Sc-2024-07-19

The current structure dataset comprises 14048 structures. Procedures to generate structures and estimate MLPs are found in A. Seko, J. Appl. Phys. 133, 011101 (2023).

Predictions using Pareto optimal MLPs

These properties are calculated for MLP equilibrium structures obtained by performing local structure optimizations from the DFT equilibrium structures. These DFT equilibrium structures are obtained by optimizing prototype structures that are included in ICSD. As a result, the structure type of the converged structure may sometimes differ from the one shown in the legend.

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

Sc-2024-07-19 shows large prediction errors. Distributed MLPs should be carefully used. MLPs are often accurate for reasonable structures, but it is sometimes inaccurate for unrealistic structures.

Pareto optimals (on convex hull)

Name	Time	RMSE	Predictions	Files
polymlp-00003	0.032 / 0.003	28.206 / 0.1705	–	–
polymlp-00035	0.125 / 0.014	9.878 / 0.0986	–	–
polymlp-00064	0.158 / 0.016	7.975 / 0.0907	–	–
polymlp-00067	0.188 / 0.017	6.945 / 0.0817	–	–
polymlp-00096	0.234 / 0.022	6.319 / 0.0796	–	–
polymlp-00068	0.452 / 0.034	4.633 / 0.0710	predictions	polymlp.lammps polymlp.in
polymlp-00040	0.585 / 0.039	4.058 / 0.0714	predictions	polymlp.lammps polymlp.in
polymlp-00069	0.798 / 0.044	3.588 / 0.0647	predictions	polymlp.lammps polymlp.in
polymlp-00098	1.034 / 0.059	3.191 / 0.0631	predictions	polymlp.lammps polymlp.in
polymlp-00185	1.496 / 0.079	3.024 / 0.0640	predictions	polymlp.lammps polymlp.in
polymlp-00100	1.713 / 0.092	2.949 / 0.0610	predictions	polymlp.lammps polymlp.in
polymlp-00187	2.592 / 0.123	2.768 / 0.0620	predictions	polymlp.lammps polymlp.in
polymlp-00192	2.618 / 0.126	2.766 / 0.0620	predictions	polymlp.lammps polymlp.in
polymlp-00166	7.253 / 0.481	2.436 / 0.0618	predictions	polymlp.lammps polymlp.in
polymlp-00284	9.954 / 0.682	2.396 / 0.0580	predictions	polymlp.lammps polymlp.in
polymlp-00285	12.267 / 0.814	2.368 / 0.0605	predictions	polymlp.lammps polymlp.in

Units:

• Time: [ms] (1core/36cores)

• RMSE: [meV/atom]/[eV/ang.]

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

• All Pareto optimal MLPs are available here.