Game Theory 2: Nash, Correlated, Wardrop, and Existence Theorems

A single-file HTML note on solution concepts and equilibrium existence, and expanded using the study questions that came up in discussion: Nash equilibrium, mixed strategies, correlated equilibrium, Kakutani, Wardrop equilibrium, and existence results for finite and infinite games.

Normal-form games Mixed strategies Correlated equilibrium Fixed-point proofs Wardrop Infinite games

Table of contents

1. Nash equilibrium and mixed Nash equilibrium core definitions

A strategic-form game is

\[ G = \bigl(I, (S_i)_{i \in I}, (u_i)_{i \in I}\bigr), \]

where \(I\) is the set of players, \(S_i\) is player \(i\)'s strategy set, and \(u_i : \prod_{j\in I} S_j \to \mathbb{R}\) is player \(i\)'s payoff.

Pure-strategy Nash equilibrium

Definition
\[ s^* = (s_i^*)_{i\in I} \text{ is a Nash equilibrium if } u_i(s_i^*, s_{-i}^*) \ge u_i(t_i, s_{-i}^*) \quad \forall i,\ \forall t_i \in S_i. \]
Interpretation. At a Nash equilibrium, every player's chosen strategy is a best response to the other players' chosen strategies. No one can gain by deviating alone.

Mixed strategy

A mixed strategy of player \(i\) is a probability distribution over \(S_i\):

\[ \sigma_i \in \Delta(S_i). \]

If players randomize independently, the mixed-strategy profile is \(\sigma=(\sigma_i)_{i\in I}\in \prod_i \Delta(S_i)\), and expected payoff is

\[ u_i(\sigma) = \sum_{s \in S} \left(\prod_{j\in I}\sigma_j(s_j)\right) u_i(s). \]

Mixed-strategy Nash equilibrium

Definition
\[ \sigma^* \text{ is a mixed Nash equilibrium if } u_i(\sigma_i^*, \sigma_{-i}^*) \ge u_i(\tau_i,\sigma_{-i}^*) \quad \forall i,\ \forall \tau_i \in \Delta(S_i). \]
Pure Nash equilibrium says "best response to a fixed action profile." Mixed Nash equilibrium says "best response to opponents' probability distributions."

Example: Matching Pennies

Player 1 \ Player 2 H T
H (1,-1) (-1,1)
T (-1,1) (1,-1)

No pure Nash equilibrium. In every cell, one player wants to switch.

  • If the outcome is \((H,H)\), player 2 wants to switch to \(T\).
  • If the outcome is \((H,T)\), player 1 wants to switch to \(T\).
  • If the outcome is \((T,H)\), player 1 wants to switch to \(H\).
  • If the outcome is \((T,T)\), player 2 wants to switch to \(H\).

Deriving the mixed equilibrium

Let \(q\) be the probability that player 2 plays \(H\). Player 1 must be indifferent between \(H\) and \(T\):

\[ u_1(H,q) = q(1) + (1-q)(-1) = 2q-1, \] \[ u_1(T,q) = q(-1) + (1-q)(1) = 1-2q. \]

Set them equal:

\[ 2q-1 = 1-2q \quad \Longrightarrow \quad q=\tfrac12. \]

By symmetry, player 2 is indifferent only if player 1 also chooses \(H\) with probability \(p=\tfrac12\).

Conclusion. Matching Pennies has no pure equilibrium, but it has a mixed Nash equilibrium: \[ \sigma_1^*(H)=\sigma_2^*(H)=\tfrac12. \]
pure H pure T mixed 50-50 the whole line segment is the mixed-strategy space
For a player with two pure actions, the mixed-strategy set \(\Delta(\{H,T\})\) is a line segment. Endpoints are pure strategies; interior points are genuine randomizations.

2. Correlated equilibrium joint randomization + recommendations

In a mixed Nash equilibrium, players randomize independently. A correlated equilibrium allows a joint distribution over action profiles, so recommendations can be statistically related.

