S-Supporting Equilibria in Game Theory

• S-supporting equilibria are a family of game-theoretic notions defined by structural constraints on mixed supports, prescribed allocations, or control sets to ensure strategic stability.
• They are applied in diverse settings including win–lose bimatrix games, market design, repeated additive games, and strategic control, offering both theoretical and algorithmic insights.
• Analytical techniques such as graph-theoretic reformulations, Lipschitz smoothing, and direct control methods reveal trade-offs, lower bound barriers, and support-size bounds in equilibrium approximation.

Searching arXiv for the cited papers and closely related work on S-supporting equilibria across game-theoretic contexts. “S-supporting equilibria” is a context-dependent expression in contemporary game theory. In one line of work on bimatrix games, an $\epsilon$-well-supported Nash equilibrium is called $S$-supporting when each player’s mixed strategy has support of cardinality at most $S$ (Anbalagan et al., 2015). In market design, an equilibrium is “$x^*$-supporting” when anonymous prices and agent-specific budgets support a prescribed Pareto-optimal allocation $x^*$ (Andelman et al., 2021). In repeated additive games with reactive strategies, non-empty subsets $S$ of the action set index equilibrium classes in which self-play uses exactly the actions in $S$ (Lesigang et al., 26 Jun 2026). A further usage studies which subset of players must be fixed at a target equilibrium so that all remaining players best-respond to it (Polevoy et al., 2023). This suggests that the expression is not a single standardized definition but a family of related notions organized around what is being “supported”: a bounded mixed support, a target allocation, a self-play action set, or an equilibrium profile itself.

1. Terminological scope

The main usages can be organized as follows.

Context	Meaning of “supporting”	Equilibrium object
Win-lose bimatrix games	Both mixed supports have cardinality $\le S$	$\epsilon$-WSNE
Competitive equilibrium with unequal budgets	Prices and budgets support a target allocation $x^*$	$S$0
Repeated additive games	Self-play uses exactly the actions in $S$1	Symmetric reactive NE class
Strategic-form control	A player subset makes the target equilibrium a best response for everyone else	Direct control set

In bimatrix approximation, the relevant object is the support of a mixed strategy: if $S$2 is an $S$3-well-supported Nash equilibrium, then “$S$4-supporting” means that both $S$5 and $S$6 have cardinality at most $S$7 (Anbalagan et al., 2015). In Fisher-style markets with divisible goods, the central question is instead whether one can choose anonymous prices and agent-specific budgets so that a given feasible allocation $S$8 is individually optimal and market-clearing (Andelman et al., 2021). In repeated additive games with reactive strategies, $S$9 is a subset of the action set, and an $S$0-supporting equilibrium is one whose stationary self-play distribution is supported on $S$1 (Lesigang et al., 26 Jun 2026). In direct-control models, “supporting” refers neither to mixed-strategy support size nor to action subsets, but to a coalition of players whose forced compliance makes the target equilibrium a best response for the remaining players (Polevoy et al., 2023).

A common source of confusion is therefore terminological rather than mathematical. Results about $S$2-supporting $S$3-WSNE do not transfer directly to $S$4-supporting competitive equilibria or to $S$5-supporting reactive equilibria, because the underlying state spaces, feasibility constraints, and equilibrium predicates are different.

2. $S$6-supporting $S$7-well-supported Nash equilibria in win-lose games

For a bimatrix game $S$8 with $S$9, a pair of mixed strategies $x^*$0, $x^*$1 is an $x^*$2-well-supported Nash equilibrium if every pure strategy played with positive probability is within $x^*$3 of a best response. Equivalently,

$x^*$4

for all $x^*$5 with $x^*$6 and all $x^*$7, and

$x^*$8

for all $x^*$9 with $x^*$0 and all $x^*$1. Such an $x^*$2-WSNE is $x^*$3-supporting if both supports have cardinality at most $x^*$4 (Anbalagan et al., 2015).

For win-lose games, every instance can be encoded as a directed bipartite graph $x^*$5 with row vertices $x^*$6 and column vertices $x^*$7, where

$x^*$8

Assuming every vertex has out-degree at least $x^*$9, the central characterization states that for any integer $S$0 and any $S$1 with $S$2, the game admits an $S$3-WSNE with both supports of cardinality at most $S$4 if and only if the graph contains either a directed cycle of length at most $S$5, or an undominated set of $S$6 vertices all on one side of the bipartition (Anbalagan et al., 2015).

