The article “Unsupervised Representation Learning of Kohn–Sham States and Consequences for Downstream Predictions of Many-Body Effects” presents a novel approach using variational autoencoders (VAEs) to represent complex electronic structures in a compressed, low-dimensional space. By encoding Kohn-Sham wavefunctions from density functional theory (DFT) into a latent space, the model captures essential features without manual feature engineering. This latent representation is then used in supervised neural networks to predict quasiparticle band structures with high accuracy under the GW approximation. The study demonstrates that VAEs can efficiently capture the underlying physics of electronic states, allowing for accurate band structure predictions across various materials, thus facilitating advancements in electronic structure calculations and potentially aiding in materials discovery.
For more details, please continue reading the full article under the following link:
https://www.nature.com/articles/s41467-024-53748-7
Please consult also the Quantum Server Marketplace platform for the outsourcing of computational science R&D projects to external expert consultants through remote collaborations:
#materials #materialsscience #materialsengineering #computationalchemistry #modelling #chemistry #researchanddevelopment #research #MaterialsSquare #ComputationalChemistry #Tutorial #DFT #simulationsoftware #simulation