Idea

A mediator draws a random pure profile \(R=(R_1,\dots,R_n)\) from a joint distribution \(\pi \in \Delta(S)\). Player \(i\) sees only the recommendation \(R_i\).

After seeing \(R_i=s_i\), player \(i\) forms a posterior belief about the others' recommendations and decides whether following \(s_i\) is optimal.

Mediator draws R ~ pi over full profiles Player 1 sees only R1 Player 2 sees only R2 recommend s1 recommend s2
The mediator samples a full profile, but each player sees only his or her component.

Formal definition

Definition
\[ \pi \in \Delta(S) \text{ is a correlated equilibrium if, for every player } i, \] \[ \sum_{s_{-i}\in S_{-i}} \Pr(R=s \mid R_i=s_i)\, \bigl[u_i(s_i,s_{-i}) - u_i(t_i,s_{-i})\bigr] \ge 0 \] \[ \text{for all recommendations } s_i \text{ with } \Pr(R_i=s_i)>0 \text{ and all deviations } t_i \in S_i. \]
1
Fix a recommendation \(s_i\). Player \(i\) is told: "play \(s_i\)."
2
Use conditional probability. The term \(\Pr(R=s \mid R_i=s_i)\) is the posterior belief about the full profile after seeing the recommendation.
3
Compare follow vs deviate. The difference \[ u_i(s_i,s_{-i}) - u_i(t_i,s_{-i}) \] is the gain from following the recommendation instead of switching to \(t_i\).
4
Average using the posterior. The whole sum is an expected payoff difference. The inequality says the expected gain from following is nonnegative.
How to read the formula in one line. Given the recommendation I actually receive, I do not want to switch to any other action.

Why writing \(\Pr(R=s \mid R_i=s_i)\) is not strange

Write the full profile as \(s=(s_i,s_{-i})\). Once \(s_i\) is fixed, the event \(R=s\) is exactly the same as the event \(R_{-i}=s_{-i}\). So

\[ \Pr(R=s \mid R_i=s_i) = \Pr(R_{-i}=s_{-i} \mid R_i=s_i). \]

The lecture writes \(R=s\) because \(\pi\) is defined on the full profile space \(S\), but conceptually the conditioning is about beliefs over \(s_{-i}\).

Equivalent linear characterization

Using conditional probability,

\[ \Pr(R=s \mid R_i=s_i) = \frac{\pi(s_i,s_{-i})}{\sum_{t_{-i}\in S_{-i}} \pi(s_i,t_{-i})}. \]

If \(\Pr(R_i=s_i)>0\), multiplying both sides by the denominator gives the equivalent condition

\[ \sum_{s_{-i}\in S_{-i}} \pi(s_i,s_{-i}) \bigl[u_i(s_i,s_{-i}) - u_i(t_i,s_{-i})\bigr] \ge 0. \]

This is useful because it is linear in \(\pi\), which is why the set of correlated equilibria is convex.

Example: Traffic intersection game

Player 1 \ Player 2 L R
U (5,1) (0,0)
D (4,4) (1,5)

Pure and mixed Nash equilibria

There are two pure Nash equilibria: \((U,L)\) and \((D,R)\).

To find the mixed equilibrium, let player 1 choose \(U\) with probability \(p\), and player 2 choose \(L\) with probability \(q\).

\[ u_1(U,q)=5q,\qquad u_1(D,q)=4q + 1(1-q)=1+3q. \] \[ 5q = 1+3q \quad\Longrightarrow\quad q=\tfrac12. \] \[ u_2(L,p)=1p + 4(1-p)=4-3p, \] \[ u_2(R,p)=0p + 5(1-p)=5-5p. \] \[ 4-3p = 5-5p \quad\Longrightarrow\quad p=\tfrac12. \]

So the mixed equilibrium gives each player expected payoff \(5/2\).

