---

name: experiment-testing-standards description: Enforces experiment testing standards: real implementation testing, no simplified approximations, exact formula matching. Use when creating experiments, testing algorithms, or validating mathematical implementations.

Experiment Testing Standards

Core Principle

Experiments must test the REAL implementation, never simplified approximations.

Mandatory Requirements

✅ REQUIRED

• Real Implementation Logic - Use exact same calculations, formulas, and logic as production code
• Full Feature Testing - Test complete implementations, not simplified versions
• Exact Formula Matching - Quantum state calculations, compatibility formulas must match production exactly
• Real Data - Use real Big Five OCEAN data converted to SPOTS dimensions

❌ FORBIDDEN

• Simplified Calculations - Never use simplified approximations
• Linear Approximations - Never use linear approximations of nonlinear operations
• Mock Implementations - Never use mock/stub implementations in experiments
• Partial Testing - Never test only part of feature with simplified logic

Example Patterns

❌ BAD: Simplified Approximation

Copy# Simplified distance decay (NOT the real implementation)
location_compatibility = exp(-decay_rate * distance_km)

✅ GOOD: Real Implementation

Copy# Real quantum state inner product (matches production code)
location_state = create_location_quantum_state(
    lat, lon, type, accessibility, vibe
)
compatibility = abs(inner_product(location_state_a, location_state_b)) ** 2

Real Data Requirement

All experiments MUST use real Big Five OCEAN data:

Copyfrom shared_data_model import load_and_convert_big_five_to_spots

# Load and convert Big Five OCEAN to SPOTS 12
spots_profiles = load_and_convert_big_five_to_spots(
    max_profiles=1000,
    data_source='auto',
    project_root=project_root
)

❌ FORBIDDEN:

• Synthetic data generation
• Hardcoded personality dimensions
• Simplified approximations

Checklist

• Experiment uses exact same formulas as production code
• Quantum state calculations match production implementation
• No simplified approximations used
• All mathematical operations match production code
• Uses real Big Five OCEAN data (not synthetic)
• Experiment logic documented with references to production code

Reference

• .cursorrules - Experiment Testing Standards section
• docs/plans/test_refactoring/TEST_WRITING_GUIDE.md