The characterization isolates two distinct mechanisms for small-support well-supported equilibria. A short directed cycle supplies a compact alternating structure of mutually rewarding actions. An undominated set supplies a side of the game for which no single opponent action covers all supported actions simultaneously. The proof decomposes accordingly: one lemma constructs an equilibrium from an undominated set, another from a short directed cycle, and the converse shows that any $S$7-WSNE with small supports forces one of those two combinatorial configurations. In the converse direction, if neither support set is undominated, then both players have true best responses of payoff $S$8 against the opponent’s mix; in the induced subgraph on the supports, every vertex has out-degree at least $S$9, hence there is a directed cycle of length at most $S$0 (Anbalagan et al., 2015).

This graph-theoretic reformulation is significant because it converts a mixed-strategy existence question into a finite structural question about bipartite digraphs. In the win-lose setting, support-bounded $S$1-WSNE are therefore governed by cycle structure and domination structure rather than by generic fixed-point arguments.

3. Lower bounds and barriers for small supports

The main lower bound for well-supported equilibria is negative in the strongest possible constant-support sense: for every fixed $S$2 and any $S$3, there are win-lose bimatrix games for which every $S$4-WSNE has support size greater than $S$5 (Anbalagan et al., 2015). The proof uses additive number theory. Haight’s theorem yields, for any $S$6, a modulus $S$7 and a subset $S$8 with $S$9 and $\le S$0. From this, one constructs a non-bipartite digraph of girth at least $\le S$1 in which every pair of vertices is dominated by some third vertex, and then converts it into a bipartite digraph by doubling vertices and adding self-arcs. The resulting win-lose game has no cycle of length at most $\le S$2 and every set of $\le S$3 rows or columns is dominated, so the characterization theorem excludes every $\le S$4-WSNE with supports of size at most $\le S$5 (Anbalagan et al., 2015).

This construction has two further consequences. First, it shows that the minimum support required for $\le S$6-WSNE is unbounded as $\le S$7 grows, even though the argument is existential rather than quantitative in terms of elementary closed forms. Second, it refutes graph-theoretic conjectures of Daskalakis, Mehta and Papadimitriou, and of Myers, by producing $\le S$8-digraphs for all $\le S$9 rather than forcing the conjectured dichotomy between short cycles and small undominated sets (Anbalagan et al., 2015).

A complementary barrier was established earlier for the regime $\epsilon$0. In win-lose games on $\epsilon$1 strategies, every $\epsilon$2-WSNE may require each player’s support to have size at least $\epsilon$3 (Anbalagan et al., 2013). The construction uses a bipartite digraph derived from a random tournament and a “$\epsilon$4-covered” property ensuring that every $\epsilon$5-set on either side has a common in-neighbour. The contradiction argument combines this covering property with a bipartite Caccetta-Häggkvist-type theorem: if every vertex in a balanced bipartite digraph on $\epsilon$6 vertices has in-degree greater than $\epsilon$7, then the graph contains a directed $\epsilon$8-cycle. Since the constructed game has no directed $\epsilon$9- or $x^*$0-cycles, any $x^*$1-WSNE with $x^*$2 must have larger supports (Anbalagan et al., 2013).

The same paper also supplies the positive side of the support-size trade-off: in every bimatrix game with payoffs in $x^*$3, for any $x^*$4 there is an $x^*$5-WSNE in which each player mixes uniformly on a multiset of

$x^*$6

pure strategies (Anbalagan et al., 2013). Thus the lower bound is polylogarithmic and the upper bound is polylogarithmic, though with different exponents and under different parameter regimes. The paper also proves a sharp “no size-$x^*$7” phenomenon: for every $x^*$8, there exist win-lose games for which no pair of mixed strategies with supports of size at most two forms a $x^*$9-WSNE, even though every bimatrix game with payoffs in $S$00 admits a $S$01-approximate Nash equilibrium with supports of cardinality at most two (Anbalagan et al., 2013). This is one of the clearest separations between approximate Nash equilibrium and its well-supported variant.

