My research focuses on AI for Science, Scientific Intelligence, neutrino physics, and intelligent research systems. I am interested in how artificial intelligence can accelerate scientific discovery, and how large-scale scientific experiments can, in turn, contribute to the development of next-generation intelligent systems.

AI for Science and Scientific Intelligence

I investigate the application of machine learning, deep learning, and agent-based systems to fundamental science, including scientific agents, research workflow automation, scientific knowledge discovery, agent harnesses, and reliable intelligent systems.

Research Agents and Intelligent Scientific Workflows

I explore agent systems designed for scientific research, focusing on hypothesis generation, experiment design, evidence verification, knowledge accumulation, and research workflow automation. My long-term goal is to build next-generation research agents that can support scientific discovery.

Neutrino Physics and Particle Astrophysics

I participate in major international neutrino experiments, including Borexino and JUNO, studying solar neutrinos, reactor neutrino oscillations, and the fundamental properties of neutrinos.

Large-Scale Scientific Computing and Statistical Inference

My research includes large-scale data analysis, high-performance computing, GPU acceleration, statistical inference, and scientific data processing methods, providing reliable computational infrastructure for complex scientific problems.

Selected Publications and Contributions

(1) JUNO Solar Oscillation Measurement (under review, 2026)

First high-precision measurement of solar neutrino oscillations with JUNO.

Personal Contributions

• Co-convener of the statistical analysis effort
• Led the design and validation of the analysis framework
• Developed a semi-model-independent analysis approach
• Led systematic uncertainty evaluation and ToyMC validation studies
• Contributed key statistical inference methodologies and robustness studies
• Helped drive the analysis from concept validation to the final physics result

(2) Experimental Evidence of Neutrinos Produced in the CNO Fusion Cycle in the Sun

Nature, 2020

One of the world's major scientific breakthroughs of 2020.

First experimental evidence of solar neutrinos produced by the CNO fusion cycle.

Personal Contributions

• Independently proposed and implemented key statistical analysis techniques
• Developed new background suppression and signal extraction methods
• Improved signal significance by approximately three orders of magnitude (~1000×)
• Provided critical technical support for the final CNO neutrino discovery
• Established analysis methods that became the foundation for subsequent studies

(3) Comprehensive Measurement of pp-Chain Solar Neutrinos

Nature, 2018

First comprehensive precision measurement of solar pp-chain neutrinos.

Personal Contributions

• Independently designed and implemented a GPU-accelerated statistical fitting framework
• Improved large-scale likelihood fitting performance by approximately three orders of magnitude (~1000×)
• Enabled extensive statistical studies and systematic uncertainty evaluations
• Provided critical computational infrastructure for the final physics results
• Resulted in a dedicated methodological publication in JINST and was subsequently adopted in later analyses

Principal Investigator of multiple national research projects, leading international teams in neutrino physics, scientific computing, and AI for Science, with cumulative research funding exceeding RMB 10 million.

现指导学生

薛景秦  硕士研究生  070202-粒子物理与原子核物理  

范良前进  硕士研究生  070202-粒子物理与原子核物理  

李典  硕士研究生  070202-粒子物理与原子核物理