic iteration converges linearly with dynamic damping when its limiting fixed-damping iteration is stable. For general strongly convex, possibly nonsmooth objectives, we establish an $O(1/k)$ bound on the expected objective gap of the last iterate, using a conditionally unbiased stochastic first-order oracle with bounded second moment. The result holds with dynamic damping from any $γ_0>0$. We show that NAG-GS can also be written as a Heavy-Ball iteration with explicitly determined step sizes and momentum coefficients. We test the stochastic result on nonsmooth hinge-loss classification problems, and further evaluate NAG-GS on quadratic problems, logistic regression, convolutional neural networks, RoBERTa, and Vision Transformers. The experiments show that NAG-GS can remain effective at learning rates where standard optimizers become unstable or lose accuracy.

math.OC↗
arXiv2026

Perturbed Gradient Descent Algorithms are Small-Disturbance Input-to-State Stable

This article investigates the robustness of gradient descent algorithms under perturbations. The concept of small-disturbance input-to-state stability (ISS) for discrete-time nonlinear dynamical systems is introduced, along with its Lyapunov characterization. The conventional linear Polyak-Lojasiewicz (PL) condition is then extended to a nonlinear version, and it is shown that the gradient descent algorithm is small-disturbance ISS provided the objective function satisfies the generalized nonlinear PL condition. This small-disturbance ISS property guarantees that the gradient descent algorithm converges to a small neighborhood of the optimum under sufficiently small perturbations. As a direct application of the developed framework, we demonstrate that the LQR cost satisfies the generalized nonlinear PL condition, thereby establishing that the policy gradient algorithm for LQR is small-disturbance ISS. Additionally, other popular policy gradient algorithms, including natural policy gradient and Gauss-Newton method, are also proven to be small-disturbance ISS.

math.OC↗
arXiv2026

Data-Driven Distributionally Robust Planning of Coastal Distribution Networks and PV-Storage-EV Stations for Low-Carbon Energy Transition

Coastal distribution networks (DNs) are becoming key infrastructure for the low-carbon energy transition as electric vehicle (EV) charging, photovoltaic (PV) generation, energy storage, and marine renewable procurement are increasingly coupled at the planning stage. Existing planning models, however, usually treat network expansion, PV-storage-EV station (PSES) deployment, low-carbon operation, and coastal asset degradation separately. This paper proposes a data-driven distributionally robust planning framework for coastal DNs and PSESs. The framework jointly optimizes radial network topology and PSES siting while incorporating salt-spray degradation costs, carbon-emission-flow (CEF)-based nodal carbon accounting, and tidal-generation-driven procurement costs. To represent heterogeneous uncertainty, the model combines continuous box uncertainty for load and PV availability with probability ambiguity over EV-charging scenarios generated from historical charging profiles. The resulting two-stage distributionally robust optimization problem is reformulated through exact relaxation of ESS mode binaries and KKT-based decomposition, and is solved by an inexact column-and-constraint generation (i-C&CG) algorithm. Case studies on a realistic 47-node coastal feeder show that salt-spray exposure can alter the selected topology and PSES siting, while the proposed planning scheme improves voltage quality, reduces annualized total cost, and reveals the cost--carbon trade-off of tidal procurement. The tailored i-C&CG algorithm obtains the same optimal design as conventional C&CG with about an order-of-magnitude reduction in solution time.

math.OC↗