Au-2021-08-04-all-icsd¶

The current structure dataset comprises 12928 structures generated from unique ICSD prototype structures composed of single elements with zero oxidation state. A more detailed procedure is found in Phys. Rev. B 99, 214108 (2019). The procedure to estimate interatomic potentials from the dataset is found in Phys. Rev. B 99, 214108 (2019) and Phys. Rev. B 102, 174104 (2020).

Improvement from **-dataset-10000-all-icsd

More robust for structures with a small interatomic distance
More robust for structures with a large interatomic distance
More complex potential models are included.
MLPs are estimated without using DFT stress tensors.
MLPs are estimated by using small regression weights for energetically unstable structures.

Predictions using Pareto optimal MLPs

../../_images/prediction-ecoh-volume116.png

The cohesive energy and volume are obtained by performing a local structure optimization from the DFT equilibrium structure. In addition, the DFT equilibrium structure is obtained by optimizing a prototype structure included in ICSD, and the prototype is used as the structure legend in the figure. Therefore, the structure type of the converged structure is sometimes different from that shown in the legend even if the potential energy surface predicted by MLP is almost the same as the true one.

The other properties predicted by each Pareto optimal MLP are available from column Predictions in the following table.

Pareto optimals¶
Name	Time [ms] (1core/36cores)	RMSE [meV/atom]/[eV/A]	Predictions	Files
pair-14	0.017 / 0.038	50.135 / 0.1749	–	–
pair-27	0.018 / 0.005	21.534 / 0.0972	–	–
pair-28	0.027 / 0.007	17.692 / 0.0935	–	–
pair-31	0.050 / 0.007	17.068 / 0.0915	–	–
pair-32	0.051 / 0.010	15.052 / 0.0875	–	–
pair-33	0.069 / 0.012	14.200 / 0.0835	–	–
pair-34	0.096 / 0.016	13.760 / 0.0832	–	–
pair-37	0.136 / 0.018	11.775 / 0.0800	–	–
pair-38	0.171 / 0.022	11.154 / 0.0780	–	–
gtinv-300	0.228 / 0.023	4.0930 / 0.0283	predictions	`mlp.lammps` `input` `log`
gtinv-235	0.281 / 0.023	3.8650 / 0.0264	predictions	`mlp.lammps` `input` `log`
gtinv-303	0.339 / 0.029	3.5431 / 0.0296	predictions	`mlp.lammps` `input` `log`
gtinv-175	0.376 / 0.031	3.1135 / 0.0273	predictions	`mlp.lammps` `input` `log`
gtinv-240	0.393 / 0.030	3.0781 / 0.0272	predictions	`mlp.lammps` `input` `log`
gtinv-312	0.472 / 0.036	2.7237 / 0.0254	predictions	`mlp.lammps` `input` `log`
gtinv-190	0.523 / 0.038	2.4203 / 0.0236	predictions	`mlp.lammps` `input` `log`
gtinv-255	0.573 / 0.040	2.3863 / 0.0236	predictions	`mlp.lammps` `input` `log`
gtinv-195	0.748 / 0.053	2.1391 / 0.0233	predictions	`mlp.lammps` `input` `log`
gtinv-260	0.787 / 0.053	2.1083 / 0.0233	predictions	`mlp.lammps` `input` `log`
gtinv-205	1.307 / 0.088	2.0804 / 0.0247	predictions	`mlp.lammps` `input` `log`
gtinv-270	1.323 / 0.091	2.0760 / 0.0246	predictions	`mlp.lammps` `input` `log`
gtinv-177	1.475 / 0.092	1.7407 / 0.0205	predictions	`mlp.lammps` `input` `log`
gtinv-242	1.506 / 0.092	1.7276 / 0.0205	predictions	`mlp.lammps` `input` `log`
gtinv-342	1.775 / 0.083	1.6772 / 0.0220	predictions	`mlp.lammps` `input` `log`
gtinv-220	1.890 / 0.117	1.5351 / 0.0214	predictions	`mlp.lammps` `input` `log`
gtinv-285	2.031 / 0.123	1.5155 / 0.0214	predictions	`mlp.lammps` `input` `log`
gtinv-261	2.341 / 0.142	1.4972 / 0.0202	predictions	`mlp.lammps` `input` `log`
gtinv-225	2.394 / 0.148	1.4784 / 0.0216	predictions	`mlp.lammps` `input` `log`
gtinv-290	2.494 / 0.149	1.4703 / 0.0216	predictions	`mlp.lammps` `input` `log`
gtinv-340	2.596 / 0.116	1.4082 / 0.0196	predictions	`mlp.lammps` `input` `log`
gtinv-197	2.819 / 0.158	1.3123 / 0.0180	predictions	`mlp.lammps` `input` `log`
gtinv-262	2.839 / 0.166	1.3089 / 0.0179	predictions	`mlp.lammps` `input` `log`
gtinv-343	2.986 / 0.131	1.2878 / 0.0199	predictions	`mlp.lammps` `input` `log`
gtinv-267	3.516 / 0.196	1.2720 / 0.0185	predictions	`mlp.lammps` `input` `log`
gtinv-202	3.656 / 0.193	1.2650 / 0.0185	predictions	`mlp.lammps` `input` `log`
gtinv-360	4.081 / 0.224	1.2595 / 0.0202	predictions	`mlp.lammps` `input` `log`
gtinv-344	4.588 / 0.248	1.1027 / 0.0181	predictions	`mlp.lammps` `input` `log`
gtinv-222	6.837 / 0.369	1.0570 / 0.0167	predictions	`mlp.lammps` `input` `log`
gtinv-287	6.946 / 0.371	1.0496 / 0.0167	predictions	`mlp.lammps` `input` `log`
gtinv-227	8.349 / 0.454	1.0351 / 0.0167	predictions	`mlp.lammps` `input` `log`
gtinv-292	8.473 / 0.461	1.0342 / 0.0167	predictions	`mlp.lammps` `input` `log`
gtinv-224	14.066 / 0.687	0.9943 / 0.0161	predictions	`mlp.lammps` `input` `log`
gtinv-289	14.825 / 0.712	0.9896 / 0.0161	predictions	`mlp.lammps` `input` `log`
gtinv-294	17.107 / 0.825	0.9699 / 0.0161	predictions	`mlp.lammps` `input` `log`
gtinv-229	17.857 / 0.829	0.9674 / 0.0161	predictions	`mlp.lammps` `input` `log`

Column “Time” shows the time required to compute the energy and forces for 1 MD step and 1 atom, which is estimated from a simulation of 10 runs for a structure with 284 atoms using a workstation with Intel(R) Xeon(R) CPU E5-2695 v4 @ 2.10GHz. Note that the MLPs should be carefully used for extreme structures. The MLPs often return meaningless values for them.

All Pareto optimal MLPs are available here.