Figure1

Exploring phase transitions in CdSe: a machine learning and swarm intelligence approach

Figure 1. Graph of comparison of neural network predictions for (A) energies and forces and (B) stress versus the DFT calculated energies, forces and stresses for the 751 structures in the testing set. The black dashed line represents perfect agreement between the prediction by the neural network and the DFT calculated value. MAEs of the energies and the forces were 1.8 meV/atom and 33 meV/Å, respectively. The stress plot shows predictions versus DFT values for each Voigt component of the stress, with average errors for every component below $$ 10^{-5} $$ eV/Å3. DFT: Density functional theory; MAEs: mean absolute errors.

Journal of Materials Informatics
ISSN 2770-372X (Online)
Follow Us

Portico

All published articles are preserved here permanently:

https://www.portico.org/publishers/oae/

Portico

All published articles are preserved here permanently:

https://www.portico.org/publishers/oae/