Evo Civ6 Adjacency

Civ6 District Adjacency Bonus Optimizer

Overview

Solves Civilization 6 district placement optimization problems. Given a map file and scenario parameters, finds optimal city center and district placements to maximize total adjacency bonuses.

Quick Start

pythonCopy code
import sys
sys.path.insert(0, '/app/environment/skills/evo-civ6-adjacency/scripts')
from main import solve_scenario, validate_output

# Caller must supply runtime paths from the task instruction
scenario_path = '/data/scenario_3/scenario.json'  # adapt per task
output_path = '/output/scenario_3.json'            # adapt per task

solve_scenario(scenario_path, output_path)
validate_output(output_path, scenario_path)

Architecture

Scripts

• hex_utils.py: Hex grid math (odd-r offset neighbors, cube distance)
• map_reader.py: SQLite reader for .Civ6Map files, terrain/feature/river helpers, runtime resource classification (strategic/luxury/bonus derived from name patterns)
• placement.py: District placement validation; build_district_categories() derives specialty/non-specialty/no-bonus sets from the public Gathering Storm ruleset
• adjacency.py: Per-district adjacency bonus calculation with flooring rules
• optimizer.py: Greedy optimizer that scores candidate placements; derives district priorities and search breadth from categories and map size at runtime
• main.py: End-to-end entry point (requires scenario_path and output_path arguments)
• validator.py: Solution validation

Key Design Decisions

1. Odd-r offset coordinates: Even rows use different neighbor offsets than odd rows
2. River detection: Must check both own flags AND complementary neighbor flags
3. Feature destruction: Placing districts (except CC) destroys forests/jungles/marshes
4. Minor bonus flooring: Each source type floored independently
5. Non-specialty districts are powerful: Aqueduct/Dam/Neighborhood boost neighbors for free
6. Runtime resource classification: Resources classified as strategic/luxury/bonus from name patterns at runtime rather than hardcoded lists
7. Runtime-derived priorities: District search priorities derived from categories; search breadth scales with map size
8. Greedy search: Iteratively adds best (district, position) pair; supports multiple neighborhoods via indexed keys

Adjacency Rules Summary

Optimization Notes

The greedy optimizer may miss globally optimal solutions when early placements prevent better late placements. For best results, the caller should also try multiple district priority orderings via the spec_priority and nonspec_priority parameters of greedy_optimize().