What question did this study set out to answer?

This research aims to develop a machine learning method for predicting solid-state electronic properties more efficiently than traditional calculations.

May 21, 2014Physical Review BOpen Access

How to represent crystal structures for machine learning: Towards fast prediction of electronic properties

View Full Paper

Ask AI

Bookmark

Authors

KSKristof T. SchüttPfizer (Germany)HGHenning GlaweMax Planck Institute for the Structure and Dynamics of Matter FBFelix BrockherdeMax Planck Institute of Microstructure Physics

Discussion

Member takes

Implication

Randomized trial explores novel crystal representations to enhance electronic property predictions in solids, indicating improved accuracy.

Key Points

This research aims to develop a machine learning method for predicting solid-state electronic properties more efficiently than traditional calculations.
High-throughput density functional calculations were performed as a training set.
A novel crystal structure representation was designed to improve learning accuracy.
Predictions focused on the density of electronic states at the Fermi energy.
Traditional data representations like the Coulomb matrix were found inadequate for periodic solids.
The proposed crystal structure representation significantly improved prediction accuracy.
Competitive prediction accuracies were achieved within an unrestricted class of spd systems.

Cite This Study

Schütt et al. (2014) studied this question.

synapsesocial.com/papers/6a085b77280cd4e998e8b873 https://doi.org/10.1103/physrevb.89.205118

View Full Paper

Ask AI

Bookmark

Implication

Randomized trial explores novel crystal representations to enhance electronic property predictions in solids, indicating improved accuracy.

How to represent crystal structures for machine learning: Towards fast prediction of electronic properties

Discussion

Member takes

Implication

Key Points

Cite This Study

How to represent crystal structures for machine learning: Towards fast prediction of electronic properties

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider