Publications by authors named "Honggeng Tao"
Article Synopsis
- Deep-learning density functional theory (DFT) has the potential to speed up material discovery and change how materials research is conducted.
- Current approaches in this area mainly depend on supervised learning, which keeps neural networks and DFT developments separate.
- This research introduces a new theoretical framework that integrates neural network optimization with DFT computations, allowing for more efficient, physics-informed unsupervised learning, and demonstrates its effectiveness through new algorithms in a differential DFT code.
Universal materials model of deep-learning density functional theory Hamiltonian.
Sci Bull (Beijing)
August 2024
Realizing large materials models has emerged as a critical endeavor for materials research in the new era of artificial intelligence, but how to achieve this fantastic and challenging objective remains elusive. Here, we propose a feasible pathway to address this paramount pursuit by developing universal materials models of deep-learning density functional theory Hamiltonian (DeepH), enabling computational modeling of the complicated structure-property relationship of materials in general. By constructing a large materials database and substantially improving the DeepH method, we obtain a universal materials model of DeepH capable of handling diverse elemental compositions and material structures, achieving remarkable accuracy in predicting material properties.
View Article and Find Full Text PDFNonreciprocity is important in both optical information processing and topological photonics studies. Conventional principles for realizing nonreciprocity rely on magnetic fields, spatiotemporal modulation, or nonlinearity. Here we propose a generic principle for generating nonreciprocity by taking advantage of energy loss, which is usually regarded as harmful.
View Article and Find Full Text PDF