Start tetrahedral (ice/LDL) → melt → does it collapse to HDL?
Temperature (model units — melts ≈1300 K):
mean q (blue) & coordination/6 (green) vs T. Ice/LDL = high q, coord≈4; HDL = low q, coord≈5, d5 pulls in.
Reading it. Cold, unmelted: is q high (~0.7–0.9) and coordination ≈4?
→ the isotropic Z=2 water holds the imposed tetrahedron. On heating: q falls, and if a
5th neighbour pulls in (d5 → ~3.2, coordination → ~5) that's the collapse to HDL.
Two well-separated states along the melt = the two structures, captured in one run. Caveats: coarse grid (h≈0.18 Å); frozen corners; interior-core only; Andersen model-time.