Flows and Decompositions of Games: Harmonic and Potential Games

In this paper we introduce a novel flow representation for finite games in strategic form. This representation allows us to develop a canonical direct sum decomposition of an arbitrary game into three components, which we refer to as the potential, harmonic and nonstrategic components. We analyze natural classes of games that are induced by this decomposition, and in particular, focus on games with no harmonic component and games with no potential component. We show that the first class corresponds to the well-known potential games. We refer to the second class of games as harmonic games, and study the structural and equilibrium properties of this new class of games. Intuitively, the potential component of a game captures interactions that can equivalently be represented as a common interest game, while the harmonic part represents the conflicts between the interests of the players. We make this intuition precise, by studying the properties of these two classes, and show that indeed they have quite distinct and remarkable characteristics. For instance, while finite potential games always have pure Nash equilibria, harmonic games generically never do. Moreover, we show that the nonstrategic component does not affect the equilibria of a game, but plays a fundamental role in their efficiency properties, thus decoupling the location of equilibria and their payoff-related properties. Exploiting the properties of the decomposition framework, we obtain explicit expressions for the projections of games onto the subspaces of potential and harmonic games. This enables an extension of the properties of potential and harmonic games to “nearby” games. We exemplify this point by showing that the set of approximate equilibria of an arbitrary game can be characterized through the equilibria of its projection onto the set of potential games.

💡 Research Summary

The paper introduces a novel representation of finite strategic‑form games as flows on a graph, where each pure‑strategy profile is a node and a unilateral deviation by a player corresponds to a directed edge. The weight on an edge is the payoff difference experienced by the deviating player, analogous to a voltage drop in an electrical network. This “flow” viewpoint makes the entire payoff structure a vector in a high‑dimensional space and naturally leads to a decomposition of any game into three orthogonal components: a potential component, a harmonic component, and a non‑strategic component.

Potential component.
If the flow is both divergence‑free (conserved at each node) and curl‑free (no circulation around any closed loop), it belongs to the potential subspace. In this case there exists a single scalar function φ (the potential) such that every unilateral payoff change equals the change in φ. This recovers the classic class of potential games. The authors construct an explicit orthonormal basis for this subspace and give a closed‑form projection matrix that maps any game onto its nearest potential game. Because φ’s minima are pure Nash equilibria, potential games always admit at least one pure equilibrium.

Harmonic component.
When the flow is conserved but exhibits non‑zero curl, it lies in the harmonic subspace. Here the circulation around some cycles is non‑vanishing, reflecting genuine conflicts among players’ interests. Games whose entire flow resides in this subspace are called harmonic games. The paper proves that the harmonic subspace has a dimension equal to (|S|−1)(|S|−2)/2 for two‑player games (and a corresponding formula for the general n‑player case) and shows that a generic harmonic game has no pure Nash equilibrium. Moreover, best‑response dynamics in harmonic games can generate cyclic or chaotic trajectories, underscoring the intrinsic instability of this class.

Non‑strategic component.
The third component consists of flow‑free terms: payoff adjustments that are identical for all players and independent of the strategy profile. These terms do not affect best‑response conditions, so they leave the set of Nash equilibria unchanged. However, they influence welfare‑related measures such as social welfare, price of anarchy, and Pareto efficiency. The authors demonstrate that by adding a suitable non‑strategic term one can improve or degrade the efficiency of equilibria without altering their locations.

Decomposition algorithm.
Using the orthonormal bases for the three subspaces, the authors present a linear‑algebraic algorithm that projects any game matrix onto each component. The algorithm requires only matrix multiplications and a few inverses, making it scalable to large games (especially when the payoff matrices are sparse). The decomposition is unique because the three subspaces are mutually orthogonal under the inner product induced by the flow representation.

Applications to approximate equilibria.
A key insight is that the potential projection of a game, denoted P(G), captures the “nearest” potential game to G. The authors prove that every ε‑Nash equilibrium of G corresponds to an exact Nash equilibrium of P(G) with ε equal to the norm of the harmonic component of G. Consequently, the set of approximate equilibria of any game can be fully characterized by studying the exact equilibria of its potential projection. This bridges the gap between the well‑understood equilibrium structure of potential games and the more complex landscape of general games.

Implications and future directions.
The decomposition cleanly separates the two central concerns of game theory: existence of equilibria (handled by the potential part) and efficiency of outcomes (handled by the non‑strategic part), while the harmonic part quantifies the degree of strategic conflict that obstructs equilibrium existence. This framework suggests new design principles: minimizing the harmonic component of a mechanism should increase the likelihood of pure equilibria, whereas adjusting the non‑strategic component can improve welfare without affecting equilibrium locations. Potential applications include mechanism design for networks, congestion games, and multi‑agent reinforcement learning, where one can monitor the evolution of the three components over time to diagnose instability or inefficiency.

In summary, the paper provides a mathematically rigorous, computationally tractable, and conceptually illuminating decomposition of finite games into potential, harmonic, and non‑strategic parts. By doing so, it extends classic potential‑game results to a broader class of “near‑potential” games, offers explicit tools for analyzing approximate equilibria, and opens a rich avenue for future research on game design, dynamics, and efficiency.