Deep-learning density functional perturbation theory

He Li,Zechen Tang,Jingheng Fu,Wen-Han Dong,Nianlong Zou,Xiaoxun Gong,Wenhui Duan,Yong Xu

DOI: https://doi.org/10.1103/PhysRevLett.132.096401

2024-01-31

Abstract:Calculating perturbation response properties of materials from first principles provides a vital link between theory and experiment, but is bottlenecked by the high computational cost. Here a general framework is proposed to perform density functional perturbation theory (DFPT) calculations by neural networks, greatly improving the computational efficiency. Automatic differentiation is applied on neural networks, facilitating accurate computation of derivatives. High efficiency and good accuracy of the approach are demonstrated by studying electron-phonon coupling and related physical quantities. This work brings deep-learning density functional theory and DFPT into a unified framework, creating opportunities for developing ab initio artificial intelligence.

Computational Physics,Materials Science

What problem does this paper attempt to address?

This paper proposes a new framework that uses deep learning to accelerate and improve the computation of density functional perturbation theory (DFPT). DFPT is an important method for predicting the response of materials to various perturbations (such as lattice vibrations and strain) from first principles, but it is computationally expensive, limiting high-throughput materials research. In this paper, the researchers utilize neural networks to learn and compute the key quantity - the change in the Kohn-Sham potential under a unit perturbation, and accurately calculate the derivatives of physical quantities through automatic differentiation techniques. This approach demonstrates efficiency and accuracy in the calculation of electron-phonon coupling (EPC) and related physical quantities, and is expected to integrate deep learning with DFT and DFPT into a unified framework, promoting the application of artificial intelligence in materials science and accelerating the discovery of new materials.

Deep-learning density functional perturbation theory

Deep-Learning Density Functional Perturbation Theory

Deep-learning density functional theory Hamiltonian for efficient ab initio electronic-structure calculation

Improving density matrix electronic structure method by deep learning

Neural-network Density Functional Theory Based on Variational Energy Minimization

Deepks: A Comprehensive Data-Driven Approach Toward Chemically Accurate Density Functional Theory

Deep learning density functional theory Hamiltonian in real space

A deep equivariant neural network approach for efficient hybrid density functional calculations

A deep learning framework to emulate density functional theory

Deep learning-based density functionals empower AI for materials

Efficient hybrid density functional calculation by deep learning

Density functionals from deep learning

A Time-Dependent Random State Approach for Large-Scale Density Functional Calculations

Accelerating finite-temperature Kohn-Sham density functional theory with deep neural networks

Many-core Acceleration of the First-Principles All-Electron Quantum Perturbation Calculations.

Pushing the frontiers of density functionals by solving the fractional electron problem

Improving the efficiency of ab initio electronic-structure calculations by deep learning

Deep Neural Network Representation of Density Functional Theory Hamiltonian

Density functional theory plus dynamical mean field theory within the framework of linear combination of numerical atomic orbitals: Formulation and benchmarks