Chapter 3: Nash Equilibrium

Algorithmic Game Theory
Formal definitionBest responseDominated strategiesMixed strategiesExistence theoremCournot and BertrandEquilibrium selectionCorrelated equilibriumMinimaxSecurity gamesPPAD
Sercan KÜLCÜ · Giresun University · Computer Engineering

Foundations

Warm-up Game: The Prisoner's Dilemma

Rules: Sam and John are questioned in separate rooms. Without knowing the other's choice, each either stays Silent or Betrays, and serves the sentence shown in the table. Payoff = −(years in prison), so bigger is better.

JohnSamSilentBetraySilentBetray(−1,−1)(−3,0)(0,−3)(−2,−2)?Sam?Johnchoices are hidden,then revealed together

Cell (Sam, John): payoff = −(years in prison)

Question 1In Sam's place, not knowing what John will do, would you stay silent or betray?
Question 2Against a computer playing Tit-for-Tat (silent in round one, then copies your previous move), what is the fewest total years you can serve over ten rounds?

Foundations

Learning Outcomes

DefinitionState the formal definition of Nash equilibrium and relate it to best responses.
Pure equilibriaMark best responses in a payoff matrix, apply iterated elimination of dominated strategies, and find equilibria of continuous games such as Cournot and Bertrand.
Mixed equilibriaCompute mixed equilibria by hand and in code using the indifference principle; find the minimax value of zero-sum and security games.
Existence and complexityInterpret Nash's existence theorem, the problem of multiple equilibria, and the PPAD-completeness of computing an equilibrium.

Foundations

Who Was John Nash?

  • American mathematician John Forbes Nash Jr. (1928–2015).
  • 1950: a short Princeton PhD thesis, Non-Cooperative Games, and a two-page note in PNAS proving existence.
  • 1994 Nobel Prize in Economics (shared with Harsanyi and Selten).
  • 2015 Abel Prize for his work on partial differential equations.
  • His life inspired the film A Beautiful Mind (2001).
Portrait of an elderly John Nash holding his glasses.

John F. Nash Jr.

His contribution: he extended von Neumann's solution for two-player zero-sum games to general n-player (non-zero-sum) games.

Foundations

Strategic Behavior and Applications

  • Each player tries to maximize their own payoff.
  • But each payoff also depends on what the others choose: “What should I do?” cannot be separated from “What will they do?”
Price competitionTwo firms cut prices; both lose profit.Prisoner's Dilemma structure
Advertising / arms racesEveryone spends; relative positions stay the same.dominant strategy: spend
Penalty kicksKicker and goalkeeper must be unpredictable.mixed strategy, zero-sum
Choosing a laneIf everyone moves to the fast lane, it slows down.congestion game
Choosing a standardEveryone gains by adopting the same technology.coordination, multiple equilibria
Sharing resourcesShared bandwidth, servers, roads.Price of Anarchy

Foundations

A State Nobody Regrets

7 and 3 peopleTill 1Till 2last one moves: 4th, not 7th5 and 5 peopleTill 1Till 2a mover drops to 6th:nobody moves = equilibrium
  • The cafeteria has two tills; everyone wants a shorter wait and may switch queues freely.
  • 7 – 3: the student at the back of the long queue gains by switching → this state will not last.
  • 5 – 5: whoever switches ends up further back → nobody moves.
  • Regret test: “Had I seen what everyone else did, would I change my choice?” If everyone answers no, the state is a Nash equilibrium.
An equilibrium is self-enforcing: everyone correctly anticipates the others' behavior and does the best they can in response. It does not mean “the best outcome” or “a fair outcome”.

Foundations

Nash Equilibrium: Formal Definition

Normal-form game G = (N, (Si), (ui)):

  • N = {1, …, n} players; Si is player i's strategy set.
  • Strategy profile s = (s1, …, sn); s−i: the strategies of everyone except i.
  • ui(s): player i's payoff (ui is the payoff function).
Nash equilibrium: a profile s* such that for every player i and every si ∈ Si
ui(si*, s−i*) ≥ ui(si, s−i*)
Unilateral deviationWith the others fixed, no player can raise their payoff by switching strategy.
Self-enforcing agreementIf you believe the others will stick to it, you stick to it too.
Stability ≠ efficiencyAn equilibrium guarantees neither fairness nor efficiency (Pareto optimality).

Foundations

Common Mistakes: Defining Nash Equilibrium

“The equilibrium is the best outcome.”No. In the Prisoner's Dilemma the equilibrium (B, B) is worse for both players than (S, S). Equilibrium means stability, not efficiency.
“They gain if both switch together.”The definition only checks unilateral deviations. A joint deviation (a coalition) does not break a Nash equilibrium.
“One satisfied player is enough.”The test is run for every player; if even one player wants to deviate, the cell is not an equilibrium.
“The row player also picks the column.”Each player changes only their own strategy: the row player moves up and down within a column, the column player left and right within a row.
“Every game has exactly one pure equilibrium.”Matching Pennies has none; Stag Hunt has two pure equilibria and one mixed.
Cost or payoff?For costs such as prison years, smaller is better. Negate them to get payoffs: 2 years → −2.

Best Response

Best Response

  • Player i's best response to the others' choice s−i is the set of strategies that maximize i's payoff:
    BRi(s−i) = argmaxsi ∈ Si ui(si, s−i)
  • Equivalently: s* is a Nash equilibrium ⇔ si* ∈ BRi(s−i*) for every i. Equilibrium = mutual best responses.
1. Row playerIn each column, mark the row player's largest payoff (u1).
2. Column playerIn each row, mark the column player's largest payoff (u2).
3. IntersectionCells with both payoffs marked are the pure-strategy Nash equilibria.

Cost on an m × n matrix: O(mn), since each cell is compared with one column maximum and one row maximum.

Best Response

Finding the Equilibrium by Marking Best Responses

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)Payoff = −(years in prison)Sam is row, John is column playerCell: (Sam's payoff, John's payoff)

Payoffs are prison years with a minus sign (−1 = 1 year). Each player wants their own number to be larger.

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)John Silent → Sam: Betray (0 > −1)

Suppose John stays Silent. Sam compares only the first numbers in that column: 0 > −1, so his best response is Betray.

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)John Silent → Sam: Betray (0 > −1)John Betray → Sam: Betray (−2 > −3)

If John Betrays: −2 > −3, so Sam's best response is again Betray.

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)John Silent → Sam: Betray (0 > −1)John Betray → Sam: Betray (−2 > −3)⇒ Betray dominant for Sam

Both marks sit in the same row: whatever John does, Betray is better. It is a dominant strategy.

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)John Silent → Sam: Betray (0 > −1)John Betray → Sam: Betray (−2 > −3)⇒ Betray dominant for SamSam Silent → John: Betray (0 > −1)

Now John. If Sam stays Silent, John compares the second numbers: 0 > −1, so Betray.

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)John Silent → Sam: Betray (0 > −1)John Betray → Sam: Betray (−2 > −3)⇒ Betray dominant for SamSam Silent → John: Betray (0 > −1)Sam Betray → John: Betray (−2 > −3)

If Sam Betrays: −2 > −3, so Betray for John too. The game is symmetric, so this is no surprise.

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)John Silent → Sam: Betray (0 > −1)John Betray → Sam: Betray (−2 > −3)⇒ Betray dominant for SamSam Silent → John: Betray (0 > −1)Sam Betray → John: Betray (−2 > −3)⇒ Betray dominant for John too

The only cell with both marks is d = (Betray, Betray): nobody wants to deviate alone. Nash equilibrium (−2, −2).

JohnSilentBetraySamSilenta−1−1(,)b−30(,)Betrayc0−3(,)d−2−2(,)John Silent → Sam: Betray (0 > −1)John Betray → Sam: Betray (−2 > −3)⇒ Betray dominant for SamSam Silent → John: Betray (0 > −1)Sam Betray → John: Betray (−2 > −3)⇒ Betray dominant for John tooa = (−1, −1) is better for both!

Yet a = (Silent, Silent) would give each only 1 year. The equilibrium is not Pareto optimal: that is the dilemma.

Best Response

Simple Example: Mark the Best Responses

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Each cell holds (u1, u2): the row player's payoff, then the column player's. Nothing marked yet.

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

If the column player plays X, the row player compares u1: A:4, B:2, C:1 → best is A (4).

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

If the column player plays Y, the row player compares u1: A:0, B:3, C:2 → best is B (3).

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

If the column player plays Z, the row player compares u1: A:1, B:0, C:3 → best is C (3).

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

If the row player plays A, the column player compares u2: X:3, Y:2, Z:1 → best is X (3).

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

If the row player plays B, the column player compares u2: X:1, Y:3, Z:2 → best is Y (3).

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

If the row player plays C, the column player compares u2: X:4, Y:0, Z:2 → best is X (4).

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cells with both payoffs marked: (A, X) = (4, 3) and (B, Y) = (3, 3) → two pure Nash equilibria. In cells such as (C, Z), at least one player wants to deviate.

