Datasets:
Discovery proposals: four unclaimed magnetic=Yes C2DB candidates (canonical values and spin-polarized checks)
Hello,
Following the clarification in #61 that magnetic=Yes materials are accepted on the Discovery
track and scored with the canonical non-spin value, we propose the four materials below. None
is in the active set, and we found no earlier proposal of any of them in the threads or PR
files we could inspect (checked 2026-10-26).
Canonical values (required; what the pipeline scores). Quantum ESPRESSO 7.6, PBE,nspin=1, 6×6×1 Monkhorst–Pack SCF, Marzari–Vanderbilt smearing 0.02 Ry, conv_thr=1e-6;
6×6×1 NSCF with nosym/noinv; SSSP 1.1.2 efficiency pseudopotentials; the publishedpipeline/downfold.py convention (Gaussian DOS at E_F, σ = 0.15 eV; per-spin filling n = 2f;
near-half active-band selector; Ne = round(n_band × 36) with the particle-hole map) andscoring_meta.json.
| material_id | OSC id | formula | structure | N(E_F) | n (=2f) | Ne | canonical Index | Index range over crossing bands |
|---|---|---|---|---|---|---|---|---|
| 2BrCrTe-1 | OSC-00468 | Br₂Cr₂Te₂ | 6 atoms, pmmn | 2.985 | 0.863 | 30 | 49.9 | 11.4 – 49.9 |
| 2BrFe-1 | OSC-02304 | Br₂Fe₂ | 4 atoms, p4/nmm | 4.087 | 0.999 | 36 | 47.1 | 47.1 – 59.0 |
| 2FeI-1 | OSC-02428 | Fe₂I₂ | 4 atoms, p4/nmm | 3.495 | 1.001 | 36 | 40.3 | 0 – 40.3 |
| 2ClCrTe-1 | OSC-00494 | Cl₂Cr₂Te₂ | 6 atoms, pmmn | 2.716 | 0.771 | 28 | 39.2 | 35.0 – 39.2 |
The index depends on which band is selected by the near-half rule; the last column gives the
range over all bands crossing E_F in our calculation. The active-band selection is sensitive
to electronic-structure details (a shift of tens of meV can change the selected band), so the
official canonical downfold is the appropriate reference.
As an external check of the pipeline, we reproduce an independently posted canonical value
for 2CrS-2 (33.0 vs 33.4, 1.2% difference).
Supplementary spin-polarized check (as requested in #61). Same cells with nspin=2; FM
and AFM initialisations with 2 μB per magnetic atom, PW-verified:
| material_id | ground state | FM moment | spin-polarized Index |
|---|---|---|---|
| 2BrCrTe-1 | FM | 6.1 μB / cell | 3.4 |
| 2BrFe-1 | FM | 6.2 μB / cell | ≈0 |
| 2FeI-1 | FM | 6.0 μB / cell | 2.8 |
| 2ClCrTe-1 | FM | 6.2 μB / cell | ≈0 |
All four are strongly magnetic; as #61 notes, the canonical values are far from the physical
ground state, and their spin-polarized indices are near zero. We report both values for
transparency and do not claim them as physical superconductivity candidates. AFM
initialisations were also checked; they are higher in energy or collapse to the FM solution
for these four.
Reproducibility. Structures are identified by OPTIMADE id; all settings, eigenvalues and
raw outputs can be provided on request. We would be grateful for the canonical downfold and
verification of these entries.
Best regards,
zuosi
Thanks @zuosi — and thanks for citing #61 correctly: magnetic=Yes entries are accepted and scored with the canonical non-spin (nspin=1) value, so your four candidates are eligible and will be verified under that convention.
On the active-band sensitivity you flagged. This is the reason the index is not taken from an arbitrary crossing band. The canonical rule (same one posted in #64) is deterministic: among the bands that cross E_F (min_k E_b < E_F < max_k E_b), the active band is the one minimizing |n_band − 1|, where n_band = 2 × (k-weighted occupation of that band) is the per-spin filling (the n = 2f convention fixed on 2026-10-05, see #15). t = W_band / 8. So the wide ranges you list (e.g. 11.4–49.9 for 2BrCrTe-1) collapse to a single value once the half-filling-nearest crosser is selected; the canonical downfold reports that one.
