> ## Documentation Index
> Fetch the complete documentation index at: https://docs.strategist.gg/llms.txt
> Use this file to discover all available pages before exploring further.

# Mathematical Foundations

## Foundational Mathematical Framework

Based on **Rasmusen's "Games and Information"** and advanced game theory literature, THE STRATEGIST implements rigorous mathematical models for personal optimization.

### Core Nash Equilibrium Definition for 4-Agent Architecture

**For Strategist's 4-agent system:**

**∀i ∈ {Optimizer, Protector, Explorer, Connector}, ∀s'ᵢ ∈ Sᵢ:**
**uᵢ(sᵢ*, s*₋ᵢ) ≥ uᵢ(s'ᵢ, s\*₋ᵢ)\*\*

Where no internal agent can improve its utility by unilaterally changing strategy, given other agents' optimal strategies.

**Mathematical Interpretation:**

```
Players: i = {1, 2, 3, 4} = {Optimizer, Protector, Explorer, Connector}
Strategies: sᵢ ∈ Sᵢ (agent i's strategy set)
Utility: uᵢ: S₁ × S₂ × S₃ × S₄ → ℝ (agent i's payoff function)
Equilibrium: s* = (s₁*, s₂*, s₃*, s₄*) where no unilateral deviation improves payoff
```

## Advanced Equilibrium Concepts for Personal Optimization

### Subgame Perfect Equilibrium

**Critical for temporal decision-making.** From **Rasmusen's analysis**, ensures credible long-term strategies by eliminating non-credible threats.

**Mathematical Definition:**
A strategy profile is subgame perfect if it induces Nash equilibrium behavior in every subgame.

**Personal Application:**
Eliminates strategies like "I'll exercise tomorrow if I don't today" because the "tomorrow self" has no incentive to follow through.

**Backward Induction Algorithm:**

```
For time T, T-1, T-2, ..., 1:
  At time t, choose action that maximizes:
  uᵢ(aₜ) + δ · Vᵢ(t+1 | aₜ)
  
Where δ = discount factor, Vᵢ = continuation value
```

### Correlated Equilibrium

**From Aumann's theory** - enables coordination through shared information sources.

**Mathematical Framework:**

```
Correlation Device: μ ∈ Δ(A₁ × ... × A₄)
Conditional Strategies: σᵢ(aᵢ | recommendation)
Incentive Compatibility: No agent wants to deviate from recommendation
```

**Personal Application:**
External data sources (weather, stress levels, social commitments) coordinate agent strategies more effectively than independent decision-making.

**Example:**

```json theme={null}
{
  "correlation_device": "weather_forecast",
  "agent_strategies": {
    "rain_day": {
      "optimizer": "indoor_productivity_focus",
      "protector": "cozy_safety_prioritization",
      "explorer": "creative_indoor_projects",
      "connector": "intimate_social_connections"
    },
    "sunny_day": {
      "optimizer": "outdoor_efficiency_tasks", 
      "protector": "vitamin_d_health_optimization",
      "explorer": "adventure_seeking_behaviors",
      "connector": "large_group_social_activities"
    }
  }
}
```

### Evolutionary Stable Strategies (ESS)

**From Maynard Smith's biological framework**, adapted for personal habit formation.

**Strategy s* is ESS if:*\*

* **π(s*, s*) > π(s', s\*)\*\* OR
* **π(s*, s*) = π(s', s\*) AND π(s\*, s') > π(s', s')\*\*

**Personal Interpretation:**
A habit configuration is evolutionarily stable if:

1. It performs better against itself than any alternative habit
2. If it performs equally well, it performs better against the alternative than the alternative performs against itself

**Replicator Dynamics:**

```
ẋᵢ = xᵢ[f(eᵢ,x) - f(x,x)]

Where:
xᵢ = frequency of behavior i in your behavioral repertoire
f(eᵢ,x) = fitness of behavior i against current behavioral mix
f(x,x) = average fitness across all behaviors
```