Column playerRow playerXYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Deviation test at (A, X): if the row player switches to B or C, they get 2 or 1 (< 4); if the column player switches to Y or Z, they get 2 or 1 (< 3). Nobody gains ✓

Best Response

Step by Step: The Underlining Method (with Ties)

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

Ayşe and Bora pick a language for a joint project; cells are (Ayşe, Bora). Rule: for Ayşe, underline the largest first number in each column; for Bora, the largest second number in each row.

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

If Bora picks Python, Ayşe looks only at the first numbers: 4, 0, 1 → underline the largest, 4 (Python).

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

If Bora picks Java: 1, 2, 0 → the largest is 2 (Java).

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

If Bora picks C: 1, 1, 1 → a three-way tie, so underline all three. Every tied option is a best response; marking only one is a common mistake.

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

Now Bora. If Ayşe picks Python, Bora looks at the second numbers: 2, 1, 0 → 2 (Python).

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

If Ayşe picks Java: 1, 4, 2 → 4 (Java).

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

If Ayşe picks C: 0, 2, 1 → 2 (Java).

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

Cells with both numbers underlined: (Python, Python) = (4, 2) and (Java, Java) = (2, 4). In (C, C) only the first number is underlined: Bora switches to Java and gets 2 instead of 1.

BoraAyşePythonJavaCPythonJavaC(4,2)(1,1)(1,0)(0,1)(2,4)(1,2)(1,0)(0,2)(1,1)

Strict elimination is not enough here: Java removes column C, then Python removes row C. The remaining 2×2 has no dominated strategy, so two equilibria survive. Underlining found both directly.

Best Response

Common Mistakes: Marking Best Responses

Column playerRow playerLeftRightTopBottom(3,1)(0,0)(1,0)(2,2)row player: same column,compares first numbersColumn playerRow playerLeftRightTopBottom(3,1)(0,0)(1,0)(2,2)column player: same row,compares second numbers
Wrong directionSearching along the row for the row player's best. They only change rows, so compare within the same column.
Wrong numberUsing the first numbers for the column player. The column player wants the second number to be large.
One mark for a tieIf the maximum appears twice, both are best responses; mark both.
One mark is enoughA cell marked for only one player is not an equilibrium; it needs two marks.
Hunting the biggest cellAn equilibrium is not the cell with the largest numbers but the cell of mutual best responses.
Skipping the checkRun the deviation test on every candidate: does any player get more by switching?

Best Response

Three Players: The Group Project Game

Three students choose Join or Skip at once. If at least two join, the project succeeds and everyone gets 3; joining costs 1. k: number of joiners.

JoinersPayoffsUnilateral deviationEquilibrium?
k = 3 (everyone)3 − 1 = 2 eachA quitter still gets the project: 3 > 2No
k = 2joiners 2, the other 3A quitting joiner sinks it: 0 < 2; the skipper joining gets 2 < 3Yes (3 profiles)
k = 1joiner −1, others 0The joiner quits: 0 > −1No
k = 0 (nobody)0 eachA lone joiner gets −1: −1 < 0Yes
Four pure equilibriaThe three profiles where exactly two join, plus nobody joining. With n players the test is the same: each player, others fixed.
Coordination failure“Nobody joins” gives everyone 0, yet no one can break it alone; joint deviations are outside the Nash definition.

Best Response

Code Trace: Finding Pure Nash Equilibria

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

U1 is the row player's payoff matrix, U2 the column player's. m = n = 3, eq = [].

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (A, X): U1 = 4 is the column maximum and U2 = 3 the row maximum → condition true.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Append (0, 0): eq = [(0, 0)].

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (A, Y): U1 = 0 < 3 and U2 = 2 < 3 → false, not an equilibrium.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (A, Z): U1 = 1 < 3 and U2 = 1 < 3 → false, not an equilibrium.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (B, X): U1 = 2 < 4 and U2 = 1 < 3 → false, not an equilibrium.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (B, Y): U1 = 3 is the column maximum and U2 = 3 the row maximum → condition true.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Append (1, 1): eq = [(0, 0), (1, 1)].

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (B, Z): U1 = 0 < 3 and U2 = 2 < 3 → false, not an equilibrium.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (C, X): U1 = 1 < 4 → false, not an equilibrium.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (C, Y): U1 = 2 < 3 and U2 = 0 < 4 → false, not an equilibrium.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

Cell (C, Z): U2 = 2 < 4 → false, not an equilibrium.

XYZABC(4,3)(0,2)(1,1)(2,1)(3,3)(0,2)(1,4)(2,0)(3,2)

All 9 cells scanned: return [(0, 0), (1, 1)] → (A, X) and (B, Y). Total work O(mn·(m+n)); O(mn) if the maxima are precomputed.

def pure_nash(U1, U2):
    m, n, eq = len(U1), len(U1[0]), []
    for i in range(m):
        for j in range(n):
            br1 = max(U1[k][j] for k in range(m))
            br2 = max(U2[i][l] for l in range(n))
            if U1[i][j] == br1 and U2[i][j] == br2:
                eq.append((i, j))
    return eq

Best Response

Continuous Strategies: Cournot Duopoly

034612034612y₁ (firm 1)y₂ (firm 2)BR₁: y₁ = (12 − y₂)/2BR₂: y₂ = (12 − y₁)/2cartel (3, 3)Nash (4, 4)
  • Two firms simultaneously choose quantities y₁, y₂ ≥ 0 (an infinite strategy set). Price P = 12 − (y₁ + y₂), zero cost.
  • Profit u₁ = y₁ (12 − y₁ − y₂). Setting the derivative 12 − 2y₁ − y₂ to 0 ⇒ BR₁(y₂) = max{0, (12 − y₂)/2} (quantities cannot be negative; if y₂ ≥ 12, it is best to produce nothing).
  • Nash equilibrium = where the best-response lines cross: y₁ = y₂ = 4, price 4, profit 16 each.
  • A cartel (3 each) gives both 18, but it is not an equilibrium: BR₁(3) = 4.5 and the deviating firm earns 20.25.
If the firms take turns best-responding, output moves (6, 3), then (4.5, 3.75), … and approaches (4, 4). Cournot, 1838.

Best Response

How Much Should I Produce Given My Rival's Output?

u₁(y₁) = y₁(12 − y₁ − y₂)  ⇒  BR₁(y₂) = max{0, (12 − y₂)/2}

The more my rival produces, the less I should produce; once y₂ ≥ 12, producing nothing is best. The only point where both firms best-respond to each other is y₁ = y₂ = 4.

Best Response

Bertrand: Competing on Price Instead of Quantity

  • Two firms sell the same good; unit cost c = 2 TL, market size 100 units. They choose prices simultaneously: the cheaper firm takes the whole market, and equal prices split it in half.
  • At (5, 5) each earns (5 − 2) · 50 = 150. Undercutting to 4.99 yields ≈ 2.99 · 100 ≈ 299, so deviating pays. The same holds at every common price above c.
  • Pricing below c loses money. The unique Nash equilibrium is p₁ = p₂ = c = 2, with zero profit.
2345678910p = cundercutting any common price slightly paysprice (TL)
Competing onEquilibriumFirm profit
Cournot (quantity), P = 12 − Y, zero costy = 4 each, price 416 each
Bertrand (price), cost cp = c0
Common mistake: “With two firms, profits stay high.” Under price competition even two firms drive price down to cost (the Bertrand paradox); profit survives only through product differentiation, capacity limits or repeated interaction. This holds for continuous prices; with 1 TL steps, prices just above c can also be equilibria.

Dominated Strategies

Dominant and Dominated Strategies

Strict dominancesi′ strictly dominates si if ui(si′, s−i) > ui(si, s−i) for every s−i.A rational player never plays a strictly dominated strategy.
Weak dominance≥ for every s−i and > for at least one.Deleting it may lose equilibria; order can matter.
  • A dominant strategy dominates all of the player's other strategies (Betray in the Prisoner's Dilemma); no need to predict the opponent. If every player has one, that profile is a dominant-strategy equilibrium. Not every game has one (RPS).
  • Iterated elimination (IESDS): delete strictly dominated strategies and repeat. Assumes rationality is common knowledge.
  • Under strict dominance order does not matter and no equilibrium is lost. A single remaining cell is the unique Nash equilibrium.
  • Test against each opposing choice; “better on average” is not enough. In the 2/3 game, numbers above 60 sometimes tie with 60: that is weak dominance.

