Datasets:
OSC open discovery: Cu2S2, Ag2Se2, Ni2SSe (OSC-00571, OSC-00418, OSC-00749) — organizer-equivalent DFT, both filling conventions
Three non-magnetic C2DB metals outside the active set. They are not top-of-table candidates (16–19 on
either convention); they are proposed because they hold up under both the present downfold.py and
the corrected filling of #15 / #18, and because the numbers come from a pipeline that now reproduces
the organizers' values almost exactly.
Pipeline. QE 7.5 PBE SCF, SSSP-efficiency 1.1 pseudopotentials, 6×6×1 k, Marzari–Vanderbilt smearing
0.02 Ry, nspin = 1, then the public pipeline/downfold.py unchanged ("current") and with the two-line
patch of #18 ("corrected"). Calibration on 43 active materials: N(E_F) within 1.4% of the official
values (median 0.0%, 39/43 within 1%), n_band within 0.003, and solver_delta identical for 43/43.
Cross-checks on the same footing: Ag₂S₂ (2AgS-3, #17) gives N(E_F) = 1.345 → 18.2 current / 19.4
corrected; our own AuI₂ and Ni₂Se₂ (#14) give 16.1 / 19.0 and 18.6 / 15.1 (see the update there).
| material | C2DB uid | OSC id | N(E_F) | current: n_band → δ → index | corrected: n → δ → index | e_hull |
|---|---|---|---|---|---|---|
| Cu₂S₂ | 2CuS-1 | OSC-00571 | 1.153 | 0.973 → 0.056 → 15.7 | 0.860 → 0.167 → 19.3 | 0.111 |
| Ag₂Se₂ | 2AgSe-4 | OSC-00418 | 1.276 | 0.974 → 0.056 → 17.3 | 0.913 → 0.111 → 18.1 | 0.220 |
| Ni₂SSe | 1SSeNi2-1 | OSC-00749 | 1.263 | 0.924 → 0.111 → 17.9 | 0.921 → 0.111 → 17.9 | 0.083 |
Candidates that score higher only under the present convention (As₂W 20.5 → 1.8, I₂Pd₂ 19.2 → 0.0,
Li₂Rh 18.4 → 0.0, IrLi₂ 18.3 → 0.0, Cu₂Sr₂ 17.7 → 0.0 after the fix) are deliberately left out.
Files follow the submissions/ format of #9/#10/#17, with both indices recorded (provisional_score =
corrected convention, score_current_pipeline = present pipeline). Fillings sit on the solver's Ne grid,
so an official value may still differ by one Ne step. Inputs and outputs available on request.
🤖 Generated with Claude Code