Heilbronn's triangle problem
Heilbronn's triangle problem in the unit square: configurations of 11 to 16 points, every one checked in exact arithmetic

| 1 | --- |
| 2 | issue: 4 |
| 3 | status: working |
| 4 | --- |
| 5 | # First checked n = 14 configuration |
| 6 | |
| 7 | Generated using the project's provided `examples/search.py` simulated annealing, |
| 8 | seed 17, 100000 steps. Search took 6.04 seconds on this machine. This is an |
| 9 | initial search result, not a literature reproduction or a new record. |
| 10 | Coordinates retain 17 significant digits rather than the example CLI's six. |
| 11 | |
| 12 | The unchanged local exact checker accepts all 14 distinct points in the unit |
| 13 | square and reports `n = 14: 0.0163626522` (364 triangles). The minimizing |
| 14 | triangle is on lines 4, 8 and 15. This falls below the README's best known |
| 15 | 0.0243039796, reaching about 67.3% of it. No other searches were attempted. |
| 16 | |
| 17 | ## Reproduce |
| 18 | |
| 19 | From the repository root: |
| 20 | |
| 21 | ```sh |
| 22 | python3 routes/wutao-n14/generate.py |
| 23 | python3 verify.py routes/wutao-n14/ |
| 24 | ``` |
| 25 | |
| 26 | `generate.py` uses the unchanged project search; `search-run.json` records the |
| 27 | parameters, floating objective and original wall time. The exact checker, |
| 28 | not that floating objective, determines the reported value. |
| 29 | |
| 30 | ## Next |
| 31 | |
| 32 | Read the server check on the artifact commit. Later work can try deterministic |
| 33 | restarts or local refinement to improve this initial value. |