A correlated device does better

Suppose a public coin recommends:

  • \((U,L)\) with probability \(1/2\)
  • \((D,R)\) with probability \(1/2\)

Expected payoff becomes

\[ \tfrac12(5,1) + \tfrac12(1,5) = (3,3). \]

Why follow the recommendation?

  • If player 1 is told \(U\), he infers player 2 will play \(L\), and \(U\) is optimal.
  • If player 1 is told \(D\), he infers player 2 will play \(R\), and \(D\) is optimal.
  • The same logic holds for player 2.
Conclusion. This is a correlated equilibrium and it yields \((3,3)\), which is strictly better than the mixed Nash payoff \((5/2,5/2)\).
Stronger mediator from the lecture. A mediator can recommend \((U,L)\), \((D,L)\), and \((D,R)\) each with probability \(1/3\). The induced expected payoff is \[ \tfrac13(5,1) + \tfrac13(4,4) + \tfrac13(1,5) = (10/3,10/3), \] and the recommendations are still self-enforcing. This shows correlated equilibrium can achieve payoffs that are outside the convex hull of Nash-equilibrium payoffs and strictly better than randomizing only between Nash equilibria.
Fact Why it is true
Every mixed Nash equilibrium is a correlated equilibrium Independent randomization is a special case of a joint distribution over profiles.
The set of correlated equilibria is convex The defining inequalities are linear in \(\pi\).
The CE set contains the convex hull of mixed Nash equilibria Because mixed Nash equilibria are inside the CE set and the CE set is convex.

3. Existence of a mixed strategy equilibrium in finite games the theorem and the roadmap

Nash's theorem
\[ \text{Every finite strategic-form game has a mixed-strategy Nash equilibrium.} \]
Why this matters. In a finite game, pure equilibrium can fail completely, as Matching Pennies shows. Nash's theorem says equilibrium still exists once players are allowed to randomize.

The key idea

Do not search directly for \(\sigma^*\). Instead, define the best-response correspondence \(B\) and look for a fixed point:

\[ \sigma^* \in B(\sigma^*). \]

If that happens, each player's mixed strategy is a best response to the others, which is exactly the definition of a mixed Nash equilibrium.

Finite game finite pure strategy sets Mixed strategy space Sigma = product of simplices compact, convex, nonempty Best responses B(sigma) nonempty, convex, closed graph Kakutani fixed point sigma* mixed NE
This is the logical chain behind Nash's existence theorem in finite games. Section 5 below gives the exact definitions of compactness, convexity, simplex, correspondence, and closed graph that make each arrow valid.

The two technical tools that make the proof work

Weierstrass's theorem. A continuous function on a nonempty compact set attains its maximum and minimum. This guarantees best responses exist.
Kakutani's fixed-point theorem. A set-valued map with nonempty, convex values and closed graph on a compact convex set has a fixed point. This turns "best responses exist" into "an equilibrium exists."

The full proof appears in Section 5 below, after Section 4 on Wardrop equilibrium.

4. Wardrop equilibrium equilibrium for a continuum of users

Wardrop equilibrium is the network-flow analogue of Nash equilibrium when there is a continuum of infinitesimal users, such as many drivers in traffic.

All routes that are actually used have equal and minimal cost. No infinitesimal user can improve by switching routes.

Basic route-based definition

Let \(f_p\) be the flow on path \(p\), and let \(c_p(f)\) be the cost of that path under total flow vector \(f\). A feasible flow \(f\) is a Wardrop equilibrium if

\[ f_p > 0 \quad \Longrightarrow \quad c_p(f) \le c_q(f)\ \text{for every feasible path } q. \]

So any route that carries positive flow must be a cheapest route.

O D Route 1: c1(f1) = f1 Route 2: c2(f2) = 1/2 Total demand = 1 Wardrop equilibrium: f1 = 1/2, f2 = 1/2
In this example, if both routes are used, they must have equal cost. Since \(c_2 = 1/2\), we need \(c_1(f_1)=f_1=1/2\), hence \(f_1=f_2=1/2\).

