HPC Map

Distance map for German motorways to nearest eligible HPC chargers.

Architecture (current)

• Graph data: GraphHopper cache built from Germany OSM PBF
• Compute core: JVM offline engine (distance-core) running multi-source Dijkstra on GraphHopper graph cache
• Orchestration: Python pipeline (pipeline/run_pipeline.py)
• Outputs:
  • hpc_distance_segments*.geojson
  • hpc_distance*.mbtiles
  • hpc_sites*.geojson
  • run_metadata*.json
• Rendering: MapLibre distance vector tiles + clustered HPC GeoJSON

Prerequisites

• Python 3.9+
• Java 17+
• GraphHopper graphhopper-web-11.0.jar in tools/graphhopper/
• Germany PBF at data/raw/osm/germany-latest.osm.pbf
• tippecanoe for MBTiles generation

Build Graph Cache (one-time)

Mac/Linux:

docker compose up graphhopper --remove-orphans

Windows:

powershell -ExecutionPolicy Bypass -File .\scripts\start_graphhopper.ps1

This imports once and reuses tools/graphhopper/graph-cache/ on subsequent starts.

Run Pipeline

Mac/Linux:

python -m pipeline.run_pipeline --config config/default.yaml

Windows:

powershell -ExecutionPolicy Bypass -File .\scripts\run_pipeline.ps1

Docker:

docker compose run --rm pipeline

Start App Stack

docker compose up --build api tileserver frontend

• API: http://localhost:8000
• Tile server: http://localhost:8080
• Frontend: http://localhost:5173

Notes

• Legacy Overpass/sample-point/candidate-preselection route code is removed from pipeline flow.
• HPC MBTiles output is removed; HPC stations are served as GeoJSON for frontend clustering.
• distance-core currently runs via distance-core/run_distance_core.sh (Linux/macOS) or scripts/run_distance_core.ps1 (Windows).
• Gradle wrapper files are scaffolded; if you need wrapper-based build/test commands, generate distance-core/gradle/wrapper/gradle-wrapper.jar once via local Gradle (gradle wrapper --gradle-version 8.10.2).