4. Small-support equilibria beyond well-supported Nash

The small-support perspective extends well beyond $S$02-WSNE. For $S$03-player games with $S$04 actions per player, every game admits an $S$05-approximate Nash equilibrium in which each player mixes over at most

$S$06

pure actions (Babichenko, 2013). Babichenko’s proof uses population-splitting and Lipschitz smoothing: each original player is replaced by a population of $S$07 subplayers, the smoothed game becomes $S$08-Lipschitz in every opponent’s action, and concentration for Lipschitz functions then implies the existence of a pure equilibrium in the population game, which corresponds to a $S$09-uniform $S$10-equilibrium in the original game. For graphical games of maximum degree $S$11, the same argument yields

$S$12

so the dependence shifts from the total number of players to the interaction degree (Babichenko, 2013).

Approximate correlated equilibrium and approximate coarse correlated equilibrium admit analogous polylogarithmic support bounds. In every $S$13-player $S$14-action game and for every $S$15, there exists an $S$16-CCE with support size at most

$S$17

and there exists an $S$18-CE with support size at most

$S$19

for any chosen failure probability $S$20 (Babichenko et al., 2013). The probabilistic method behind both results is similar: start from an exact equilibrium, sample profiles independently, and apply Hoeffding’s inequality plus a union bound over regret constraints. For $S$21-CCE, this leads to a randomized algorithm with expected number of iterations at most $S$22 once an exact correlated equilibrium is available; each iteration checks the $S$23 deviation inequalities in $S$24 time (Babichenko et al., 2013).

These results show that “small support” is a robust theme across equilibrium notions, but not a uniform one. Approximate Nash, $S$25-WSNE, $S$26-CE, and $S$27-CCE impose different deviation classes, and the resulting support-size bounds differ accordingly. The support-minimization problem also changes with the equilibrium concept: finding an exact correlated equilibrium of minimum support is NP-hard under Cook reductions even in two-player zero-sum games (Babichenko et al., 2013).

5. Supporting arbitrary Pareto-optimal allocations in competitive equilibrium

In a market with additive valuations over heterogeneous divisible resources, an allocation $S$28 is “supported” when there exist anonymous prices $S$29 and agent-specific budgets $S$30 such that $S$31 is a competitive equilibrium (Andelman et al., 2021). The model has agents $S$32, goods $S$33 with supplies $S$34, and additive utilities

$S$35

The tuple $S$36 is $S$37-supporting if: each agent solves

$S$38

no agent overspends, and market clearing holds: $S$39

The core theorem states that given any Pareto-optimal allocation $S$40, one can compute in polynomial time prices $S$41 and budgets $S$42 so that $S$43 is an equilibrium supporting $S$44 (Andelman et al., 2021). The construction has three phases.

First, a cycle-elimination procedure transforms the sharing graph of the allocation into a forest while preserving Pareto optimality and each agent’s utility. If the bipartite sharing graph $S$45 contains a cycle, an infinitesimal reallocation around that cycle can eliminate one strictly interior edge without decreasing any agent’s utility. Repeating this produces a new Pareto-optimal allocation $S$46 with the same utility vector and a cycle-free sharing graph (Andelman et al., 2021).

Second, each tree component is priced locally. Choosing an arbitrary agent $S$47 as root, prices on adjacent items are initialized by $S$48, and then propagated through the tree by

$S$49

whenever agent $S$50 holds both the already priced item $S$51 and another item $S$52. This enforces indifference among all goods held by the same agent within the tree: $S$53 Each agent budget is then set to exactly finance the assigned bundle: $S$54 Lemmas 3.3–3.4 show that under these local prices and budgets, no agent in the tree can profitably reallocate budget from one good in the tree to another (Andelman et al., 2021).

Third, the locally priced trees are scaled to prevent cross-tree deviations. One chooses a scalar $S$55 per tree and sets

$S$56

A continuous gain map $S$57 is defined on the simplex of scaling vectors, and Brouwer’s fixed-point theorem yields $S$58 with $S$59. At such a fixed point, all cross-tree gains vanish. Equivalently, the paper gives a linear-program formulation: $S$60 together with $S$61 and $S$62 (Andelman et al., 2021).

Algorithmically, cycle elimination can be done in $S$63, pricing each tree is a single BFS or DFS in $S$64, and the scaling step can be solved in polynomial time by LP. The same framework supports the Rawlsian max-min allocation, which can itself be found by a single LP in $S$65, or by sorting in $S$66 for two agents and in $S$67 for two goods (Andelman et al., 2021). In this literature, “supporting equilibrium” means implementability of a target Pareto-optimal allocation rather than sparsity of a mixed strategy.

