Best practices & limitations

Practical guidance for getting physically reliable amorphous structures out of AmorphGen — and an honest account of where the underlying methods break down.

Prefer NVT annealing over NPT melt-quench with foundation MLIPs

The single most important recommendation: with universal/foundation MLIPs, generate amorphous structures by random placement + low-temperature annealing (or the hybrid route), rather than a full crystal melt-quench under NPT.

Universal MLIPs (MACE-MP, CHGNet, SevenNet, …) are trained almost entirely on near-equilibrium crystalline data. A high-temperature liquid is out of that distribution on every axis at once — large forces, broken/over-coordinated bonds, high kinetic energy — so the model extrapolates and the melt is unreliable. The failure is worst under NPT, because the barostat acts on the stress tensor, which is even more sensitive to out-of-distribution force error than the energy: wrong stresses drive the cell to expand, collapsing the predicted density.

AmorphGen’s Random structure generation and Hybrid workflow routes are built around the reliable regime: they start from in-distribution disordered configurations and anneal at low temperature, in a fixed cell (NVT-like) when you keep the auto-estimated density. This avoids the unstable liquid entirely.

Tip

If you must run a melt-quench, prefer NVT (fixed cell) over NPT, keep the melt temperature as low as the chemistry allows, and treat any large mid-quench volume change as a red flag rather than a result.

Decision guide

Situation	Recommended route
New composition, no crystal needed	`--random-gen` + `--relax` (NVT-like, fixed cell)
Want a diverse amorphous ensemble cheaply	`--hybrid-ensemble` (random → anneal → quench)
You have a crystal and want classic MQ	Full pipeline, but prefer NVT stages; watch the density
MLIP keeps over-expanding under NPT	Switch to random-gen / hybrid, or fix the cell

Trust the relaxed density, set it when the estimate is shaky

The auto-estimated density is a starting cell heuristic (class-aware sphere packing on Shannon/Cordero/Goldschmidt radii). It is good for common oxides but approximate for unusual chemistries.

Warning

For elements missing from the radii tables, or compositions far from the tuned material classes, the auto density can be off. AmorphGen prints NOTE: Auto density is approximate for this composition when confidence is low — in that case pass --target-density (or a cell_length) explicitly, or relax with a cell filter and trust the relaxed density, not the estimate.

Dense rutile-type dioxides are the usual culprits: the generic metal_oxide packing factor under-predicts them, which is why rutile-type MO₂ oxides (TiO₂, SnO₂, RuO₂, IrO₂, OsO₂, …) are routed to a denser rutile_dioxide class. They are identified geometrically — an MO₂ whose 4+ cation radius is below the rutile/fluorite cutoff (~0.70 Å) — so fluorite dioxides (ZrO₂, HfO₂, CeO₂) correctly stay metal_oxide. See Validation for the validated density ranges.

Known limitations

Foundation-MLIP reliability — accuracy on amorphous/liquid configurations, and on elements sparsely represented in training data, is not guaranteed. High-temperature liquid sampling under NPT is the main failure mode (see above); validate against AIMD or experiment for any new chemistry (see Validation).
Density estimation is approximate — the auto density is a class-aware sphere-packing estimate for the starting cell, tuned on common material classes. It can be off for unusual chemistries; override with --target-density and trust the relaxed density.
Element coverage of the radii tables — minimum-separation and density estimation use Shannon/Cordero/Goldschmidt radii for a curated element set. Elements outside it fall back to approximate values, degrading the auto density (e.g. set an explicit --target-density, or add the element to amorphgen/utils/radii.py).
Single-point relaxation leaves voids — a 0 K relax of one random structure can stay porous; use the hybrid (anneal + quench) route for densification.
Ensembles, not single structures — one structure is not statistically representative of an amorphous phase. Generate an ensemble (e.g. -n 20) and average for any reported property.
Structure generation only — AmorphGen produces relaxed atomic structures, not electronic-structure or transport properties; those need a separate DFT/post-processing step on the generated models.
Classical potentials need parameters — the Lennard-Jones and Buckingham+Coulomb backends require user-supplied parameters; they are a fast starting point, not a substitute for a fitted potential.
Resume granularity — stage-level only. A run killed mid-stage re-runs that whole stage on --resume; the partial trajectory is preserved but not continued from.