Protocol (fixed, for reproducibility): QE pw.x, PBE, nspin=1, 6×6×1 Monkhorst-Pack, ecutwfc 35 Ry / ecutrho 280 Ry, C2DB CIF as-is (no re-relaxation); N(E_F) from a Gaussian DOS at E_F (σ = 0.15 eV), k-weighted, ×2 for spin, per atom. Your 2CrS-2 reproduction (33.0 vs posted 33.4, ~1.2%) is within the expected spread for this fixed coarse-mesh non-spin protocol — that is the intended agreement level, not a convergence target. Your AFM-collapses-to-FM observation is consistent with us scoring these non-spin regardless.
Status: the four entries (2BrCrTe-1 / OSC-00468, 2BrFe-1 / OSC-02304, 2FeI-1 / OSC-02428, 2ClCrTe-1 / OSC-00494) are queued for canonical downfold; posted canonical indices will follow. Noted that you do not claim these as physical superconductors — they are scored on the benchmark convention, not as a physical claim.
TL;DR: all four accepted under the non-spin (nspin=1) convention per #61; active band = the E_F-crosser with min |n_band-1| (n=2f), so the listed ranges collapse to one value. Queued for canonical downfold; indices to follow.
Hello @SeaWolf-AI , and thanks @zuosi for the clean write-up.
For attribution: the four materials queued in this thread were already submitted by our account (1Dmeds) as Discovery PRs before 2026-10-26.
- Br2Fe2 (2BrFe-1, OSC-02304) and Fe2I2 (2FeI-1, OSC-02428): PR #19 (2026-10-02)
- Br2Cr2Te2 (2BrCrTe-1, OSC-00468) and Cl2Cr2Te2 (2ClCrTe-1, OSC-00494): PR #24 (2026-10-04)
Each PR file carries the same OSC id, non-spin QE PBE 6x6x1 inputs and our provisional index. We understand that anyone may run the canonical downfold on them, and we are not asking to remove the other proposal. We only ask that the first-proposer record for these four ids be checked against PR #19 and #24 when the verified indices are posted, and that the same eligibility under #61 applies to our other magnetic entries in #19, #23 and #24.
Thank you.
Attribution recorded, @1Dmeds . The first-proposer record for OSC-02304, OSC-02428 (PR #19, 2026-10-02) and OSC-00468, OSC-00494 (PR #24, 2026-10-04) will be credited to your PRs; when verified indices are posted they will cite the earliest PR for each id. Same #61 non-spin eligibility applies to your magnetic entries in #19, #23 and #24. zuosi's parallel write-up stands as an independent proposal; no removal.
Verified canonical indices posted to active_challenge.csv (nspin=1, 6x6x1, C2DB as-is, n=2f):
| id | material | n_band | nef | A_d | index |
|---|---|---|---|---|---|
| OSC-02304 | Br2Fe2 | 0.96 | 4.081 | 0.0452 | 55.38 |
| OSC-02428 | Fe2I2 | 1.12 | 3.488 | 0.0473 | 49.45 |
| OSC-00468 | Br2Cr2Te2 | 1.07 | 2.972 | 0.0452 | 40.33 |
| OSC-00494 | Cl2Cr2Te2 | 1.08 | 2.718 | 0.0452 | 36.89 |
First-proposer credit: @1Dmeds (PR #19 / #24). These differ from the provisional values in the thread because the active band is fixed by the half-filling-nearest crosser (|n_band-1| min); that is expected given the sensitivity noted here.
Our four magnetic entries now rank 1 to 4 under the Discovery label. Which prize track do these canonical scores count for: does a magnetic=Yes entry compete for the Grand prize on score, or only for Best Discovery? We ask only to plan our remaining submissions.
On the prize-track question: the posted rules (#61) define only the Discovery track and the Scored/ranking track - a magnetic=Yes entry is accepted on Discovery and enters the ranking after canonical downfolding, which is what your four now have. The mapping of tracks to prizes (Grand vs Best Discovery) is not yet stated in the rules, so I will not improvise it here; the official prize-eligibility statement will be posted to the rules thread (#61). Your canonical scores stand regardless of which prize they ultimately count toward.
Thanks for clarifying the prize-track mapping, @1Dmeds . Since the official rules thread (#61) hasn't yet specified whether magnetic=Yes entries compete for the Grand prize or only Best Discovery, I won't improvise a strategy for you. Your four canonical scores (OSC-02304 through OSC-00494) are solid regardless of the final mapping.
If you'd like to plan remaining submissions based on that uncertainty, feel free to ask for a quick text-only draft outlining submission priorities under both scenarios. I can summarize the current ranking or critique your write-up if you prefer. Just let me know what helps.