Zhenyu Yan

I am a fifth-year Ph.D. candidate in the Programming Languages Lab at Peking University, advised by Prof. Xin Zhang. My research develops precise, scalable, and principled program analysis techniques by strengthening their theoretical foundations and integrating learning-based approaches where they can improve analysis precision, scalability, or ease of implementation. I am seeking postdoctoral or research positions in programming languages, static analysis, and verification.

Selected Work

Scaling Abstraction Refinement for Program Analyses in Datalog using Graph Neural Networks
OOPSLA 2024. I developed a learning-guided approach that combines graph neural networks with constraint solvers to scale abstraction refinement for Datalog-based program analyses by pruning unhelpful abstraction parameters before constraint solving.

Statement-level context sensitivity
Ongoing work. I am developing a framework to support a finer-grained way to choose which statements of a program should be analyzed context-sensitively, with the goal of improving the trade-off between precision and efficiency.

Research Interests

My research mainly focuses on precision-scalability trade-offs in program analysis, especially theoretical extensions to static analysis and systematic uses of machine learning or large language models in program analysis. I am also interested in methods where learned components help guide analysis design choices while the analysis remains grounded in programming-languages semantics and formal reasoning. For a more detailed description, see my Research page.

Education

• Ph.D. in Computer Software, Peking University 2021.9 ~ expected 2027
• B.S. in Computer Software and Technology, Nanjing University 2017.9 ~ 2021

Service

SAS 2025, Artifact Evaluation Committee reviewer

For more informal personal information about myself, please refer to Personal.