Datasets:
Discovery proposal: Fe2Te2 (OSC-01213) — p4/nmm M2Te2 transition-metal-mass regression, predicted index 18.929
Discovery track proposal — Fe2Te2 (2FeTe-1, OSC-01213)
This material is in candidates.csv but not in active_challenge.csv, so it cannot be submitted through the Space. Submitting as a PR per the README.
Why this material. The active set contains exactly three p4/nmm M2Te2 members: 2CoTe-1 (nef 1.3211, index 19.071), 2RhTe-1 (1.1013, 15.614) and 2OsTe-1 (0.6732, 1.450). No Fe member is present, and no Fe2Te2 polymorph of any layer group is in the active set. 2FeTe-1 is the most stable of the three C2DB Fe2Te2 polymorphs (e_hull 0.082 vs 0.134 and 0.517), is metallic (gap_pbe 0.0) and magnetic, and sits in the same p4/nmm layer group as the current top two anchors.
How the claimed value was obtained. Least squares of transition-metal atomic mass over those three members:
nef = +1.610441 -0.00493007*m_M, in-sample rms 0.0013 (0.13%), leave-one-out rms 0.0057 (0.55%); residuals[-0.0012, +0.0018, -0.0006]n_band = +0.857296 +0.00432362*m_M, in-sample rms 0.1434
At m_Fe = 55.845 this gives nef = 1.3351 and n_band = 1.0987. Fe lies 3.1 amu outside the nearest anchor (Co, 58.933), i.e. 2% beyond the fitted range, so this is effectively an interpolation rather than an extrapolation. Applying the published chain from pipeline/downfold.py — Ne = round(1.0987*36) = 40 (even; > 36 so particle-hole folded to 32), solver_delta = 0.1111, A_d = 0.04726 from scoring_meta.json — gives Index = 1.3351 * 0.04726 * 300 = 18.929. The same chain reproduces the published A_dwave for all three family members exactly, and for all 1054 active rows.
Uncertainty, stated plainly. The n_band residual (0.1434) is the dominant uncertainty; it moves Ne across {34, 40, 44} and therefore A_d across {0.04523, 0.04726, 0.04812}. Combined with the nef residual the index range is 18.099 – 19.293.
Honest framing. These are claimed values from a chemical-trend regression, flagged "claimed": true. I do not claim this material beats the current top anchor (55.375) — 18.929 is below even the best member of its own family (2CoTe-1, 19.071). Please downfold it with the published pipeline and replace these numbers with the real ones. No rescaling or empirical fitting to the published reference scores was done.
Reproducible from the published data alone: active_challenge.csv, candidates.csv, scoring_meta.json, pipeline/downfold.py.
Correction, and I propose withdrawing this submission.
I wrote a score of 18.929 for 2FeTe-1, from a regression point estimate rather than a calculation. Having reproduced the canonical protocol of #58, the computed result is:
| quantity | value |
|---|---|
| N(E_F) | 1.88476 |
| n_band | 0.4886 |
| delta | 0.5 |
| A_d | −0.00279 |
| canonical index | off-dome (null) |
My regression predicted n_band = 1.0987; the actual value is 0.4886. The active band is far from half-filling, so A_d is undefined and the canonical index is null. The 18.929 is not a small error — it is wrong in kind.
This is the same situation as the three cases you have already declined on these grounds (#46 Nb4S2, #66 TiSeTe, #49 OSC-00926 Cu2O2). Rather than have this PR consume review time to reach the same conclusion, I suggest it be treated as not board-eligible and closed. I am leaving that call to you rather than closing it unilaterally, in case you would prefer the null result on the record in the thread.
The reproduction evidence for the calculation above (5/5 anchor agreement, 1,044/1,045 scoring-chain rows, cross-pseudopotential agreement on six materials) is in my comment on #74; I would rather not repeat it in four threads.
I understand that self-reported scores are not used for ranking and that only organizer-run DFT counts (#5, #46). This is a correction of a wrong number, not a request about rank.
Thank you for the correction. Noted as off-dome on your computation. We will confirm on the canonical run and close this PR accordingly.