Do the store's two DFT settings give the same energy for one cell?

EXPERIMENTS.md · lines 15791–15805

E230 — do the store's two DFT standards agree on one decoration? (pre-registered 2026-09-22 19:4x)

The QE store holds rows on two standards, each internally consistent: v2 (50/400 Ry; k matched to 0.212 Å⁻¹; references at 50/400, k 10³) — E172, E176, E177, E179, E180 — and v3 (60/720 Ry; 0.16 Å⁻¹; references k 13³) — everything since E183. Their elemental references differ by +8.6 (Ta) to +42.2 (Nb) meV/atom in absolute energy, which cancels only if the alloy's cutoff/k error matches its references'. The one warning sign: near-random Mo₅₀Ta₅₀ at 54 atoms gives −89.8 on v2 (E180) and −119.8 on v3 (E192_T800) — 30 meV/atom, but on different decorations. E230 removes the decoration: E191_MoTa_rand0's exact cell (16 atoms, a = 6.492 Å) at the v2 standard (50/400 Ry, k 5³ by the v2 k-density rule; v2 references), against its v3 value −123.1. Prediction: |E_form(v2) − E_form(v3)| > 15 meV/atom. Decision: > 15 → the five v2 rows leave every refit until recomputed at v3, and conclusions drawn on them (E172–E180) are re-read; 5–15 → kept with a per-standard uncertainty; < 5 → the standards agree and E180 vs E192_T800 is decoration/SRO. Chain runs/e230_chain.sh starts only after every other pw.x chain has finished (≈ 15 min of pw.x).

EXPERIMENTS.md · lines 15926–15928

E230 RESULT. The same decoration at the v2 standard: −127.5 vs v3 −123.1 → 4.4 meV/atom, under the 5 meV line: the standards agree and the five v2 rows stay; E180 vs E192_T800's 30 meV is decoration and v5-sampled SRO.