Dominated Strategies

Back to the Game: The Prisoner's Dilemma

JohnSamSilentBetraySilentBetray(−1,−1)(−3,0)(0,−3)(−2,−2)

Payoff = −years: each player's best response marked

  • Silent or betray? Theory's answer: Betray is dominant: if John is silent, 0 > −1; if he betrays, −2 > −3. The unique Nash equilibrium is (Betray, Betray): (−2, −2), two years each.
  • In practice: in one-shot play people cooperate far more often than theory predicts. A meta-analysis of experiments from 1958–1992 finds roughly half of participants cooperate, and communication raises that rate markedly (Sally, 1995). Trust, fairness and reciprocity also matter.
Sam's 10-round plan against John's Tit-for-TatSam's total years
Always Silent10 × 1 = 10
Always Betray0 + 9 × 2 = 18
Silent 9 rounds, Betray the last9 × 1 + 0 = 9 (minimum)

In the repeated game, betrayal is punished in later rounds; only the last round goes unpunished. This “last-round effect” is what backward induction is about.

Dominated Strategies

Iterated Elimination: A 3×3 Example

Column playerRow playerXYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Start: no row strictly dominates another (e.g. A vs B: 3 = 3). Look at the column player first.

Column playerRow playerXYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Over the remaining rows, Z strictly dominates X (u2: 4 > 2, 3 > 2, 6 > 1) → delete X.

Column playerRow playerXYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Over the remaining columns, C strictly dominates B (u1: 2 > 0, 6 > 5) → delete B.

Column playerRow playerXYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Over the remaining rows, Z strictly dominates Y (u2: 4 > 3, 6 > 0) → delete Y.

Column playerRow playerXYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Over the remaining columns, C strictly dominates A (u1: 6 > 1) → delete A.

Column playerRow playerXYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

One cell remains: (C, Z) = (6, 6). It is the game's unique Nash equilibrium (marking best responses gives the same answer).

Dominated Strategies

Step by Step: Each Elimination Enables the Next

Column playerRow playerLeftCenterRightTopMiddleBottom(1,5)(5,6)(6,2)(3,5)(6,0)(2,2)(2,2)(0,2)(1,5)

Neither player has a dominant strategy, and at first there is only one strict dominance. Compare rows using only the first numbers, columns using only the second.

Column playerRow playerLeftCenterRightTopMiddleBottom(1,5)(5,6)(6,2)(3,5)(6,0)(2,2)(2,2)(0,2)(1,5)

Middle beats Bottom in every column: 3 > 2, 6 > 0, 2 > 1 → delete Bottom.

Column playerRow playerLeftCenterRightTopMiddleBottom(1,5)(5,6)(6,2)(3,5)(6,0)(2,2)(2,2)(0,2)(1,5)

In the remaining rows, Left beats Right: 5 > 2 (Top), 5 > 2 (Middle) → delete Right. Not possible while Bottom remained: there Right 5 > Left 2.

Column playerRow playerLeftCenterRightTopMiddleBottom(1,5)(5,6)(6,2)(3,5)(6,0)(2,2)(2,2)(0,2)(1,5)

In the remaining columns, Middle beats Top: 3 > 1, 6 > 5 → delete Top. While Right remained, Top survived there with 6 > 2.

Column playerRow playerLeftCenterRightTopMiddleBottom(1,5)(5,6)(6,2)(3,5)(6,0)(2,2)(2,2)(0,2)(1,5)

Only the Middle row is left: Left 5, Center 0 → delete Center.

Column playerRow playerLeftCenterRightTopMiddleBottom(1,5)(5,6)(6,2)(3,5)(6,0)(2,2)(2,2)(0,2)(1,5)

The survivor (Middle, Left) = (3, 5) is the unique Nash equilibrium. (Top, Center) = (5, 6) is better for both but not an equilibrium: the row player switches to Middle and gets 6.

Dominated Strategies

Code Trace: Iterated Elimination of Dominated Strategies

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Remaining rows and columns: R = [0, 1, 2] (A, B, C), S = [0, 1, 2] (X, Y, Z).

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

No row is dominated (x = []). Column X is dominated by Z → y = [0].

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

S.remove(0): delete column X; the next round rechecks the rows.

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Row B is dominated by C → x = [1]. Columns are checked too: y = [].

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

R.remove(1): delete row B; the loop restarts on the smaller game.

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

No row is dominated (x = []). Column Y is dominated by Z → y = [1].

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

S.remove(1): delete column Y; the next round rechecks the rows.

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

Row A is dominated by C → x = [0]. Columns are checked too: y = [].

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

R.remove(0): delete row A; the loop restarts on the smaller game.

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

No row or column is dominated: x = y = [].

XYZABC(3,2)(3,3)(1,4)(3,2)(0,4)(5,3)(2,1)(2,0)(6,6)

return [2], [2] → only (C, Z) is left. Four eliminations removed X, B, Y, A in turn.

def eliminate(U1, U2):
    R, S = [0, 1, 2], [0, 1, 2]   # remaining
    while True:
        x = [r for r in R for k in R if
             all(U1[k][c] > U1[r][c] for c in S)]
        y = [c for c in S for d in S if
             all(U2[r][d] > U2[r][c] for r in R)]
        if x: R.remove(x[0])      # drop a row
        elif y: S.remove(y[0])    # drop a column
        else: return R, S

Dominated Strategies

Dominance by a Mixed Strategy

u1LeftRight
Top40
Middle11
Bottom04
  • No pure row dominates Middle: it beats Top under Right (1 > 0) and Bottom under Left (1 > 0).
  • Mixing Top and Bottom ½–½ gives the row player 2 in every column: 2 > 1 and 2 > 1 → Middle is strictly dominated by a mixed strategy and can be deleted.
  • In the plot, Middle's line (1) lies below the upper envelope of Top and Bottom (at least 2) for every q: Middle is a best response to no belief.
12340½1q = P(column player Left)expected payoffTop: 4qBottom: 4 − 4qMiddle: 1½Top + ½Bottom: 2
Common mistake: “No single pure strategy beats it, so it cannot be deleted.” Mixtures count too; when eliminating, try pure strategies first, then simple mixtures such as ½–½.

Dominated Strategies

The Trap of Weak Dominance

ColumnRowLeftRightTopBottom(1,1)(0,1)(1,1)(1,1)
  • For the row player, Bottom weakly dominates Top (1 = 1, 1 > 0). The column player is always indifferent.
  • Pure Nash equilibria: (Top, Left), (Bottom, Left), (Bottom, Right), three in all.
  • Deleting Top loses the (Top, Left) equilibrium; different elimination orders can leave different results.
  • Lesson: strict dominance is safe; weak dominance can lose equilibria.
Strict elimination does not always shrink a game to one cell: in Battle of the Sexes nothing is dominated. Then we turn to best-response marking and mixed strategies.

Dominated Strategies

Common Mistakes: Eliminating Dominated Strategies

Wrong numbersComparing two rows using the column player's payoffs. Row ↔ first number, column ↔ second number.
“Mostly bigger” is enoughStrict dominance needs bigger in every remaining column; a single tie or loss stops the elimination.
Counting deleted strategiesA deleted column no longer takes part. In an elimination chain, a strategy (e.g. Top) may become removable only after another (Right) is gone.
Mistaking weak for strictWith a tie, the dominance is weak; deleting may lose equilibria and the result may depend on the order.
Dominant ≠ dominatedA dominant strategy beats all others; a dominated one loses to at least one. Elimination works even without a dominant strategy.
Every survivor is an equilibriumIf elimination leaves several cells (e.g. a 2×2 in the language game), find the equilibria of the reduced game by underlining.

Mixed Strategies

Mixed Strategies and the Indifference Principle

  • A mixed strategy assigns a probability to each pure strategy (e.g. Rock 1/3, Paper 1/3, Scissors 1/3). Payoffs become expected payoffs. In 2×2 games, p = σ1(first row), q = σ2(first column).
  • In games with no pure equilibrium (Matching Pennies, RPS, penalty kicks, inspection, bluffing), players want to be unpredictable. The RPS equilibrium is (1/3, 1/3, 1/3): every action earns 0. Any other mix can be exploited.
  • Support: the pure strategies played with positive probability.
Indifference principle: in equilibrium, each player earns the same expected payoff from every pure strategy in their support (strategies outside it earn no more). Otherwise they would shift all probability to the better one.
A surpriseThe row player's mix is computed from the column player's payoffs (and vice versa).
InterpretationProbabilities can be read as “rolling dice” or as behavior shares in a large population (e.g. directions of penalty kicks).

Mixed Strategies

Computing a Mixed Equilibrium: The Inspection Game

ManagerWorkerInspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)
Inspection gameRow: worker, column: manager. Wage 4, effort cost 2, output 6, inspection cost 1. A shirker who is caught gets no wage. Example: in Work–Inspect the worker gets 4 − 2 = 2 and the manager 6 − 4 − 1 = 1.

