Candidates shipped with the paper, screened here¶
The 26 candidate structures in examples/perov5/paper_results, relaxed with MACE-MP-0 and scored with the protocol's stability, uniqueness and novelty criteria. Their targets and budget are those of the paper, so the record is not ranked.
| field | value |
|---|---|
| Dataset | Perov-5, protocol paper |
| Type | model |
| Inputs | structure, property:heat_all, property:dir_gap |
| Parameters | – |
| Training data | Perov-5 train (11,356) |
| Status | Computed here on 2026-10-05 |
| Evidence level | MLIP validated |
| Added | 2026-10-05 by Anand Babu (UCLouvain) |
| Links | paper · code |
Inverse design¶
| metric | value | definition |
|---|---|---|
| SUN | 0.654 | stable, unique and novel candidates divided by the budget of 54; a candidate that was not delivered counts as a failure |
| Stable | 1.000 | fraction of delivered candidates whose MACE-MP-0 formation energy after relaxation is at most 0.10 eV/atom, against elemental reference phases (the criterion of meidnet screen and of the paper) |
| Unique | 0.923 | fraction of delivered candidates whose composition does not repeat an earlier one |
| Novel | 0.692 | fraction of delivered candidates whose composition is not in the data set (training, validation and test splits) |
| ΔHf | -1.27 eV/atom | median MACE-MP-0 formation energy of the delivered candidates |
| Delivered | 26 | candidates delivered out of the budget of 54 |
| SUN count | 17 | stable, unique and novel candidates |
Candidates¶
Every candidate with its MLIP formation energy, novelty and, where Perov-5 has the same sites, the DFT band gap.
| candidate | formula | target gap (eV) | ΔHf (eV/atom) | stable | novel | DFT gap (eV) |
|---|---|---|---|---|---|---|
| chalcogenide_chalcogenide_T1_R1_2 | LaCrTe3 | -1.485 | yes | yes | – | |
| chalcogenide_chalcogenide_T1_R1_3 | NaNbTe3 | -1.208 | yes | yes | – | |
| chalcogenide_chalcogenide_T2_R1_1 | KNbTe3 | -1.338 | yes | yes | – | |
| chalcogenide_chalcogenide_T2_R1_2 | BaTe3W | -1.310 | yes | yes | – | |
| chalcogenide_chalcogenide_T2_R1_3 | BaTe3Mo | -1.429 | yes | yes | – | |
| halide_halide_T1_R1_1 | NaMnI3 | -0.900 | yes | yes | – | |
| halide_halide_T1_R1_2 | NaNiI3 | -0.723 | yes | yes | – | |
| halide_halide_T1_R1_3 | RbMnI3 | -1.123 | yes | yes | – | |
| halide_halide_T1_R1_4 | NaNiI3 | -0.723 | yes | yes | – | |
| halide_halide_T1_R2_4 | NaZnI3 | -0.928 | yes | yes | – | |
| halide_halide_T2_R1_1 | RbPbI3 | -1.240 | yes | yes | – | |
| halide_halide_T2_R1_2 | KMnI3 | -1.083 | yes | yes | – | |
| halide_halide_T2_R1_3 | NaCoI3 | -0.586 | yes | yes | – | |
| halide_halide_T2_R1_4 | NaSnI3 | -0.936 | yes | yes | – | |
| halide_halide_T3_R1_1 | CsMnI3 | -1.160 | yes | yes | – | |
| halide_halide_T3_R1_2 | NaFeI3 | -0.553 | yes | yes | – | |
| halide_halide_T3_R2_3 | CsPbI3 | -1.302 | yes | yes | – | |
| halide_halide_T3_R2_4 | RbSnI3 | -1.194 | yes | yes | – | |
| oxide_oxide_T1_R1_1 | SrZrO3 | -3.478 | yes | no | – | |
| oxide_oxide_T1_R1_2 | LaMnO3 | -2.467 | yes | no | – | |
| oxide_oxide_T1_R1_3 | LaScO3 | -3.629 | yes | no | – | |
| oxide_oxide_T1_R1_4 | NaTaO3 | -2.902 | yes | no | – | |
| oxide_oxide_T2_R1_1 | LaMnO3 | -2.467 | yes | no | – | |
| oxide_oxide_T2_R1_2 | BaMnO3 | -1.919 | yes | no | – | |
| oxide_oxide_T2_R1_3 | SrVO3 | -2.416 | yes | no | – | |
| oxide_oxide_T2_R1_4 | BaPbO3 | -1.970 | yes | no | – |
Outputs¶
- inverse_design:
benchmarks/runs/perov5/paper-candidates/inverse_design
Reproduce¶
python scripts/benchmarks.py paper-candidates perov5
Computed on Intel64 Family 6 Model 197 Stepping 2, GenuineIntel (16 threads), CPU only.