Education: 

2006-2010    B.S., School of Physics, Shandong University

2010-2015    Ph.D, College of Chemistry and Molecular Engineering, Peking University

Employment: 

2016-2017    Postdoc, Hong Kong Baptist University (Adviser: Michel. A. Van Hove)

2018-2020    Postdoc, Duke University (Adviser: Volker Blum)

2021-            Associate Professor (Principal Investigator), School of Physics, Beihang University

We are an electronic structure theory research group within the School of Physics at Beihang University (BUAA), Beijing. Our research focuses on condensed-matter electronic structure theory, including theoretical method development, large-scale first-principles code development, and computational materials applications. By combining electronic structure theory, high-performance computing, and artificial intelligence (AI), we aim to develop advanced computational approaches for materials research.

Current Research Directions:

1. Large-Scale DFT Code Development (ASPIRES and FHI-aims)

We have been consistently developing first-principles simulation codes for accurate and efficient electronic structure calculations of molecules and solids. As core developers of the FHI-aims code package, an all-electron density functional theory (DFT) framework based on numeric atom-centered orbitals (NAOs), we are responsible for the development and maintenance of its relativistic electronic structure methods and related computational functionalities.

Our ongoing in-house development is the ASPIRES (Ab-initio Simulation Platform for Intelligent Research on Electronic Structures) code package, a next-generation all-electron DFT framework specifically designed for intelligent and automated electronic structure simulations. Built upon the Python and PyTorch ecosystems, ASPIRES natively combines high-performance scientific computing with machine learning (ML) technologies, aiming to provide an open-source platform for the development of AI-oriented electronic structure algorithms and the inverse design of materials.

2. AI for Electronic Structure and Materials Modeling

We explore the integration of AI with electronic structure theory to accelerate first-principles simulations and materials design. By combining ML, automatic differentiation, graph neural networks (GNNs), and electronic structure theory, our research interest focuses on two major directions:

(i) AI-accelerated electronic structure methods. Ongoing efforts include the development of ML-based NAO basis functions and GNN models for electron density prediction. These methods aim to significantly reduce the computational cost of electronic structure calculations at the self-consistent field (SCF) level while maintaining physical fidelity.

(ii) ML for materials property prediction and screening. We have developed interpretable ML models for predicting elastic and mechanical properties of inorganic materials, particularly ceramics and amorphous systems. Recent works combine stacked ensemble learning, symbolic regression, and deep neural networks, achieving high accuracy with physically meaningful descriptors. Applications include rare-earth-doped silicate ceramics, and high-throughput screening of mechanical materials. We also explore the use of convolutional neural networks (CNNs) trained on large datasets to uncover how structural features, such as bond lengths and angles, influence the spin-orbit splitting magnitude of frontier bands in 2D systems.

3. Relativistic DFT Methods Based on the Four-Component Dirac Equation

Materials containing heavy elements exhibit numerous novel properties, placing them at the forefront of physics and chemistry. Compared to light elements, heavy atoms in the lower rows of the periodic table have much more complex electronic structures, leading to a long-standing challenge for ab initio calculations. This complexity often arises from two aspects:

(i) Relativistic effects, which conceptually include scalar relativistic (SR) effects and spin-orbit coupling (SOC) contributions; (ii) Strong correlation effects, usually associated with the electron localization of unfilled d/f shells of transition metals and lanthanides/actinides.

In previous studies, we have developed a quasi-four-component (Q4C) relativistic band theory for extended systems (with hundreds of atoms in a unit cell). This approach enables accurate fully-relativistic all-electron calculations for electronic structures with much cheaper computational costs. Building on this framework, our group is currently working on:

(i) Fully-relativistic DFT+U methods to capture the interplay between electron correlation and relativistic effects; (ii) non-collinear Q4C formalism, aiming to treat materials with open-shell electrons and magnetic properties. (iii) finite-nucleus Q4C methods, extending relativistic theory to accurately describe near-nucleus electronic structures and emerging applications such as nuclear spectroscopy and nuclear clock studies.

4. First-Principles and Machine Learning Studies of Materials

We apply our theoretical frameworks and software tools to investigate a broad range of functionally interesting materials. These covers various material systems, such as rare-earth and actinide compounds, perovskites, π-electron and carbon-based systems, structural ceramics, and amorphous materials. Particular interests include the roles of SOC and relativistic effects in quantum materials, structure-property relationships in functional materials, and the application of ab initio simulations and ML approaches for materials discovery and design.

Openings

We typically recruit 1-2 undergraduate researchers, 1-2 Mphil students (with intent to pursue PhD), and 1-2 PhD students per year.

Please note that there are no openings for CSC international students at the Master's degree level. Therefore, all inquiries will not receive a response—apologies!

For postdoc or visiting scholar positions, please contact the PI for more details. We welcome postdocs with backgrounds in either ① first-principles code development (FORTRAN and MPI skills required) or ② AI for electronic structure methods. The salary starts at 350,000 CNY/Y (~50,000 USD/Y) and can reach as high as 700,000 CNY/Y (~100,000 USD/Y) based on experience, with all insurances covered by the institution.

Contact: rdzhao@buaa.edu.cn

Group Members:

Han Wang (PhD student, 2023-)

Yanmiao Han (PhD student, 2024-)

Zheng Pan (MPhil&PhD student, 2022-)

Wenhao Li (MPhil&PhD student, 2022-)

Xinyi Tan (MPhil&PhD student, 2023-)

Zhaoyang Zhang (MPhil&PhD student, 2024-; undergraduate 2020-2024)

Huanhuan Cao (MPhil&PhD student, 2024-)

Deyang Liang (MPhil student, 2025-; undergraduate 2021-2025)

Lishi Zheng (undergraduate 2023)

Zhenxu Luo (undergraduate 2023)

Luokuan Feng (undergraduate 2023)

Xinyan Dai (undergraduate 2024)

Wentao Wang (undergraduate 2024)

Yunxiang Guo (undergraduate 2024)

Alumni: Jiaqi Lu (undergraduate 2018),  Peilei Zuo (undergraduate 2021), Shengri Liu (undergraduate 2021), Siyuan Zhang (undergraduate 2022)

Group Website (temporary...)