Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Engineering Social Optimality via Utility Shaping in Non-Cooperative Games under Incomplete Information and Imperfect Monitoring

David Smith, Jie Dong, Yizhou Yang

TL;DR

The paper tackles decentralized decision-making under incomplete information and imperfect public monitoring. It introduces a message-free blueprint that engineers social optimality by embedding shadow prices or KKT-aligned penalties into private utilities, turning the stage game into an exact potential game whose unique Nash equilibrium aligns with the planner’s welfare $W(p)$ (including constraints). A curvature condition based on a single-inflection compressed/stretched exponential response ensures strong monotonicity and contraction, enabling a Bayesian equilibrium characterized as a strongly-monotone SVI; with noisy updates and drift, explicit tracking bounds are derived. Two computational studies—multi-tier supply chains and a non-cooperative agentic-AI compute market—demonstrate near-centralized welfare, elimination of steady-state constraint violations, and faster convergence compared to baselines, even under discrete quantization. The proposed deployable rules provide a practical, scalable path to achieving social optima with limited messaging, broad applicability across demand response, cloud/edge scheduling, transportation pricing, and biosecurity/agriculture, while acknowledging scope, limitations, and guardrails for robust deployment.

Abstract

In this paper, we study decentralized decision-making where agents optimize private objectives under incomplete information and imperfect public monitoring, in a non-cooperative setting. By shaping utilities-embedding shadow prices or Karush-Kuhn-Tucker(KKT)-aligned penalties-we make the stage game an exact-potential game whose unique equilibrium equals the (possibly constrained) social optimum. We characterize the Bayesian equilibrium as a stochastic variational inequality; strong monotonicity follows from a single-inflection compressed/stretched-exponential response combined with convex pricing. We give tracking bounds for damped-gradient and best-response-with-hysteresis updates under a noisy public index, and corresponding steady-state error. The framework accommodates discrete and continuous action sets and composes with slower discrete assignment. Deployable rules include: embed prices/penalties; publish a single public index; tune steps, damping, and dual rates for contraction. Computational experiments cover (i) a multi-tier supply chain and (ii) a non-cooperative agentic-AI compute market of bidding bots. Relative to price-only baselines, utility shaping attains near-centralized welfare, eliminates steady-state constraint/capacity violations when feasible, and accelerates convergence; with quantization, discrete equilibria track continuous ones within the mesh. The blueprint is portable to demand response, cloud/edge scheduling, and transportation pricing and biosecurity/agriculture. Overall, utility shaping plus a public index implements the constrained social optimum with stable equilibria under noise and drift-an operations-research-friendly alternative to heavy messaging or full mechanism design.

Engineering Social Optimality via Utility Shaping in Non-Cooperative Games under Incomplete Information and Imperfect Monitoring

TL;DR

The paper tackles decentralized decision-making under incomplete information and imperfect public monitoring. It introduces a message-free blueprint that engineers social optimality by embedding shadow prices or KKT-aligned penalties into private utilities, turning the stage game into an exact potential game whose unique Nash equilibrium aligns with the planner’s welfare (including constraints). A curvature condition based on a single-inflection compressed/stretched exponential response ensures strong monotonicity and contraction, enabling a Bayesian equilibrium characterized as a strongly-monotone SVI; with noisy updates and drift, explicit tracking bounds are derived. Two computational studies—multi-tier supply chains and a non-cooperative agentic-AI compute market—demonstrate near-centralized welfare, elimination of steady-state constraint violations, and faster convergence compared to baselines, even under discrete quantization. The proposed deployable rules provide a practical, scalable path to achieving social optima with limited messaging, broad applicability across demand response, cloud/edge scheduling, transportation pricing, and biosecurity/agriculture, while acknowledging scope, limitations, and guardrails for robust deployment.

Abstract

In this paper, we study decentralized decision-making where agents optimize private objectives under incomplete information and imperfect public monitoring, in a non-cooperative setting. By shaping utilities-embedding shadow prices or Karush-Kuhn-Tucker(KKT)-aligned penalties-we make the stage game an exact-potential game whose unique equilibrium equals the (possibly constrained) social optimum. We characterize the Bayesian equilibrium as a stochastic variational inequality; strong monotonicity follows from a single-inflection compressed/stretched-exponential response combined with convex pricing. We give tracking bounds for damped-gradient and best-response-with-hysteresis updates under a noisy public index, and corresponding steady-state error. The framework accommodates discrete and continuous action sets and composes with slower discrete assignment. Deployable rules include: embed prices/penalties; publish a single public index; tune steps, damping, and dual rates for contraction. Computational experiments cover (i) a multi-tier supply chain and (ii) a non-cooperative agentic-AI compute market of bidding bots. Relative to price-only baselines, utility shaping attains near-centralized welfare, eliminates steady-state constraint/capacity violations when feasible, and accelerates convergence; with quantization, discrete equilibria track continuous ones within the mesh. The blueprint is portable to demand response, cloud/edge scheduling, and transportation pricing and biosecurity/agriculture. Overall, utility shaping plus a public index implements the constrained social optimum with stable equilibria under noise and drift-an operations-research-friendly alternative to heavy messaging or full mechanism design.

Paper Structure

This paper contains 52 sections, 13 theorems, 29 equations, 4 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Organization.
  3. Related Work
  4. Pricing, decentralization, and proportional fairness
  5. Power control, response curves, and monotone games
  6. Variational inequalities, stochastic programs, and discrete convexity
  7. Problem description
  8. Methodology
  9. Physics-informed response and utility shaping
  10. Information about these prior mechanisms.
  11. Curvature certificate and properties
  12. Bayesian equilibrium as an SVI
  13. Repeated play and tracking under imperfect monitoring
  14. Computational experiments
  15. Multi-tier supply chains
  16. ...and 37 more sections

Key Result

Theorem 1

Let $X=\prod_{i=1}^N X_i$ be convex and compact. Consider Eq. (4) with $v_i$ concave and increasing, $c_i$ convex, and define the welfare/potential Then:

Figures (4)

  • Figure 1: Message-free loop: a single public index focalizes decentralized responses; private payoffs are shaped with KKT-aligned penalties so selfish updates implement the planner’s first-order conditions.
  • Figure 2: Representative reliability curves for exponentials used in supply-chain experiments and Agentic AI experiment
  • Figure 3: Supply-chain welfare gap to centralized $W^*$ (mean over 100 runs, 500 iterations). Unified planner welfare; methods share identical iteration/compute budgets. Dashed line at 0 marks centralized optimum, $\kappa=3,\beta=0.8$ in $y(\cdot)=\exp(-(\kappa/\cdot)^\beta)$
  • Figure 4: Agentic-AI experiment (100-run average; unified planner welfare). Welfare gap to centralized $W^*$ for utility shaping (ours) and price-only (agents; no shaping). Dashed line marks zero gap.

Theorems & Definitions (38)

  • Theorem 1: Exact potential & uniqueness for Eq. (4)
  • proof
  • Proposition 1: Strict concavity and discrete robustness for Eq. (5)
  • proof
  • Remark 1: Operational takeaway: implementing the (possibly constrained) social optimum, under the stated curvature/monotonicity and convexity assumptions
  • Proposition 2
  • proof
  • Proposition 3: Sigmoidal certificate $\Rightarrow$ exact potential & unique NE
  • proof
  • Conjecture 1: Minimality/necessity for social optimality
  • ...and 28 more