Arrows: in every cell some player gains by deviating → best responses form a cycle, no pure Nash equilibrium.
InspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)0½¾1−4−2024q (probability of Inspect)expected payofflines appear once p and q are chosen
  1. The worker plays Work with probability p, the manager Inspect with probability q.
InspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)0½¾1−4−2024q (probability of Inspect)expected payoffWorkShirkWorker's payoff
  1. The worker plays Work with probability p, the manager Inspect with probability q.
  2. Worker: E[Work] = 2q + 2(1−q) = 2;  E[Shirk] = 0·q + 4(1−q) = 4 − 4q
InspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)0½¾1−4−2024q (probability of Inspect)expected payoffWorkShirkWorker's payoff
  1. The worker plays Work with probability p, the manager Inspect with probability q.
  2. Worker: E[Work] = 2q + 2(1−q) = 2;  E[Shirk] = 0·q + 4(1−q) = 4 − 4q
  3. Worker indifferent: 2 = 4 − 4q ⇒ q* = 1/2
InspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)0½¾1−4−2024p (probability of Work)expected payoffInspectDon'tManager's payoff
  1. The worker plays Work with probability p, the manager Inspect with probability q.
  2. Worker: E[Work] = 2q + 2(1−q) = 2;  E[Shirk] = 0·q + 4(1−q) = 4 − 4q
  3. Worker indifferent: 2 = 4 − 4q ⇒ q* = 1/2
  4. Manager: E[Inspect] = 1·p − 1·(1−p) = 2p − 1;  E[Don't] = 2p − 4(1−p) = 6p − 4
InspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)0½¾1−4−2024p (probability of Work)expected payoffInspectDon'tManager's payoff
  1. The worker plays Work with probability p, the manager Inspect with probability q.
  2. Worker: E[Work] = 2q + 2(1−q) = 2;  E[Shirk] = 0·q + 4(1−q) = 4 − 4q
  3. Worker indifferent: 2 = 4 − 4q ⇒ q* = 1/2
  4. Manager: E[Inspect] = 1·p − 1·(1−p) = 2p − 1;  E[Don't] = 2p − 4(1−p) = 6p − 4
  5. Manager indifferent: 2p − 1 = 6p − 4 ⇒ p* = 3/4
InspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)0½¾1−4−2024p (probability of Work)expected payoffInspectDon'tManager's payoff
  1. The worker plays Work with probability p, the manager Inspect with probability q.
  2. Worker: E[Work] = 2q + 2(1−q) = 2;  E[Shirk] = 0·q + 4(1−q) = 4 − 4q
  3. Worker indifferent: 2 = 4 − 4q ⇒ q* = 1/2
  4. Manager: E[Inspect] = 1·p − 1·(1−p) = 2p − 1;  E[Don't] = 2p − 4(1−p) = 6p − 4
  5. Manager indifferent: 2p − 1 = 6p − 4 ⇒ p* = 3/4
  6. Equilibrium: (p*, q*) = (3/4, 1/2). Expected payoffs: worker 2, manager 1/2.
InspectDon'tWorkShirk(2,1)(2,2)(0,−1)(4,−4)0½¾1−4−2024p (probability of Work)expected payoffInspectDon'tManager's payoff
  1. The worker plays Work with probability p, the manager Inspect with probability q.
  2. Worker: E[Work] = 2q + 2(1−q) = 2;  E[Shirk] = 0·q + 4(1−q) = 4 − 4q
  3. Worker indifferent: 2 = 4 − 4q ⇒ q* = 1/2
  4. Manager: E[Inspect] = 1·p − 1·(1−p) = 2p − 1;  E[Don't] = 2p − 4(1−p) = 6p − 4
  5. Manager indifferent: 2p − 1 = 6p − 4 ⇒ p* = 3/4
  6. Equilibrium: (p*, q*) = (3/4, 1/2). Expected payoffs: worker 2, manager 1/2.
  7. With a harsher penalty (a caught shirker gets −2): q* becomes 1/3, while p* = ¾ stays the same!

Mixed Strategies

When the Penalty Changes, Whose Probability Moves?

Worker indifferent: 2 = c·q + 4(1 − q) ⇒ q* = 2/(4 − c)  ·  Manager indifferent: 2p − 1 = 6p − 4 ⇒ p* = 3/4

c changes the worker's payoff, yet the worker stays at p* = 3/4; the manager's q* moves. Each player's mix is set by the opponent's payoffs.

Mixed Strategies

Step by Step: A Mixed Equilibrium by Indifference

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)

Ayşe hides (row), Bora seeks (column); cells show fun points (Ayşe, Bora). Bora gets 3 for finding her in the garden, 1 in the house; Ayşe gets 2 for not being found in the garden, 1 in the house.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)Deviation cycle (Ayşe, Bora)(Garden, Garden)(Garden, House)(House, House)(House, Garden)Ayşe→HouseBora→HouseAyşe→GardenBora→Garden

After underlining, no cell has two marks: best responses form a cycle (arrows). No pure Nash equilibrium → look for a mixed one.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)p = P(Ayşe Garden), q = P(Bora Garden)

Variables: Ayşe picks Garden with probability p, Bora with probability q. If Ayşe mixes, both options must give her the same payoff; that equation gives q.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)p = P(Ayşe Garden), q = P(Bora Garden)Ayşe: E[Garden] = 0·q + 2·(1 − q) = 2 − 2q

Ayşe in the Garden: Bora searches the garden with prob. q (0) and the house with 1 − q (2). Expected payoff 2 − 2q.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)p = P(Ayşe Garden), q = P(Bora Garden)Ayşe: E[Garden] = 0·q + 2·(1 − q) = 2 − 2qAyşe: E[House] = 1·q + 0·(1 − q) = q

Ayşe in the House: 1 if Bora searches the garden (q), 0 if the house. Expected payoff q.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)p = P(Ayşe Garden), q = P(Bora Garden)Ayşe: E[Garden] = 0·q + 2·(1 − q) = 2 − 2qAyşe: E[House] = 1·q + 0·(1 − q) = q2 − 2q = q ⇒ 2 = 3q ⇒ q* = 2/3

Equate and solve: 2 − 2q = q ⇒ q* = 2/3. Bora must search the garden with probability 2/3 to keep Ayşe indifferent.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)p = P(Ayşe Garden), q = P(Bora Garden)Ayşe: E[Garden] = 0·q + 2·(1 − q) = 2 − 2qAyşe: E[House] = 1·q + 0·(1 − q) = q2 − 2q = q ⇒ 2 = 3q ⇒ q* = 2/3Bora: E[Garden] = 3·p + 0·(1 − p) = 3pBora: E[House] = 0·p + 1·(1 − p) = 1 − p

Same for Bora: given Ayşe's p, his two options give E[Garden] = 3p and E[House] = 1 − p.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)p = P(Ayşe Garden), q = P(Bora Garden)Ayşe: E[Garden] = 0·q + 2·(1 − q) = 2 − 2qAyşe: E[House] = 1·q + 0·(1 − q) = q2 − 2q = q ⇒ 2 = 3q ⇒ q* = 2/3Bora: E[Garden] = 3·p + 0·(1 − p) = 3pBora: E[House] = 0·p + 1·(1 − p) = 1 − p3p = 1 − p ⇒ 4p = 1 ⇒ p* = 1/4

3p = 1 − p ⇒ p* = 1/4. Note: p* comes from Bora's numbers, not Ayşe's.

Bora (seeks)Ayşe (hides)GardenHouseGardenHouse(0,3)(2,0)(1,0)(0,1)p = P(Ayşe Garden), q = P(Bora Garden)Ayşe: E[Garden] = 0·q + 2·(1 − q) = 2 − 2qAyşe: E[House] = 1·q + 0·(1 − q) = q2 − 2q = q ⇒ 2 = 3q ⇒ q* = 2/3Bora: E[Garden] = 3·p + 0·(1 − p) = 3pBora: E[House] = 0·p + 1·(1 − p) = 1 − p3p = 1 − p ⇒ 4p = 1 ⇒ p* = 1/4Payoffs: Ayşe 2/3, Bora 3/4

Check: 0 < 1/4 < 1 and 0 < 2/3 < 1 ✓. Payoffs: Ayşe 2 − 2·(2/3) = 2/3, Bora 3·(1/4) = 3/4. Ayşe likes the garden, yet hides there only 25% of the time.

Mixed Strategies

Code Trace: mixed_2x2(A, B)

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)

Battle of the Sexes: A holds the row player's payoffs, B the column player's (0 = Opera, 1 = Football).

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)

Denominator from the row player's payoff differences: da = 3 − 0 − 0 + 2 = 5.

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)column probability2/53/5

da ≠ 0, so the check passes. Setting u₁(Opera, q) = 3q equal to u₁(Football, q) = 2(1 − q) ⇒ q = (2 − 0)/5 = 2/5: the column player plays Opera with probability 2/5.

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)column probability2/53/5

Now the column player's payoffs: db = 2 − 0 − 0 + 3 = 5.

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)row probability3/52/5column probability2/53/5

db ≠ 0. Setting u₂(p, Opera) = 2p equal to u₂(p, Football) = 3(1 − p) ⇒ p = 3/5: the row player plays Opera with probability 3/5.

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)row probability3/52/5column probability2/53/5

0 < p, q < 1 holds: both are genuine mixtures ⇒ an interior mixed equilibrium exists, so return None is skipped.

