Abstract:We investigate first-order notions of correlated equilibria; distributions of actions for smooth, potentially non-concave games such that players do not incur any regret against small modifications to their strategies along a set of continuous vector fields. We define two such notions, based on local deviations and on stationarity of the distribution, and identify the notion of coarseness as the setting where the associated vector fields are in fact gradient fields. For coarse equilibria, we prove that online (projected) gradient decent has a universal approximation property for both variants of equilibrium. In the non-coarse setting, we instead reduce the problem of finding an equilibrium to fixed-point computation via the usual framework of $\Phi$-regret minimisation, and identify tractable instances. Finally, we study the primal-dual framework to our notion of first-order equilibria. For coarse equilibria defined by a family of functions, we find that a dual bound on the worst-case expectation of a performance metric takes the form of a generalised Lyapunov function for the dynamics of the game. Specifically, usual primal-dual price of anarchy analysis for coarse correlated equilibria as well as the smoothness framework of Roughgarden are both equivalent to a problem of general Lyapunov function estimation. For non-coarse equilibria, we instead observe a vector field fit problem for the gradient dynamics of the game. These follow from containment results in normal form games; the usual notion of a (coarse) correlated equilibria is equivalent to our first-order local notions of (coarse) correlated equilibria with respect to an appropriately chosen set of vector fields.

First-order approximation of strong vector equilibria with application to nondifferentiable constrained optimization

Global Approximations of Vector Optimization Problems in Terms of Variational Convergence

Directional Compactness, Approximations and Efficiency Conditions for Nonsmooth Vector Equilibrium Problems with Constraints

Nonlinearly Constrained Best Approximation in Hilbert Spaces: the Strong CHIP and the Basic Constraint Qualification.

Approximate solutions of vector optimization problems via improvement sets in real linear spaces

Approximate Solutions in Linear Fractional Vector Optimization

Approximate Solutions for Abstract Inequality Systems

Approximating the set of Nash equilibria for convex games

First-order (coarse) correlated equilibria in non-concave games

First-order Optimality Conditions for Two Classes of Generalized Nonsmooth Semi-Infinite Optimization

Vector Equilibrium Problems—A Unified Approach and Applications

Strong convergence of an inertial iterative algorithm for variational inequality problem, generalized equilibrium problem, and fixed point problem in a Banach space

Bounding the Error for Approximate Solutions of Almost Linear Complementarity Problems Using Feasible Vectors

Optimality conditions for robust weakly efficient solutions in uncertain optimization

Vector optimization problems with weakened convex and weakened affine constraints in linear topological spaces

First-Order Methods for Linearly Constrained Bilevel Optimization

Efficiency Conditions for Nonsmooth Vector Equilibrium Problems with Constraints via Generalized Subdifferentials

Equilibrium modeling and solution approaches inspired by nonconvex bilevel programming

Second-order optimality conditions for general nonconvex optimization problems and variational analysis of disjunctive systems

Linear convergence of first order methods for non-strongly convex optimization

On some optimality conditions for a class of problems in mathematical programming with equilibrium constraints