Skip to main content

Research

We develop and apply computational tools to tackle challenging problems in chemistry and materials science. Our research is highly interdisciplinary, integrating quantum chemistry, machine learning, condensed matter theory, and quantum information. We apply our computational frameworks to discover next-generation renewable energy solutions, addressing the energy crisis and environmental challenges. We also design quantum materials with exotic physical properties, aiming to lay the foundation for the next technological revolution.

Systems We Study

Electronic Structure

Electronic structure encodes the microscopic origins of the chemical behavior of a system. Our group develops and applies quantum chemistry methods to accurately simulate electronic structure, with a particular focus on strongly correlated systems. Strong electron correlation gives rise to exotic quantum phases and phenomena, such as the Kondo effect and unconventional superconductivity. Accurate electronic structure simulations are essential for understanding the underlying mechanisms behind these behaviors.

Quantum Materials

Quantum information is a transformative technology with the potential to reshape our future. However, major challenges remain, including limited conceptual understanding and the lack of suitable materials for robust quantum computing. Our mission is to design materials with exotic quantum properties—such as topologically protected orders, fracton phases, and many-body localization—to accelerate progress in quantum information science and technology.

New Energy Solution

Sustainable and low-cost energy resources are essential to support the rapid growth of modern technology. As a computational research group, we focus on identifying efficient solutions to meet the increasing computational demands of the HPC and GPU computing era. We use the quantum chemistry and AI platforms developed in our group to accelerate the discovery of new energy materials.

Methods We Develop & Use

Quantum Embedding

Quantum embedding methods are promising tools for scalable chemical simulations. The core idea is to apply different levels of theory to different parts of a system. In materials science, approaches like density matrix embedding theory (DMET, shown in the figure) and dynamical mean-field theory (DMFT) are widely used for their effective treatment of electron correlations. Our group develops embedding methods based on these frameworks to achieve accurate simulations of large solid-state systems.

ML-accelerated quantum chemistry

Quantum chemistry constantly balances accuracy and computational cost. Machine learning offers two powerful solutions. First, pretrained neural networks can replace the most time-consuming components of quantum chemistry methods, significantly reducing cost while preserving accuracy across diverse systems. Second, the expressiveness of neural networks makes them ideal for representing quantum states—so-called neural network quantum states (NNQS)—which map fundamental system information, such as nuclear coordinates, directly to wavefunctions. Our group advances simulation methods in both directions to accelerate progress in chemical research.

Generative AI

Chemistry is the science of creating new substances. However, traditional trial-and-error methods are too slow to keep pace with the vastness of chemical space. Recent advances in data science and artificial intelligence offer a path toward systematic and efficient exploration. Our group develops and applies AI tools, guided by quantum chemistry, to accelerate the discovery of novel functional materials.