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)row probability3/52/5column probability2/53/5

Expected payoff u₁ = 3·(2/5) + 0·(3/5) = 6/5; u₂ = 2·(3/5) = 6/5. Worse than either pure equilibrium (2 or 3).

column playerrowOperaFootballOperaFootball(3,2)(0,0)(0,0)(2,3)row probability3/52/5column probability2/53/5

Result (3/5, 2/5, 6/5). In Free Money q = 0 ⇒ None. In the Prisoner's Dilemma da = 0 ⇒ the first check returns None.

from fractions import Fraction as F
def mixed_2x2(A, B):
    da = A[0][0] - A[0][1] - A[1][0] + A[1][1]
    if da == 0: return None     # degenerate
    q = F(A[1][1] - A[0][1], da)  # row indifferent
    db = B[0][0] - B[0][1] - B[1][0] + B[1][1]
    if db == 0: return None
    p = F(B[1][1] - B[1][0], db)  # col indifferent
    if not (0 < p < 1 and 0 < q < 1):
        return None               # no interior eq.
    u1 = A[0][0] * q + A[0][1] * (1 - q)
    return p, q, u1

Mixed Strategies

Mixed Equilibria of Six Classic Games

GamePure NEMixed NE (p*, q*)(u₁, u₂)In the mixed equilibrium
Matching Pennies01/2, 1/2(0, 0)Each outcome 25%; unexploitable
Chicken29/10, 9/10(−1/10, −1/10)Crash (1/10)² = 1%; with crash cost −k, swerve (k − 1)/k
Battle of the Sexes23/5, 2/5(6/5, 6/5)Miss each other 52%
Stag Hunt (stag 4, hare 3)23/4, 3/4(3, 3)Miscoordinate 37.5%; both stag ≈ 56%
Pure Coordination21/3, 1/3(10/3, 10/3)Mismatch ≈ 44%; a better party is chosen less
Free Money2none—For q > 0, Heads is strictly better; only boundary (0, 0)
p*, q*: probability of the first strategy. Chicken: (Swerve, Swerve) = (0, 0), (Swerve, Straight) = (−1, 1), (Straight, Straight) = (−10, −10). Pure Coordination: (Party, Party) = (10, 10), (Home, Home) = (5, 5). Free Money: (Heads, Heads) = (1, 1), else 0.

Mixed Strategies

Common Mistakes: Mixed Equilibrium

Finding p from your own numbersThe row player's p makes the column player indifferent, so it is computed from the column player's payoffs.
Forgetting (1 − q)E = 0·q + 2·(1 − q) = 2 − 2q, not “2q”. Each term is multiplied by its own probability.
Not checking the resultIf p or q falls outside (0, 1), there is no interior mixed equilibrium (usually some strategy is dominated).
Skipping pure equilibriaThe mixed calculation does not replace the pure equilibria; underline first, then look for the mixed one.
“Indifferent, so any p works”A player is indifferent among their own options, but in equilibrium plays the unique p that leaves the opponent indifferent.
Equating the wrong thingsSet equal the expected payoffs of one player's two options, not the two players' payoffs.

Existence and Equilibrium Selection

Nash's Existence Theorem (1950)

Theorem. Every game with finitely many players and finite strategy sets has at least one (mixed-strategy) Nash equilibrium.
  • Proof idea: a mixed profile in which everyone best-responds to everyone always exists; fixed-point theorems (Kakutani, Brouwer) guarantee it. That profile is a Nash equilibrium.
  • The guarantee is for mixed strategies; a pure equilibrium may not exist (Matching Pennies).
  • With infinite strategy sets an equilibrium may not exist (e.g. “whoever names the largest number wins”).
0½¾10½1p = P(Work)q = P(Inspect)(¾, ½)BRworker(q)BRmanager(p)

Inspection game: the two best-response curves cross at the equilibrium

Existence and Equilibrium Selection

Step by Step: Three Equilibria in Channel Sharing

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)

Two devices want to send a packet on the same channel. If both send, they collide (−2 each); if one sends, it gets 3 and the waiting one 1; if both wait, the slot is wasted (0). This is the Chicken structure.

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)

Device 1 (first numbers): if the other Sends, −2 < 1 → Wait; if the other Waits, 3 > 0 → Send. It wants to do the opposite of the other.

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)

Same for Device 2 (second numbers). Doubly underlined cells: (Send, Wait) = (3, 1) and (Wait, Send) = (1, 3): two pure equilibria.

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)q = P(Device 2 Sends)u₁(Send, q) = −2·q + 3·(1 − q) = 3 − 5q

The third equilibrium is mixed: q = P(Device 2 Sends). Device 1's expected payoff from Send is −2q + 3(1 − q) = 3 − 5q.

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)q = P(Device 2 Sends)u₁(Send, q) = −2·q + 3·(1 − q) = 3 − 5qu₁(Wait, q) = 1·q + 0·(1 − q) = q

From Wait: 1 if the other sends, 0 if it waits → q.

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)q = P(Device 2 Sends)u₁(Send, q) = −2·q + 3·(1 − q) = 3 − 5qu₁(Wait, q) = 1·q + 0·(1 − q) = q3 − 5q = q ⇒ q* = 1/2; symmetry: p* = 1/2

Indifference: 3 − 5q = q ⇒ q* = 1/2. The game is symmetric, so Device 2's indifference gives p* = 1/2.

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)(Send, Wait)(Wait, Send)(½, ½)p = P(Device 1 Sends)q = P(Device 2 Sends)BR₁BR₂011

Best-response curves BR₁ (solid) and BR₂ (dashed) cross at three points: the two corners are the pure equilibria, the middle point the mixed one.

Device 2Device 1SendWaitSendWait(−2,−2)(3,1)(1,3)(0,0)Outcomes in the mixed equilibrium (½, ½):collision (S, S): ¼ → −2one sender (success): ½idle channel (W, W): ¼ → 0expected payoff: ½ (3 or 1 when pure)

The mixed equilibrium is fair but inefficient: collisions ¼ of the time, an idle channel another ¼; each device averages ½. Random backoff in network protocols is this randomization in practice.

Existence and Equilibrium Selection

How Many Equilibria? The Oddness Rule

GamePure equilibriaMixed equilibriaTotal
Prisoner's Dilemma101
Hide and Seek, Inspection, Matching Pennies011
Channel sharing, Chicken, Battle of the Sexes, Stag Hunt, Pure Coordination213
Language choice (3×3)21 (Ayşe picks Python with prob. 3/4, Bora with prob. 1/5)3
The Trap of Weak Dominance (ties)3infinitely many—
Free Money (ties)202 (even)
Oddness (Wilson, 1971)In nondegenerate games (no special ties among payoffs), the number of equilibria is finite and odd (1, 3, 5, …).
Rule of thumbFound two pure equilibria? There is very likely a mixed one too; look for it with indifference. With ties, the rule can fail.

Existence and Equilibrium Selection

Multiple Equilibria: Stag Hunt

Hunter 2Hunter 1StagHareStagHare(4,4)(0,3)(3,0)(3,3)

Stag Hunt: a stag can only be caught together (4 each); a hare guarantees 3 alone, whatever the partner does.

Hunter 2Hunter 1StagHareStagHare(4,4)(0,3)(3,0)(3,3)

Best responses coincide in two cells: (Stag, Stag) and (Hare, Hare). Which one will be played?

Hunter 2Hunter 1StagHareStagHare(4,4)(0,3)(3,0)(3,3)

The third equilibrium is mixed: if the other picks Stag with probability q, 4q = 3 ⇒ q = ¾. Stag is a best response only if you believe your partner comes with probability ≥ 75%.

Hunter 2Hunter 1StagHareStagHare(4,4)(0,3)(3,0)(3,3)

Payoff dominance: (Stag, Stag) gives both hunters more (4 > 3) → the efficiency criterion picks it.

Hunter 2Hunter 1StagHareStagHare(4,4)(0,3)(3,0)(3,3)

Risk dominance (Harsanyi–Selten): product of deviation losses at (S,S): (4−3)·(4−3) = 1; at (H,H): (3−0)·(3−0) = 9 → (Hare, Hare) is “safer”.

Players: Hunter 1 on rows, Hunter 2 on columns; cells are (Hunter 1, Hunter 2).

Existence and Equilibrium Selection

Equilibrium Selection: Which One Is Played?

Payoff dominancePick the equilibrium that makes everyone better off: (S, S) in Stag Hunt. Requires trust.
Risk dominanceUnsure what the other side will do? Pick the one with lower risk of loss: (H, H) in Stag Hunt.
Focal point (Schelling, 1960)Culture, habit or salience makes one equilibrium stand out: “If we meet in Istanbul, it's Taksim at noon.”
Correlated equilibriumAumann (1974): a shared signal (a traffic light) steers players to one of the equilibria; it resolves the Chicken game at an intersection.
Learning and evolutionIn repeated play, players learn by trial; which equilibrium they reach depends on initial behavior.
FairnessIn Battle of the Sexes each pure equilibrium favors one player; the mixed equilibrium gives both only 6/5.