## Advanced Mathematical Structures

### Utility Function Decomposition

**Agent-Specific Utility Functions:**

**Optimizer:** `u₁(s) = Σ(output_value) / Σ(resource_cost)`

**Protector:** `u₂(s) = baseline_security - Σ(risk_exposure × impact)`

**Explorer:** `u₃(s) = Σ(novelty_value × growth_potential) - stagnation_penalty`

**Connector:** `u₄(s) = Σ(relationship_quality × interaction_frequency) - social_isolation_cost`

### Strategic Complementarities and Substitutabilities

**Supermodular Games:**
When agents' strategies are strategic complements:

```
∂²uᵢ(sᵢ, s₋ᵢ) / ∂sᵢ∂sⱼ > 0 for j ≠ i
```

**Example:** Higher Optimizer effort makes Explorer effort more valuable (learning accelerates with better systems)

**Submodular Games:**
When agents' strategies are strategic substitutes:

```
∂²uᵢ(sᵢ, s₋ᵢ) / ∂sᵢ∂sⱼ < 0 for j ≠ i
```

**Example:** Higher Protector effort makes Explorer effort less necessary (safety reduces need for exploration)

## Existence and Uniqueness Theorems

### Nash's Existence Theorem

**For finite games:** Every finite game has at least one Nash equilibrium (possibly in mixed strategies).

**For continuous games (Strategist application):**

* **Strategy sets Sᵢ are non-empty, convex, compact subsets of Euclidean space**
* **Utility functions uᵢ are continuous and quasi-concave in sᵢ**
* **Therefore, equilibrium exists**

### Uniqueness Conditions

**Contraction Mapping:** If the best response function is a contraction, equilibrium is unique:

```
||BR(s) - BR(s')|| ≤ α||s - s'|| for α < 1
```

**Personal Application:** Stable personalities tend toward unique equilibrium configurations.

## Bayesian Games and Incomplete Information

### Type Space Construction (Harsanyi Transformation)

**When agents don't know each other's true preferences:**

```
Type Space: Θᵢ = possible preference types for agent i
Belief Function: μᵢ(θ₋ᵢ | θᵢ) = agent i's beliefs about others' types
Strategy: sᵢ(θᵢ) = action as function of type
```

**Personal Application:**
Early in personal development, agents don't know true preferences. Bayesian updating reveals personality through observed choices.

### Perfect Bayesian Equilibrium

**Requirements:**

1. **Sequential Rationality:** Agents optimize at every decision point
2. **Belief Consistency:** Beliefs derived from strategies using Bayes' rule where possible

**Personal Application:**
As you learn about yourself, agent coordination improves through better information about internal preferences.

## Mixed Strategies and Randomization

### When to Use Mixed Strategies

**Indifference Condition:** In equilibrium, agents must be indifferent between strategies in their support.

**Example: Exercise Timing**

```
Morning Gym Utility = Evening Gym Utility
p · (high_energy_workout) + (1-p) · (disrupted_schedule) = 
q · (lower_energy_workout) + (1-q) · (consistent_routine)
```

**Solve for optimal mixing probabilities p and q**

### Mixed Strategy Equilibrium Computation

**For 2×2 subgames between agents:**

```
Agent 1 mixing: u₁(Up, Left) = u₁(Down, Left) and u₁(Up, Right) = u₁(Down, Right)
Agent 2 mixing: u₂(Up, Left) = u₂(Up, Right) and u₂(Down, Left) = u₂(Down, Right)
```

## Mechanism Design for Self-Control

### VCG Mechanisms for Internal Coordination

**Truthful Reporting:** Agents honestly report internal states
**Efficient Allocation:** Resources distributed optimally across life domains
**Budget Balance:** Total energy/attention allocation equals available resources

**VCG Payment Rule:**

```
Payment by agent i = Σⱼ≠ᵢ vⱼ(outcome_without_i) - Σⱼ≠ᵢ vⱼ(actual_outcome)
```

**Personal Application:** Agents "pay" attention costs based on social welfare they create/destroy.

### Myerson's Optimal Mechanism

**Virtual Valuations:** Account for agent type distributions

```
ψᵢ(vᵢ) = vᵢ - (1-Fᵢ(vᵢ))/fᵢ(vᵢ)
```

**Revenue Maximization:** Allocate resources to agents with highest virtual valuations, not just highest reported values.

## Repeated Games and Temporal Consistency

### Folk Theorem Application

**Any individually rational payoff can be sustained if agents are sufficiently patient (δ → 1).**

**Trigger Strategies for Habits:**

* **Grim Trigger:** Permanent punishment after deviation (too harsh)
* **Tit-for-Tat:** Copy last period's action
* **Generous Tit-for-Tat:** Occasional forgiveness (optimal for personal development)

### Dynamic Programming for Life Optimization

**Bellman Equation:**

```
V(state) = max{immediate_utility + δ · E[V(next_state)]}
```

**Personal Application:**

```
V(health_state, energy_level, relationships) = 
max{u(daily_actions) + δ · E[V(tomorrow_state | daily_actions)]}
```

***

*Mathematical rigor ensures optimal strategic configurations rather than intuitive guesswork.*

**Next: [Computational Methods →](/game-mechanics/nash-computational-methods)**