Priced-link version used in congestion games

In the pricing-congestion model from the lecture, provider \(i\) chooses a price \(p_i\), link \(i\) has latency \(l_i(x_i)\), and the effective cost of using link \(i\) is

\[ p_i + l_i(x_i). \]

With total demand \(d\) and reservation utility \(R\), a flow vector \(x=(x_i)_{i=1}^I\) is a Wardrop equilibrium if

\[ \sum_{i=1}^I x_i \le d, \] \[ p_i + l_i(x_i) = \min_j \{p_j + l_j(x_j)\} \quad \text{for all } i \text{ with } x_i > 0, \] \[ p_i + l_i(x_i) \le R \quad \text{for all } i \text{ with } x_i > 0, \]

and if the minimal effective cost is strictly below \(R\), then all traffic is served:

\[ \sum_{i=1}^I x_i = d. \]
Interpretation. Used links must have minimal effective cost, and users only enter the network if that cost does not exceed the reservation utility.
Why this appears in the existence lecture. The lecture uses Wardrop flow as the lower-level behavior in a price-congestion game. Providers choose prices, users react through Wardrop equilibrium, and the resulting provider game can have either a unique pure Nash equilibrium or no pure Nash equilibrium depending on the latency functions.

Connection to Nash equilibrium

Wardrop equilibrium is like Nash equilibrium for infinitely many negligible users. A single user cannot noticeably change congestion, so the relevant deviation is an infinitesimal switch of mass from one route to another.

5. Existence of a mixed strategy Nash equilibrium in finite games proof details

This is the detailed fixed-point proof behind Nash's theorem. The main idea is not just a checklist of lemmas: it is a reusable existence pattern. Build a strategy space with the right topological shape, turn equilibrium into a fixed-point problem, and then use Kakutani to force a self-consistent profile.

Abstract template. Find a nonempty compact convex set \(X\) and a best-response correspondence \(B:X \rightrightarrows X\) with nonempty convex values and closed graph. Then Kakutani gives some \(x^*\in X\) such that \(x^*\in B(x^*)\).

Core definitions used in the proof

Definitions

Convex set. A set \(C\subseteq \mathbb{R}^m\) is convex if for all \(x,y\in C\) and every \(\lambda\in[0,1]\),

\[ \lambda x + (1-\lambda)y \in C. \]

Standard simplex. The standard \(n\)-simplex is

\[ \Delta_n = \left\{p\in \mathbb{R}^{n+1} : p_k \ge 0 \text{ for all } k,\ \sum_{k=0}^n p_k = 1\right\}. \]

For a player with \(|S_i|=m_i\) pure actions, the mixed-strategy set is exactly a standard simplex:

\[ \Sigma_i = \Delta(S_i) \cong \left\{p_i\in \mathbb{R}^{m_i} : p_i(a_i)\ge 0,\ \sum_{a_i\in S_i} p_i(a_i)=1\right\}. \]

Compactness. In finite-dimensional Euclidean space, a set is compact if and only if it is closed and bounded. So each simplex \(\Sigma_i\) is compact because it is a closed and bounded subset of \(\mathbb{R}^{m_i}\).

Homeomorphism. A map \(h:A\to B\) between topological spaces is a homeomorphism if it is bijective, continuous, and \(h^{-1}:B\to A\) is also continuous. In that case, \(A\) and \(B\) have the same topological shape.

Correspondence. A correspondence \(B:X \rightrightarrows X\) assigns to each \(x\in X\) a set \(B(x)\subseteq X\), not a single point. This is necessary because best responses need not be unique.

Closed graph. The graph of \(B\) is

\[ \operatorname{Graph}(B)=\{(x,y)\in X\times X : y\in B(x)\}. \]

We say \(B\) has a closed graph if whenever \(x^n\to x\), \(y^n\to y\), and \(y^n\in B(x^n)\) for all \(n\), then \(y\in B(x)\).

