WyFormer generated structures

WyFormer generated datasetsStructures generated by WyFormer, with various post processing. Used in the ICML 2025 paper "Wyckoff Transformer: Generation of Symmetric Crystals".The folder structure is the following: the first is the dataset which was used for training WyFormer, using only train and validation parts. Then the folder structure corresponds to transformations of the data.mp_20/WyckoffTransformer 10k formally valid Wyckoff representations generated by WyFormer trained on MP-20 dataset.mp_20/WyckoffTransformer/DiffCSP++10k 9999 structures obtained with DifCSP++; it failed for one Wyckoff representation, we consider this structure unstable. Can be considered as the "official" WyFormer sample.mp_20/WyckoffTransformer/DiffCSP++10k/CHGNet_free/DFT CHGNet pre-relaxation followed by DFT relaxation; for some structures the DFT relaxation failed, we consider them unstable. The relaxation was obtained using MP-compatiable MPGGADoubleRelaxStaticMaker. Note that material indices unfortunately got permuted at the CHGNet pre-relaxations step. Used in Table 1. Can be considered as the "official" WyFormer DFT-relaxed sample.mp_20/WyckoffTransformer/DiffCSP++10k/CHGNet_free/DFT-GGA-relax-1 same as above, but relaxed with a single invocation of MPRelaxSet. This is less precise, not strictly compatible to Materials Project, but is the same as reported in FlowMM paper and code. Used in Table 1.mp_20/WyckoffTransformer/DiffCSP++/ 1k structures obtained with DifCSP++mp_20/WyckoffTransformer/DiffCSP++/DFT/ DFT relaxation of 105 novel and unique structures, MPGGADoubleRelaxStaticMakermp_20/WyckoffTransformer/CrySPR/CHGNet_fix/ 1k structures obtained with CrySPR and CHGNet, whith a constraint during the relaxation that maintained the Wyckoff positionsmp_20/WyckoffTransformer/CrySPR/CHGNet_fix/DFT/ DFT relaxation of 105 novel and unique structures, MPGGADoubleRelaxStaticMaker.mpts_52/WyckoffTransformer/CrySPR/CHGNet_fix 1k structures generated with Wyformer trained on MPTS-52 dataset, then CrySPR and CHGNet, with a constraint during the relaxation that maintained the Wyckoff positions.Format descriptionstructure - pymatgen.core.structure.Structuregroup, species, numIons, sites - arguments to pyxtal.from_random. For */WyckoffTransformer/data.csv.gz they were generated with WyFormer, for the rest they were obtained from structures with pyxtal.from_seed. Note the the indexing within those fields is by chemical element, not by Wyckoff position.site_symmetries, elements, multiplicity, wyckoff_letters, sites_enumeration, dof - information about the Wyckoff positions, indexed by Wyckoff position. The dof is the number of degrees of freedom for the Wyckoff position, i.e. the number of free parameters in the Wyckoff position. sites_enumeration enumerates the Wyckoff position with the same site symmetry, see the paper for details. For example, for space group 2 aka P-1, Wyckoff position a has site symmetry -1 and enumeration 0, while b has site symmetry -1 and enumeration 1.sites_enumeration_augmented - possible variants of the enumeration, depend on the arbitrary choice of the space group Euclidean normalizer, e. g. unit cell center. See the preprint for details.smact_validity - "Compositional Validity" computed with SMACT. Not all structures in MP-20 conform to this criterion.structural_validity - "Structural Validity" introduced by CDVAE, whether any two atoms are closer than 0.5 Angstromscdvae_e - energy predicted by the model included in CDVAE, used for EMD(E) distribution similarity metricchgnet_energy_per_atom - energy per atom from CHGNet relaxationchgnet_e_above_hull_corrected - energy above hull from CHGNet relaxation, taking into account MP energy correctiondft_e_uncorrected - raw potential energy from DFT relaxationdft_e_corrected - potential energy from DFT relaxation, corrected with MaterialsProject2020Compatibilitydft_e_above_hull_corrected - energy above hull computed from DFT relaxation computed using 2023-02-07-ppd-mp.pkl.gz distributed by matbench-discovery as reference.entry - pymatgen.entries.ComputedEntry containing the results of the DFT run.Citatation@article{kazeev2025wyckoff, title={{Wyckoff Transformer: Generation of symmetric crystals}}, author={Kazeev, Nikita and Nong, Wei and Romanov, Ignat and Zhu, Ruiming and Ustyuzhanin, Andrey and Yamazaki, Shuya and Hippalgaonkar, Kedar}, journal={arXiv preprint arXiv:2503.02407}, year={2025}}