# 전라좌수군 견리사의 군령 ## Paper Generation System for PhD Thesis > 견리사의 (見利思義): 이익을 보면 의로움을 생각하라 > 필사즉생 (必死卽生): 죽으려 하면 살 것이요, 살려 하면 죽을 것이다 --- ## Directory Structure ``` paper_generation/ ├── README.md # This file ├── __init__.py # Module initialization ├── generate_all_chapters.py # Main entry point ├── paper_magnet.py # Paper search system ├── chap1_introduction.py # 起 정운 ├── chap2_theory.py # 承 권준 ├── chap3_empirics.py # 轉 김완 + 나대용 ├── chap4_discussion.py # 結 어영담 │ ├── P1_vagueness/ # P1: U-Shape Analysis │ ├── theory.md # 권준의 이론 정리 │ ├── empirics.py # OLS/Logit 분석 코드 │ ├── figures.py # Figure 생성 코드 │ └── output/ # 생성된 Figure (.png) │ ├── P2_trap/ # P2: Competency Trap │ ├── theory.md # 권준의 이론 정리 │ ├── simulation.py # Bayesian 시뮬레이션 코드 │ ├── figures.py # Figure 생성 코드 │ └── output/ # 생성된 Figure (.png) │ ├── P3_newsvendor/ # P3: Newsvendor Optimization │ ├── theory.md # 권준의 이론 정리 │ ├── model.py # Newsvendor 모델 코드 │ ├── figures.py # Figure 생성 코드 │ └── output/ # 생성된 Figure (.png) │ └── output/ # Generated markdown files ├── chap1_introduction.md ├── chap2_theory.md ├── chap3_empirics.md └── chap4_discussion.md ``` --- ## Quick Start ```bash # Generate all chapters python generate_all_chapters.py # Check status python generate_all_chapters.py --status # Generate specific chapter python generate_all_chapters.py --chapter 1 # View expectation management python generate_all_chapters.py --expect # Search for relevant papers python paper_magnet.py --paper P2 ``` --- ## The Fleet (전라좌수군) | Chapter | 한자 | Commander | Virtue | Bayesian Role | |---------|------|-----------|--------|---------------| | **1. Introduction** | 起 | 정운 🐢 | 利 (Speed) | Prior π(θ) | | **2. Theory** | 承 | 권준 🐅 | 思 (Structure) | Likelihood π(y\|θ) | | **3. Empirics** | 轉 | 김완 🐙 + 나대용 🐅 | 義 (Criticism) | Calibration | | **4. Discussion** | 結 | 어영담 👾 | 見 (Observation) | Generator | --- ## Three Papers (P1/P2/P3) Each chapter generates content for three papers simultaneously: | Paper | Emoji | Title | Domain Focus | |-------|-------|-------|--------------| | **P1** | ✌️ | U-Shape: When Vagueness Pays | Technology | | **P2** | 🦾 | Competency Trap: When Success Kills Options | Organization | | **P3** | 🤹 | Execution Gap: Optimal Number of Options | Competition | --- ## Paper Magnet (논문 자석) Search for theoretically resonant papers from `/Users/hyunjimoon/tolzul/Space/Library/1논문용/`: ```bash # Top resonant papers python paper_magnet.py # Papers for specific theory python paper_magnet.py --paper P1 # U-Shape python paper_magnet.py --paper P2 # Competency Trap python paper_magnet.py --paper P3 # Execution Gap # Scott-Charlie tension papers python paper_magnet.py --tension # Keyword search python paper_magnet.py --keyword "real option" ``` --- ## Workflow (견리사의 순환) ``` 利 → 思 → 義 → 見 → 利 ↓ 정운(Draft) → 권준(Structure) → 김완(Verify) → 어영담(Record) ↑ ↓ └──────────── Generator (다음 Prior) ←─────────┘ ``` --- ## Output Format Each generated markdown file contains: - P1 section (U-Shape) - P2 section (Competency Trap) - P3 section (Execution Gap) - Cross-Synthesis section --- ## Legacy Code Old code has been archived to: `/Users/hyunjimoon/tolzul/Front/On/love(cs)/strategic_ambiguity/empirics/docs_archive/legacy_paper_generation/` --- ## P1/P2/P3 Empirics & Figures (나대용 임무) ### Quick Start: Generate All Figures ```bash # P1: U-Shape Analysis cd P1_vagueness python empirics.py # Run OLS/Logit analysis python figures.py # Generate figures # P2: Competency Trap cd P2_trap python simulation.py # Run Bayesian simulation python figures.py # Generate figures # P3: Newsvendor cd P3_newsvendor python model.py # Run newsvendor optimization python figures.py # Generate figures ``` ### P1: U-Shape - When Vagueness Pays **Variables:** - `vagueness_score (V)`: 0.6 * S_cat + 0.4 * S_concdef - `survival`: Binary (survived 3+ years) - `funding`: Log(total funding amount) - `exercisability`: Hardware/Software classification **Models:** - H1: OLS regression `log(Funding) = β₀ + β₁V + β₂V² + controls` - H2: Logit with interaction `Pr(Survival) = Λ(β₀ + β₁V + β₂H + β₃(V×H) + ...)` **Output Figures:** - `P1_u_shape_survival.png`: U-shape survival vs vagueness - `P1_hw_sw_comparison.png`: Hardware vs Software interaction - `P1_coefficient_table.png`: Coefficient table visualization ### P2: Competency Trap - When Success Kills Options **Variables:** - `prior (μ₀, σ₀)`: Initial beliefs about current path - `evidence_strength`: Strength of disconfirming signal - `believer_ratio`: Proportion of like-minded investors - `switching_threshold (τ)`: Evidence threshold for pivot **Model:** Bayesian update simulation ``` μ_post = (σ_e² * μ₀ + σ₀² * y) / (σ₀² + σ_e²) τ = f(σ₀, believer_ratio) ``` **Output Figures:** - `P2_belief_lockin_diagram.png`: μ,σ → threshold heatmap - `P2_pivot_threshold_curve.png`: Pivot threshold vs σ curve - `P2_case_comparison.png`: Waymo vs Comma.ai case study ### P3: Newsvendor - Optimal Number of Options **Variables:** - `D`: Demand ~ Poisson(λ) or Normal(μ,σ²) - `C`: Commitment cost - `F`: Flexibility cost - `CR`: Commitment ratio = C / (C + F) **Model:** Newsvendor optimization ``` k* = μ_D + σ_D * Φ⁻¹(CR) ``` **Output Figures:** - `P3_cr_kstar_curve.png`: CR vs k* policy curve - `P3_sensitivity_heatmap.png`: Sensitivity to σ(D) - `P3_industry_calibration.png`: Industry-specific CR calibration - `P3_unified_framework.png`: P1/P2/P3 unified on CR-k* plane --- ## Requirements ```bash pip install numpy pandas scipy statsmodels matplotlib ``` ## Reproducibility - All scripts use `SEED = 42` for reproducibility - Dummy data generated internally (no external data required) - Relative paths used for output --- *통제사: 이순신 문현지 (Moon)* *공병대장: 나대용 (Builder)*