Why homeomorphism matters for fixed points. If \(h:A\to B\) is a homeomorphism and \(f:A\to A\) is continuous, then

\[ g = h\circ f\circ h^{-1}:B\to B \]

is continuous, and \(g(y)=y\) if and only if \(f(h^{-1}(y))=h^{-1}(y)\). So fixed points are transported by homeomorphisms. This is why Brouwer is often proved on a standard simplex or ball and then transferred to any homeomorphic model. In this note, however, we apply Kakutani directly on \(\Sigma\), so no explicit homeomorphism is needed for the Nash proof itself.

Brouwer. If \(f:X\to X\) is continuous and \(X\) is nonempty, compact, and convex, then some \(x^*\) satisfies \(f(x^*)=x^*\).
Kakutani. This is the set-valued version: if \(B:X\rightrightarrows X\) has nonempty convex values and closed graph, then some \(x^*\) satisfies \(x^*\in B(x^*)\).

Step 1: Build the compact convex stage

For each player \(i\), define the simplex of mixed strategies

\[ \Sigma_i = \Delta(S_i). \]

The full mixed strategy space is

\[ \Sigma = \prod_{i\in I} \Sigma_i. \]

By the definitions above, each \(\Sigma_i\) is a nonempty compact convex set. Finite products preserve these properties, so \(\Sigma\) is exactly the kind of space on which Kakutani can operate.

Step 2: Turn equilibrium into a fixed-point problem

Best response correspondence
\[ B_i(\sigma_{-i}) = \arg\max_{\tau_i \in \Sigma_i} u_i(\tau_i,\sigma_{-i}), \] \[ B(\sigma) = \prod_{i\in I} B_i(\sigma_{-i}). \]

A fixed point \(\sigma^* \in B(\sigma^*)\) means every player's mixed strategy is a best response to the others. So instead of directly solving many equilibrium inequalities at once, the proof reduces existence to a single fixed-point statement.

A
\(B(\sigma)\) is nonempty. For fixed \(\sigma_{-i}\), the function \(\tau_i \mapsto u_i(\tau_i,\sigma_{-i})\) is linear in \(\tau_i\), hence continuous. Since \(\Sigma_i\) is compact and nonempty, Weierstrass gives at least one maximizer. This is the first place where topology enters: compactness turns continuity into actual existence of best responses.
B
\(B(\sigma)\) is convex-valued. Suppose \(\sigma_i'\) and \(\sigma_i''\) are both best responses to \(\sigma_{-i}\). Then for every \(\tau_i\in\Sigma_i\), \[ u_i(\sigma_i',\sigma_{-i}) \ge u_i(\tau_i,\sigma_{-i}), \qquad u_i(\sigma_i'',\sigma_{-i}) \ge u_i(\tau_i,\sigma_{-i}). \] Multiply by \(\lambda\) and \(1-\lambda\), add, and use linearity: \[ u_i\bigl(\lambda \sigma_i' + (1-\lambda)\sigma_i'',\, \sigma_{-i}\bigr) = \lambda u_i(\sigma_i',\sigma_{-i}) + (1-\lambda)u_i(\sigma_i'',\sigma_{-i}) \ge u_i(\tau_i,\sigma_{-i}). \] So every convex combination of best responses is again a best response. Geometrically, linearity of expected payoff makes the argmax set inherit the convexity of the simplex.
C
\(B\) has a closed graph. Intuitively, best responses cannot jump wildly when payoffs are continuous. If \((\sigma^n,\hat{\sigma}^n)\to(\sigma,\hat{\sigma})\) and each \(\hat{\sigma}^n \in B(\sigma^n)\), then for every player \(i\) and every candidate deviation \(\tau_i\in\Sigma_i\), \[ u_i(\hat{\sigma}_i^n,\sigma_{-i}^n) \ge u_i(\tau_i,\sigma_{-i}^n). \] Passing to the limit and using continuity gives \[ u_i(\hat{\sigma}_i,\sigma_{-i}) \ge u_i(\tau_i,\sigma_{-i}), \] so \(\hat{\sigma}_i\in B_i(\sigma_{-i})\) for every \(i\). Hence \(\hat{\sigma}\in B(\sigma)\).
D
Apply Kakutani. The map \(B:\Sigma \rightrightarrows \Sigma\) has nonempty convex values, closed graph, and is defined on a compact convex nonempty set. Therefore there exists \(\sigma^*\in\Sigma\) such that \[ \sigma^* \in B(\sigma^*). \]
What the proof really shows. The existence theorem is a geometry statement. The mixed-strategy simplex supplies a compact convex stage, continuity keeps optimality stable under limits, and linearity makes best-response sets convex. Once those ingredients are present, Kakutani forces equilibrium.
Conclusion. By Kakutani's fixed-point theorem, a mixed-strategy Nash equilibrium exists in every finite game.

