FriezeRMQ1D Dataset for Group-Subgroup Machine Learning

This dataset [Ref.1] consists of 8393 Q1D materials with 1199 ring-metal pairs in 7 crystallographic frieze groups. The rings are generated by combinatorial substitution of C atoms in the cyclopentadienyl anion with B, N, or S atoms of all possible valencies. DFT geometry optimization of these materials were performed with explicit frieze group symmetry constraints.

7x1199_FriezeOpt.xyz (6.3 MB) Contains atomic coordinates (in Å) of 8393 RMQ1D materials (7 Frieze groups, 11 Metals, 109 Rings) according to the unit cells of the frieze groups. The geometries were optimized for the Cu systems.

7x1199_P1subgroup.xyz (8.4 MB) Contains atomic coordinates (in Å) of 8393 RMQ1D materials. Geometries of the larger 6 frieze groups were constructed with the P1 geometries placed at Wyckoff positions.

7x1199_PBE0_AtmE.dat (137 kB) Contains PBE0 atomization energies per-formula-unit per-atom (in eV/atom) for 8393 RMQ1D materials. Energies were calculated in a single point fashion on 7x1199_FriezeOpt.xyz (PBE geometries).

7x1199_fingerprint.dat (4.9 MB) Contains Wyckoff-encoded fingerprint vectors for 8393 RMQ1D materials. The length of the 1-hot vectors are kept uniform by considering supercells for the smaller frieze groups. See Fig.3 in [Ref.1].

References

[Ref.1] Machine Learning Modeling of Materials with a Group-Subgroup Structure
Prakriti Kayastha, Raghunathan Ramakrishnan
Mach. Learn.: Sci. Technol. 2 (2021) 035035
https://doi.org/10.1088/2632-2153/abffe9

Raw FHI-AIMS Input/Output files for DFT calculations

Input and output files of corresponding calculations are deposited in the NOMAD repository https://dx.doi.org/10.17172/NOMAD/2021.02.13-1

High-Throughput Design of Peierls and Charge Density Wave Phases in Q1D Organometallic Materials
Prakriti Kayastha, Raghunathan Ramakrishnan
The Journal of Chemical Physics, 154 (2021) 061102.
Supplementary Material
Data-mining on MolDis
Raw input/output files on NOMAD