Existence and Equilibrium Selection

Change the Assumption: What If a Player Makes Mistakes?

Prisoner's Dilemma (minus prison years)
SilentBetray
Silent(−1, −1)(−3, 0)
Betray(0, −3)(−2, −2)
Stag Hunt
StagHare
Stag(4, 4)(0, 3)
Hare(3, 0)(3, 3)
If the opponent errs with probability ε, does the best response change?
Prisoner's Dilemma (minus prison years)
SilentBetray
Silent(−1, −1)(−3, 0)
Betray(0, −3)(−2, −2)
Stag Hunt
StagHare
Stag(4, 4)(0, 3)
Hare(3, 0)(3, 3)
If the opponent errs with probability ε, does the best response change?
Prisoner's Dilemma: no changeOpponent Silent with probability p.
Silent: −1·p − 3·(1 − p) = 2p − 3
Betray: 0·p − 2·(1 − p) = 2p − 2
Betray wins by 1 for every p.
Prisoner's Dilemma (minus prison years)
SilentBetray
Silent(−1, −1)(−3, 0)
Betray(0, −3)(−2, −2)
Stag Hunt
StagHare
Stag(4, 4)(0, 3)
Hare(3, 0)(3, 3)
If the opponent errs with probability ε, does the best response change?
Prisoner's Dilemma: no changeOpponent Silent with probability p.
Silent: −1·p − 3·(1 − p) = 2p − 3
Betray: 0·p − 2·(1 − p) = 2p − 2
Betray wins by 1 for every p.
Stag Hunt: depends on the error rateOpponent slips to Hare with probability q.
Stag: 4·(1 − q)  ·  Hare: always 3
Stag is a best response only if 4(1 − q) ≥ 3: q ≤ 1/4. Otherwise play the risk-dominant Hare.
Dominant-strategy equilibria survive mistakes; coordination equilibria may not.

Existence and Equilibrium Selection

Correlated Equilibrium (Aumann, 1974): A Traffic Light at the Intersection

Driver 2Driver 1YieldGoYield(0, 0)(−1, 1)Go(1, −1)(−10, −10)Pure equilibria: (Yield, Go), (Go, Yield)Mixed equilibrium: yield 9/10crash 1/100, payoff −1/10

Two drivers reach an unsignalled intersection at once: give way (Yield) or drive through (Go). This is Chicken: each pure equilibrium favors one driver; without coordination, in the symmetric mixed equilibrium each driver yields with probability 9/10, they crash 1% of the time and the expected payoff is −1/10.

Driver 2Driver 1YieldGoYield(0, 0)μ = 1/3(−1, 1)μ = 1/3Go(1, −1)μ = 1/3(−10, −10)μ = 0The light picks one of 3 cells w.p. 1/3and tells each driver ONLY theirown recommendation.

A mediator (the traffic light) draws a cell from a joint distribution μ and privately tells each driver only their own recommendation.

Driver 2Driver 1YieldGoYield(0, 0)μ = 1/3(−1, 1)μ = 1/3Go(1, −1)μ = 1/3(−10, −10)μ = 0D1 told “Yield” ⇒ D2: Yield ½, Go ½Obey (Yield): 0·½ + (−1)·½ = −½Deviate (Go): 1·½ + (−10)·½ = −4.5−½ > −4.5 ⇒ obeying is better

If told “Yield”, Bayes' rule makes the opponent ½–½ Yield/Go; obeying (−½) beats deviating (−4.5).

Driver 2Driver 1YieldGoYield(0, 0)μ = 1/3(−1, 1)μ = 1/3Go(1, −1)μ = 1/3(−10, −10)μ = 0D1 told “Go” ⇒ D2 surely YieldsObey (Go): 1 Deviate (Yield): 01 > 0 ⇒ obeying is betterSymmetric for Driver 2.

If told “Go”, the opponent was surely told “Yield”; obeying (1) beats deviating (0). No recommendation is worth deviating from ⇒ a correlated equilibrium.

Driver 2Driver 1YieldGoYield(0, 0)μ = 1/3(−1, 1)μ = 1/3Go(1, −1)μ = 1/3(−10, −10)μ = 0Correlated equilibriumMixed NashPure Nashthe light's μ is here, payoff (0, 0)

Expected payoff 0 each (better than −1/10 in the mixed equilibrium), with no crashes. Every Nash equilibrium is a correlated equilibrium, but not conversely. Correlated equilibria can be found in polynomial time by linear programming; tuning light durations to traffic is a separate design choice.

Minimax and Complexity

Zero-Sum Games: Maximin and Minimax

Column (min)Row (max)XYZABC435216520

Zero-sum: each cell shows only the row player's payoff; the column player loses the same amount (u2 = −u1).

Column (min)Row (max)XYZABC435216520row min310

The row player is pessimistic: in each row, the worst value the opponent can leave: 3, 1, 0.

Column (min)Row (max)XYZABC435216520row min310

Maximin = max(3, 1, 0) = 3 (row A). The row player can guarantee at least 3.

Column (min)Row (max)XYZABC435216520row min310col max536

The column player is pessimistic too: in each column, the highest value the row player can get: 5, 3, 6.

Column (min)Row (max)XYZABC435216520row min310col max536

Minimax = min(5, 3, 6) = 3 (column Y). The column player can guarantee losing at most 3.

Column (min)Row (max)XYZABC435216520row min310col max536

Maximin = minimax = 3 ⇒ saddle point (A, Y): smallest in its row, largest in its column. It is a pure Nash equilibrium; the value of the game is v = 3.

Minimax and Complexity

No Saddle Point: Penalty Kicks

KeeperKickerLeftRightLeftRight60959070min6070max909504/1115060708090100p = P(kicker Left)goal %
  1. Row minima 60, 70 → maximin 70; column maxima 90, 95 → minimax 90. 70 < 90: no saddle point.
KeeperKickerLeftRightLeftRight6095907004/1115060708090100p = P(kicker Left)goal %keeper Leftkeeper Right
  1. Row minima 60, 70 → maximin 70; column maxima 90, 95 → minimax 90. 70 < 90: no saddle point.
  2. Kicker goes Left with prob. p: keeper Left ⇒ 60p + 90(1−p) = 90 − 30p; keeper Right ⇒ 95p + 70(1−p) = 70 + 25p.
KeeperKickerLeftRightLeftRight6095907004/1115060708090100p = P(kicker Left)goal %keeper Leftkeeper Rightlower envelope
  1. Row minima 60, 70 → maximin 70; column maxima 90, 95 → minimax 90. 70 < 90: no saddle point.
  2. Kicker goes Left with prob. p: keeper Left ⇒ 60p + 90(1−p) = 90 − 30p; keeper Right ⇒ 95p + 70(1−p) = 70 + 25p.
  3. The keeper minimizes the scoring rate, so the kicker's guarantee is the lower envelope of the two lines.
KeeperKickerLeftRightLeftRight6095907004/1115060708090100p = P(kicker Left)goal %keeper Leftkeeper Rightlower envelopev ≈ 79.1
  1. Row minima 60, 70 → maximin 70; column maxima 90, 95 → minimax 90. 70 < 90: no saddle point.
  2. Kicker goes Left with prob. p: keeper Left ⇒ 60p + 90(1−p) = 90 − 30p; keeper Right ⇒ 95p + 70(1−p) = 70 + 25p.
  3. The keeper minimizes the scoring rate, so the kicker's guarantee is the lower envelope of the two lines.
  4. Peak of the lower envelope: 90 − 30p = 70 + 25p ⇒ p* = 4/11 ≈ 0.36, v = 870/11 ≈ 79.1%.
KeeperKickerLeftRightLeftRight6095907004/1115060708090100p = P(kicker Left)goal %keeper Leftkeeper Rightlower envelopev ≈ 79.1
  1. Row minima 60, 70 → maximin 70; column maxima 90, 95 → minimax 90. 70 < 90: no saddle point.
  2. Kicker goes Left with prob. p: keeper Left ⇒ 60p + 90(1−p) = 90 − 30p; keeper Right ⇒ 95p + 70(1−p) = 70 + 25p.
  3. The keeper minimizes the scoring rate, so the kicker's guarantee is the lower envelope of the two lines.
  4. Peak of the lower envelope: 90 − 30p = 70 + 25p ⇒ p* = 4/11 ≈ 0.36, v = 870/11 ≈ 79.1%.
  5. Keeper goes Left with prob. q: 60q + 95(1−q) = 90q + 70(1−q) ⇒ q* = 5/11; the keeper also guarantees 79.1% ⇒ max-min = min-max.

Illustrative values (real data show a similar pattern: Palacios-Huerta, 2003). Left/Right are from the kicker's view.