Where Kakutani enters

Kakutani is needed because \(B\) is usually set-valued: a player can have several best responses at the same profile. Brouwer handles single-valued maps, while Kakutani is the fixed-point theorem designed for correspondences.

Weierstrass gives: best responses exist.
Kakutani gives: some profile is a best response to itself.

6. Existence in games with infinite strategy spaces pure existence under concavity; mixed existence more generally

When strategy sets are infinite, pure equilibrium may or may not exist. The answer depends on the shape of the payoff functions.

Debreu-Glicksberg-Fan theorem (pure strategy existence)

Theorem
\[ \text{Consider } \bigl(I,(S_i)_{i\in I},(u_i)_{i\in I}\bigr) \text{ such that for each } i: \]
  • \(S_i\) is compact and convex,
  • \(u_i(s_i,s_{-i})\) is continuous in the strategy profile,
  • \(u_i(s_i,s_{-i})\) is concave (or quasi-concave) in \(s_i\).
\[ \text{Then a pure-strategy Nash equilibrium exists.} \]
Why the proof works. It is the same Kakutani strategy as in finite games, but the source of convexity changes. In finite games, convexity is created by randomization over finitely many pure actions. Here, convexity must already be built into the pure strategy sets \(S_i\), and concavity of payoffs keeps the best-response sets convex.
Finite mixed case. The domain is \(\Sigma=\prod_i \Delta(S_i)\). The simplex geometry comes from allowing mixtures.
Infinite pure case. The domain is \(S=\prod_i S_i\). Convexity must be assumed directly, and concavity replaces linearity in the convexity step.

Proof skeleton

  1. Build the stage. Let \(S=\prod_i S_i\). Since each \(S_i\) is compact, convex, and nonempty, the full strategy space has the same topological shape needed by Kakutani.
  2. Define best responses. Set \(B_i(s_{-i}) = \arg\max_{s_i \in S_i} u_i(s_i,s_{-i})\) and \(B(s)=\prod_i B_i(s_{-i})\).
  3. Get nonemptiness from compactness. Each \(B_i(s_{-i})\) is nonempty by Weierstrass because \(S_i\) is compact and \(u_i\) is continuous.
  4. Get convexity from concavity. The argmax set of a concave function over a convex set is convex, so each \(B_i(s_{-i})\) is convex-valued.
  5. Get closed graph from continuity. The same limit argument as in the finite mixed case shows that \(B\) has a closed graph.
  6. Apply Kakutani. Therefore some \(s^*\in S\) satisfies \(s^*\in B(s^*)\), which is a pure Nash equilibrium.

Example where the theorem applies

Let each player choose a number \(x_i \in [0,1]\), and suppose player \(i\)'s payoff is

\[ u_i(x_i,x_{-i}) = -\bigl(x_i - m_{-i}\bigr)^2, \]

where \(m_{-i}\) is some average of the other players' actions. This payoff is continuous and concave in \(x_i\) because it is a negative quadratic. So the theorem guarantees a pure equilibrium exists.

