Agentic data science with the Data Agent Kit
cloud.google.com·8h ago
AI Research
arXiv CS.AI·Aug 26·5 min read
Learning the Kohn-Sham map with neural operators for quasi-linear scaling density functional theory
TL;DR
A neural operator has been developed to learn the Kohn-Sham map for orbital-free density functional theory (DFT), achieving quasi-linear scaling in electronic structure calculations. This method allows for accurate predictions of electron densities and structural properties across diverse materials without the need for traditional orbital diagonalization.
✦ Why It Matters
Consider adopting this neural operator approach for large-scale electronic structure calculations in your projects.
Key Takeaways
- Explore the Kohn-Sham map for efficient electronic structure modeling.
- Utilize the SE(3)-equivariant Fourier neural operator for density predictions.
- Implement this method for large systems to reduce computational costs.
- Test the model on diverse materials to validate its generalization capabilities.
How It Works
The method leverages a neural operator that learns the Kohn-Sham map, which directly relates the Kohn-Sham potential to electron density and noninteracting kinetic energy. By focusing on the density component, the SE(3)-equivariant Fourier neural operator processes inputs on real-space grids, allowing for stable and efficient SCF calculations without the need for auxiliary orbitals.