Minimax and Complexity

Security Games: Randomized Patrols

  • Many targets, limited resources: airports, ports, forest patrols. A defender covering k of n targets has C(n, k) pure strategies; for n = 20, k = 5 that is 15,504.
  • The attacker can observe the patrol pattern: a fixed schedule can be exploited, so the defender randomizes. The ARMOR system at Los Angeles International Airport (2007) scheduled patrols with this model (Tambe, 2011).
One guard, two targets

Airport H is worth 4, port L is worth 2. The guard covers H with probability p and L with 1 − p; attacking an uncovered target wins its value, attacking a covered one wins 0.

Equate: 4(1 − p) = 2p ⇒ p = 2/3; the attacker gets 4/3. If the guard always covered H, the attacker would take 2 from L.

01/32/3101234pattacker's payoff● p* = 2/3, payoff 4/3attack H: 4(1 − p)attack L: 2palways H: 2

Minimax and Complexity

Von Neumann's Minimax Theorem (1928)

In every finite two-player zero-sum game   maxx miny xTAy = miny maxx xTAy = v
Equilibria = minimax strategies(x*, y*) is a Nash equilibrium ⇔ x* is a maximin and y* a minimax strategy. Every equilibrium has the same value v.
InterchangeabilityIf (x, y) and (x′, y′) are equilibria, so is (x, y′): no coordination problem.
Linear programmingThe row player seeks the mix x and the largest v such that x earns at least v against every column; this is a linear program that computers solve quickly.
Link to learningIf players learn by increasing the weight of strategies that did well, their average strategies approach minimax.

Nash's theorem can be seen as a generalization of the minimax theorem to non-zero-sum n-player games; but in general games the value is not unique and LP does not solve them.

Minimax and Complexity

Code Trace: maximin(A) — 2 × n Zero-Sum

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)

A = [[4, 1, 3], [0, 3, 2]]: if the row player plays row 1 with probability p, each column gives a line. Pure maximin is 1, minimax is 3: no saddle point.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)0·1·

n = 3 columns. The candidates start with the endpoints: p = 0 and p = 1.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)0·1·1/2·

Columns X and Y: d = 6 ≠ 0 ⇒ the lines cross at p = 1/2; add it to the candidates.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)0·1·1/2·2/3·

Columns X and Z: d = 3 ≠ 0 ⇒ the lines cross at p = 2/3; add it to the candidates.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)0·1·1/2·2/3·1/3·

Columns Y and Z: d = −3 ≠ 0 ⇒ the lines cross at p = 1/3; add it to the candidates.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)0·1·1/2·2/3·1/3·

g(p) = minj [p·A[0][j] + (1 − p)·A[1][j]]: the column player answers each p with the worst column (the lower envelope).

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)001·1/2·2/3·1/3·

max tries a candidate: g(0) = 0. Best so far.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)00111/2·2/3·1/3·

max tries a candidate: g(1) = 1. Best so far.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)00111/222/3·1/3·

max tries a candidate: g(1/2) = 2. Best so far.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)00111/222/35/31/3·

max tries a candidate: g(2/3) = 5/3.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pcandidate pg(p)00111/222/35/31/34/3

max tries a candidate: g(1/3) = 4/3.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pv = 2candidate pg(p)00111/222/35/31/34/3

The largest guarantee is at p = 1/2: v = 2. Column Z's line stays above the envelope, so the column player never plays Z.

01/31/22/31024pX: 4pY: 3 − 2pZ: 2 + pv = 2candidate pg(p)00111/222/35/31/34/3

Returns (1/2, 2). The minimax side: if the column player plays X with 1/3 and Y with 2/3, the row player gets at most 2 ⇒ max-min = min-max = 2.

from fractions import Fraction as F
def maximin(A):                   # A: 2 x n
    n, cand = len(A[0]), [F(0), F(1)]
    for j in range(n):
        for k in range(j + 1, n):
            d = A[0][j] - A[1][j] - A[0][k] + A[1][k]
            if d != 0:
                cand.append(F(A[1][k] - A[1][j], d))
    g = lambda p: min(p*a+(1-p)*b for a, b in zip(*A))
    p = max((c for c in cand if 0 <= c <= 1), key=g)
    return p, g(p)

Minimax and Complexity

How Hard Is It to Compute an Equilibrium?

ProblemComplexityMethod / source
Pure Nash equilibrium (matrix)PolynomialBest-response marking, O(mn)
Two-player zero-sum, mixedPolynomialLinear programming (minimax)
Two-player general game, mixedPPAD-completeChen–Deng (2006)
3+ players, exactFIXP-completeEtessami–Yannakakis (2007)
3+ players, ε-approximate (small ε)PPAD-completeDaskalakis–Goldberg–Papadimitriou (2006)
Max-welfare equilibrium / unique?NP-hardGilboa–Zemel (1989); Conitzer–Sandholm (2008)
  • Support enumeration and Lemke–Howson work in practice but are exponential in the worst case.
  • With three players an equilibrium can be irrational, so we seek ε-approximate ones (deviating gains at most ε). Two-player games with rational payoffs have rational equilibria.
  • Further reading: for fixed ε (e.g. 0.1), two-player games admit a quasi-polynomial algorithm (Lipton–Markakis–Mehta, 2003); multiplayer games stay PPAD-hard (Rubinstein, 2015).

Minimax and Complexity

PPAD: Finding a Solution Known to Exist

  • A Nash equilibrium always exists → “is there an equilibrium?” is trivial; NP-completeness does not fit problems whose solution is guaranteed.
  • PPAD (Papadimitriou, 1994): the class of problems whose solution is known to exist, but finding it may require following a very long path step by step.
known sourcesought end = equilibriumother paths
  • Analogy: walk from the start of a path looking for its end; the end surely exists, but the path may be exponentially long.
  • Finding a Nash equilibrium is in this class. Consequence: in general games equilibria may not be efficiently computable → do players actually “find” equilibria?

Minimax and Complexity

In Practice: Approximate Equilibria and Software

  • ε-Nash equilibrium: with payoffs scaled to [0, 1], no player can gain more than ε by deviating unilaterally. Real-time systems (ad auctions, markets) often settle for this.
  • An exact equilibrium may be hard to find, but a profile where “almost nobody wants to deviate” is often found quickly; even simple profiles mixing a few strategies approximate well.
  • Further reading: a ½-approximate equilibrium built from two best responses and a 50–50 mix (Daskalakis–Mehta–Papadimitriou, 2006); small-support ε-equilibria (Lipton–Markakis–Mehta, 2003).
import nashpy as nash
A = [[2, 2], [0, 4]]    # worker: Work / Shirk
B = [[1, 2], [-1, -4]]  # manager: Inspect / Don't
game = nash.Game(A, B)
for x, y in game.support_enumeration():
    print(x, y)  # [0.75 0.25] [0.5 0.5]

The inspection game in Nashpy: support enumeration returns p* = 3/4, q* = 1/2. Gambit does the same (support enumeration, Lemke–Howson, LP) with a GUI.

There are other equilibrium concepts too, e.g. correlated equilibrium and, in sequential games, subgame perfect equilibrium.

Summary

Discussion: Equilibrium, Fairness and Rationality

Is an equilibrium fair?Not necessarily. It is only a set of mutual best responses: payoffs are unequal in Battle of the Sexes, and the Prisoner's Dilemma has an outcome better for everyone. Fairness ≠ equilibrium.
Are people rational?Real decisions are shaped by missing information, time pressure, emotions and social norms (bounded rationality, Herbert Simon). The model is a clear benchmark for studying strategic interdependence.
Ethical limitsDeception, manipulation, unfair advantage… “Anything to win?” Rules and institutions set the limits of strategic behavior — which is the subject of mechanism design.

Summary

Summary

ConceptWhat it saysHow to find it
Nash equilibriumNobody wants to deviate aloneMutual best responses
Eliminating dominated strategiesA rational player never plays a dominated strategyIteratively delete strictly dominated ones
Mixed equilibriumStrategies in the support pay equallyIndifference equations
Existence (Nash 1950)Every finite game has a mixed equilibriumKakutani / Brouwer fixed point
Multiple equilibriaTheory may not pick a single outcomePayoff / risk dominance, focal points
Correlated equilibriumFollowing the mediator is in everyone's interestObedience constraints + linear programming
MinimaxZero-sum: max-min = min-max = vSaddle point or LP
ComplexityExact (2 players) and approximate (many players) equilibria are PPAD-completeLemke–Howson, support enumeration

Summary

Checklist: Solving a Game Step by Step

Are moves observed?Yes: sequential gamebackward inductionNo: matrix1. Eliminatestrictly dominated2. Underlinepure equilibria3. Indifferencemixed equilibrium4. Verifydeviation test5. Count: usually an odd number
Read firstWho is row, who is column? Whose payoff comes first in a cell? Are the numbers payoffs or costs (years, minutes)?
Then writeWrite each equilibrium as a profile: (Middle, Left) = (3, 5); for a mixed one, give both players' probabilities and expected payoffs.

