1887
banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Automated fit of high-dimensional potential energy surfaces using cluster analysis and interpolation over descriptors of chemical environment
Rent:
Rent this article for
USD
10.1063/1.4846297
/content/aip/journal/jcp/139/23/10.1063/1.4846297
http://aip.metastore.ingenta.com/content/aip/journal/jcp/139/23/10.1063/1.4846297

Figures

Image of FIG. 1.
FIG. 1.

GABEF energy vs DFT energy (eV) for the 168 configurations of the Ag test set.

Image of FIG. 2.
FIG. 2.

GABEF energy vs DFT energy (eV) for the 104 configurations of the SnPb test set.

Image of FIG. 3.
FIG. 3.

GABEF energy vs DFT energy (eV) for the 549 configurations of the Si test set.

Image of FIG. 4.
FIG. 4.

GABEF energy vs DFT energy (eV) for the 586 configurations of the Li test set.

Image of FIG. 5.
FIG. 5.

((a)–(d)) Four low-energy Li isomers used as parent structures for the set of 7000 distorted geometries.

Tables

Generic image for table
Table I.

Computing time (seconds on a single core CPU) and RMSE (eV) measured against the fit set and the test set of Ag averaged over ten independent runs. The alternate descriptors (Eqs. (14)–(18) ) were used along with different numbers of atom types “.”

Generic image for table
Table II.

The RMSE measured against the fit set, and the RMSE and MAE measured against the test set, in ten independent runs with Ag using 60 atom types and the original descriptors (Eqs. (10)–(13) ). The RMSD of DFT energies to the DFT mean energy is 1.581 eV. The time (single core CPU) for fitting was 300 s and the time for testing was 8.5 s.

Generic image for table
Table III.

The RMSE measured against the fit set, and the RMSE and MAE measured against the test set, in ten independent runs with Ag using 60 atom types and the alternate descriptors (Eqs. (14)–(18) ). The RMSD of DFT energies to the DFT mean energy is 1.581 eV. The time (single core CPU) for fitting was 318 s and the time for testing was 9.1 s.

Generic image for table
Table IV.

RMSE and MAE obtained by averaging ten runs for varying sizes of Li fitting sets ( ) and numbers of atom types (). The top part of the table is for a set of structures generated with random displacements varying between 0 and 0.5 Å, and the bottom part is for a set of structures generated with random distortions varying between 0 and 0.3 Å (see text for details). The RMSD of DFT energies to the mean is 0.356 eV for the first set and 0.135 eV for the second set.

Loading

Article metrics loading...

/content/aip/journal/jcp/139/23/10.1063/1.4846297
2013-12-18
2014-04-16
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Automated fit of high-dimensional potential energy surfaces using cluster analysis and interpolation over descriptors of chemical environment
http://aip.metastore.ingenta.com/content/aip/journal/jcp/139/23/10.1063/1.4846297
10.1063/1.4846297
SEARCH_EXPAND_ITEM