6. $S$68-supporting classes in repeated additive games

In repeated additive games between two players with finitely many actions, Lesigang, Hilbe and Glynatsi study reactive strategies, i.e. memory-1 strategies that condition only on the opponent’s previous action (Lesigang et al., 26 Jun 2026). The stage game is additive: $S$69 and the long-run payoff under reactive strategies exists as

$S$70

A reactive strategy satisfies $S$71 for all $S$72, so it can be written as $S$73 with each $S$74 (Lesigang et al., 26 Jun 2026).

A symmetric Nash equilibrium in the class of reactive strategies is a $S$75 such that

$S$76

A key simplification is that one only needs to check deviations among pure unconditional strategies $S$77: $S$78 Moreover, if $S$79 is the stationary frequency of $S$80 under self-play and

$S$81

then

$S$82

Because

$S$83

the equilibrium conditions become linear equalities and inequalities in the strategy variables (Lesigang et al., 26 Jun 2026).

Theorem 3.1 states that a reactive strategy $S$84 is a symmetric Nash equilibrium if and only if there exists a non-empty $S$85 such that four conditions hold. First, a support condition: $S$86 Second, if the Markov chain $S$87 is not primitive, the initial move must be chosen so that the limiting distribution is supported on $S$88. Third, all actions in $S$89 give equal payoff against $S$90: $S$91 Fourth, no action outside $S$92 gives more: $S$93 Distinct subsets $S$94 give disjoint equilibrium classes, and every reactive equilibrium belongs to exactly one minimal $S$95 (Lesigang et al., 26 Jun 2026).

The case $S$96 recovers equalizer strategies. Then the zero restrictions disappear, the outside-action inequalities are vacuous, and only the equal-payoff conditions remain: $S$97 The paper identifies these with the equalizer or zero-determinant strategies of Press and Dyson and Hilbe et al. (Lesigang et al., 26 Jun 2026).

For a three-action donation game with actions $S$98 and parameters $S$99, $S$00, applying the characterization yields seven equilibrium classes. Evolutionary simulations under Imhof and Nowak dynamics show that when $S$01 increases from $S$02 to $S$03 while $S$04, $S$05, and $S$06, $S$07-supporting equilibria dominate at low $S$08, $S$09-supporting equilibria dominate at high $S$10, and the equalizer class $S$11 is rarely observed because it is far less invasion-resistant (Lesigang et al., 26 Jun 2026). The maximal number of free parameters in an $S$12-action game is

$S$13

which is symmetric in $S$14 about $S$15 (Lesigang et al., 26 Jun 2026).

7. Control sets and equilibrium attainment

A structurally different notion asks which subset of players can be forced to play a target equilibrium so that the remaining players then find that equilibrium a best response. In a finite strategic-form game $S$16 with target equilibrium $S$17 and initial profile $S$18, a subset $S$19 is a direct control set if

$S$20

The minimum size of such a set is denoted $S$21 (Polevoy et al., 2023).

This notion links support to equilibrium strength. If every coalition of size at least $S$22 is a direct control set for $S$23, and if each player weakly prefers $S$24 when the others play $S$25, namely

$S$26

then $S$27 is an $S$28-strong Nash equilibrium, and the bound is tight (Polevoy et al., 2023). Thus control sets provide a bridge between intervention requirements and robustness against coalitional deviation.

The computational picture is difficult. Deciding whether there exists a direct control set of size at most $S$29 is NP-complete, and unless $S$30 no polynomial-time algorithm approximates $S$31 within factor $S$32 for any $S$33, even when each player has only two actions (Polevoy et al., 2023). Under player-wise monotonicity, however, a Local-Ratio algorithm gives an $S$34-approximation when each influence neighborhood has size at most $S$35. The paper also identifies tractable or approximately tractable cases inside potential games: singleton congestion games under a general-position assumption and mild cost-monotonicity admit an exact polynomial-time algorithm; symmetric congestion games with decreasing costs can be searched over $S$36 in polynomial time; and in coordination games on graphs, the minimum direct control set problem is NP-complete in the two-colour case, but a target-dominating-set reduction yields an $S$37-approximation when all $S$38 are large enough (Polevoy et al., 2023).

In this literature, the “supporting” subset is a subset of players rather than actions or mixed-strategy supports. The conceptual commonality is nevertheless clear: an equilibrium is supported by a small or structured object that certifies its attainability, implementability, or self-enforcement.