Summary

Test Yourself (1/2)

Game ALeftMiddleRightTopBottom(3,1)(1,2)(2,0)(0,2)(2,3)(1,1)
Game BLeftRight
Top(2, 1)(0, 0)
Bottom(0, 0)(1, 2)
  1. Does the column player have a dominant strategy in Game A?
  2. Find Game A's pure Nash equilibria by underlining.
  3. How many equilibria does Game B have? In the mixed one, what are p = P(Top) and q = P(Left)?
  4. In Game B, are the mixed-equilibrium payoffs better than the pure ones?
Show answers
  1. Yes: Center. In row Top 2 > 1, 2 > 0; in row Bottom 3 > 2, 3 > 1.
  2. Row player's bests: Left → Top, Center → Bottom, Right → Top; column player's bests: Center, Center. Unique equilibrium (Bottom, Center) = (2, 3).
  3. Three: (Top, Left), (Bottom, Right) and a mixed one. Row indifferent: 2q = 1 − q ⇒ q = 1/3; column indifferent: p = 2(1 − p) ⇒ p = 2/3.
  4. No: in the mixed one both get 2/3; in the pure ones at least 1.

Summary

Test Yourself (2/2)

True or false?

  1. A strictly dominated strategy can be played in a Nash equilibrium.
  2. In a mixed equilibrium, the row player's probability is computed from their own payoffs.
  3. A subgame perfect equilibrium (found by backward induction on the game tree) is always a Nash equilibrium.
  4. In a three-player game, joint deviations by two players are also checked.

Sequential game

  1. Player 1 chooses Left or Right. Right → the game ends with (1, 1). Left → Player 2 chooses a → (3, 0) or b → (0, 2). Find the equilibrium and outcome by backward induction.
  2. If Player 2 promises “I'll choose a if you go Left”, should Player 1 trust it?
Show answers
  1. False: a strictly dominated strategy is a best response to nothing.
  2. False: it must make the opponent indifferent, so it comes from the opponent's payoffs.
  3. True; the converse fails (non-credible threats).
  4. False: Nash equilibrium only checks unilateral deviations.
  5. After Left, Player 2 picks b (2 > 0) → value (0, 2). Player 1: Right 1 > 0. Equilibrium (Right; b), outcome (1, 1).
  6. No: at that point a gives Player 2 0 and b gives 2; the promise is not credible.

Summary

Sample Problems

Game 1LMR
U(2, 1)(1, 2)(0, 0)
D(3, 0)(0, 1)(1, 3)
Game 2LR
U(4, 1)(0, 3)
D(1, 2)(2, 0)
  1. Game 1: pure Nash equilibria by marking best responses.
  2. Game 2: the mixed equilibrium via indifference.
  3. Zero-sum A = [[3, −1, 2], [1, 0, 4], [−2, −3, 5]]: saddle point? Value?
  4. “Every Nash equilibrium is Pareto optimal.” True?
  5. Why is existence stated for mixed strategies?
  6. Cournot, P = 15 − Y, zero cost: equilibrium quantities? In an equal-split cartel, each firm's output, and why is it not an equilibrium?
Show answers
  1. Best row in each column: D, U, D; best column in each row: M, R. Equilibria: (U, M) and (D, R).
  2. q = P(L): 4q = q + 2(1−q) ⇒ q = 2/5; p = P(U): p + 2(1−p) = 3p ⇒ p = 1/2.
  3. Maximin = max(−1, 0, −3) = 0 = minimax = min(3, 0, 5): saddle point (row 2, column 2), v = 0.
  4. False: in the Prisoner's Dilemma (B, B) is the equilibrium, yet (S, S) is better for both.
  5. A pure equilibrium may not exist (Matching Pennies). Since probabilities vary continuously, the fixed-point theorem guarantees an equilibrium among mixed profiles.
  6. BR(y) = (15 − y)/2 ⇒ y* = 5, profit 25; the cartel (3.75 each) is not an equilibrium: BR(3.75) = 5.625.

Summary

Exercises (1/2)

Game 1LeftRight
Top(4, 3)(1, 2)
Bottom(3, 1)(2, 4)
Game 2012
0(2, 5)(1, 3)(3, 1)
1(5, 2)(3, 3)(1, 2)
2(0, 0)(5, 5)(0, 0)
  1. Does Game 1 have a dominant strategy? Find its pure Nash equilibria by marking best responses.
  2. In Game 2 (row i, column j), iteratively delete strictly dominated strategies. What remains? Compare with the pure Nash equilibrium.
  3. In RPS your opponent plays Rock and Paper 50–50 (never Scissors). What is your best response and expected payoff?
Show answers
  1. No dominant strategy (row: Top vs Left, Bottom vs Right; column: Left vs Top, Right vs Bottom). Two pure equilibria: (Top, Left) = (4, 3) and (Bottom, Right) = (2, 4).
  2. Deletion order: column 2 (by 1) → row 0 (by 1) → column 0 (by 1) → row 1 (by 2). The surviving cell (2, 1) = (5, 5) is also the unique pure Nash equilibrium.
  3. E[Rock] = −1/2, E[Paper] = +1/2, E[Scissors] = 0 → always Paper, expected payoff +1/2.

Summary

Exercises (2/2)

  1. Find the maximin and minimax of the zero-sum matrices A = [[3, 1, 4], [2, 0, 1], [5, 2, 3]] and B = [[0, 2], [3, 1]]. Which has a saddle point?
  2. (X, X) = (4, 1), (Y, Y) = (1, 4), other cells (0, 0). In the mixed equilibrium, what are p* = P(row X), q* = P(column X) and the probability of failing to meet?
  3. In Chicken, if a crash cost −20 instead of −10, what would the swerve and crash probabilities be in the mixed equilibrium?
  4. In Stag Hunt (stag 4, hare 3), if a hare paid 2 instead of 3, what would the stag probability be in the mixed equilibrium? Does trust get easier?
Show answers
  1. A: row minima 1, 0, 2 → maximin 2; column maxima 5, 2, 4 → minimax 2: saddle point (row 3, column 2), value 2. B: maximin 1 < minimax 2, no saddle point; the row player mixes 1/2–1/2 for value 3/2.
  2. Row indifferent: 4q = 1 − q ⇒ q* = 1/5; column indifferent: p = 4(1 − p) ⇒ p* = 4/5. Failing to meet (4/5)(4/5) + (1/5)(1/5) = 17/25 = 68%.
  3. With crash cost k, the swerve probability is (k − 1)/k = 19/20; crash (1/20)² = 1/400.
  4. 4q = 2 ⇒ q* = 1/2 (was 3/4): a 50% chance that the partner shows up is enough; trust gets easier.

Summary

References

  • J. F. Nash, “Equilibrium points in n-person games”, PNAS 36(1), 1950; “Non-cooperative games”, Annals of Mathematics 54(2), 1951.
  • J. von Neumann, “Zur Theorie der Gesellschaftsspiele”, Mathematische Annalen 100, 1928.
  • M. J. Osborne, An Introduction to Game Theory, Oxford University Press, 2004 (Chapters 2–4).
  • N. Nisan, T. Roughgarden, É. Tardos, V. Vazirani (eds.), Algorithmic Game Theory, Cambridge, 2007 (Chapters 1–3).
  • T. Roughgarden, Twenty Lectures on Algorithmic Game Theory, Cambridge, 2016 (Lectures 13, 18 and 20).
  • C. Daskalakis, P. Goldberg, C. Papadimitriou, “The complexity of computing a Nash equilibrium”, SIAM J. Computing 39(1), 2009.
  • K. Etessami, M. Yannakakis, “On the complexity of Nash equilibria and other fixed points”, SIAM J. Computing 39(6), 2010 · R. J. Lipton, E. Markakis, A. Mehta, “Playing large games using simple strategies”, ACM EC, 2003 · A. Rubinstein, “Inapproximability of Nash equilibrium”, STOC, 2015.

Summary

References (2/2)

  • J. C. Harsanyi, R. Selten, A General Theory of Equilibrium Selection in Games, MIT Press, 1988.
  • A. R. Karlin, Y. Peres, Game Theory, Alive, AMS, 2017 (zero-sum and general games, Nash's theorem).
  • D. Easley, J. Kleinberg, Networks, Crowds, and Markets, Cambridge, 2010 (Chapter 6: Games).
  • M. Tambe, Security and Game Theory, Cambridge University Press, 2011 (security games, ARMOR).
  • D. Sally, “Conversation and cooperation in social dilemmas: a meta-analysis of experiments from 1958 to 1992”, Rationality and Society 7(1), 1995 · R. Wilson, “Computing equilibria of n-person games”, SIAM J. Applied Mathematics 21, 1971.
  • Course pages: Roughgarden CS364A · Shah CSC304 · Software: Gambit, Nashpy · Video: Game Theory 101

The End

Chapter 3: Nash Equilibrium
Previous: Chapter 2: Games and Representations · Next: Chapter 4: Mechanism Design
1 / 1 Sercan KÜLCÜ. All rights reserved.