Concave in own action chord lies below graph own action
Concavity is what makes the best-response set convex, which is exactly what Kakutani needs.

Why pure equilibrium can fail without concavity

A circle-location example: two players choose points on a circle and player 1 wants to maximize distance from player 2, while player 2 wants to minimize it. There is no pure Nash equilibrium because one player always wants to move away and the other wants to adjust in response.

Key lesson. Compactness and continuity alone are not enough for pure equilibrium in infinite games. Concavity or quasi-concavity is the crucial extra condition.

Glicksberg's theorem (mixed strategy existence in compact metric games)

More general mixed-strategy theorem
\[ \text{If each } S_i \text{ is a nonempty compact metric space and each } u_i \text{ is continuous, then a mixed-strategy Nash equilibrium exists.} \]

This theorem is more powerful because it does not require concavity in the pure action. The price is that equilibrium may exist only after players mix, and the mixed-strategy spaces become infinite-dimensional.

7. Deep-dive clarifications answers to the exact sticky points

These notes are included because they were precisely the places where the definitions started to feel abstract.

Study Q1 How should I interpret the correlated-equilibrium inequality?

Fix a player \(i\), a recommendation \(s_i\), and a possible deviation \(t_i\).

\[ \sum_{s_{-i}} \Pr(R=s \mid R_i=s_i) \bigl[u_i(s_i,s_{-i}) - u_i(t_i,s_{-i})\bigr] \ge 0 \]

This is just

\[ \mathbb{E}\!\left[ u_i(\text{follow}) - u_i(\text{deviate}) \,\middle|\, R_i=s_i \right] \ge 0. \]

So the sum is not an unweighted sum over all \(s_{-i}\). It is a conditional expectation, using the belief induced by the recommendation.

Study Q2 Why do the notes write \(R=s\) instead of \(R_{-i}=s_{-i}\)?

Because the distribution \(\pi\) is defined over the full profile space \(S\). But once your recommendation \(s_i\) is fixed, writing \[ \Pr(R=s \mid R_i=s_i) \] is equivalent to writing \[ \Pr(R_{-i}=s_{-i} \mid R_i=s_i). \] The first notation keeps the joint distribution explicit; the second is often easier to interpret.

Study Q3 Why does the convex hull of Nash equilibria sit inside correlated equilibria?

Two facts are enough:

  • Every mixed Nash equilibrium is a correlated equilibrium.
  • The correlated-equilibrium set is convex.

So if \(\pi^1,\dots,\pi^k\) are mixed Nash equilibria and \(\lambda_1,\dots,\lambda_k\) are nonnegative weights summing to 1, then \[ \sum_{m=1}^k \lambda_m \pi^m \] is again a correlated equilibrium.

Study Q4 What is Kakutani's theorem really saying?

It generalizes fixed points from ordinary functions to correspondences (set-valued maps). Instead of asking for \[ f(x)=x, \] it asks for \[ x \in f(x). \] In game theory, \(f(x)\) is the set of best responses. So a fixed point means the strategy profile is one of its own best responses, which is exactly equilibrium.

Minimal example. Let \(A=[0,1]\) and define \[ f(x)= \begin{cases} [1/2,1], & x\le 1/2,\\ [0,1/2], & x>1/2. \end{cases} \] Then \(x=1/2\) is a fixed point because \(1/2 \in f(1/2)\).
Study Q5 Why does a pure equilibrium exist in the infinite-game theorem?

Because the best-response correspondence over pure strategies satisfies Kakutani's conditions:

  • The product strategy space \(S=\prod_i S_i\) is compact, convex, and nonempty.
  • Best responses exist by Weierstrass.
  • Best-response sets are convex because the payoff is concave in own strategy.
  • The graph is closed by continuity.

Kakutani then gives \(s^* \in B(s^*)\), which means every player is best responding in pure strategies.