About Me

Hi, I'm Jiaqing! I am a first-year Computer Science Ph.D. student at Fudan University. I received my M.Sc. in Computer Science from ETH Zurich in 2025, and my B.Eng. in Electronics and Computer Science from the University of Edinburgh in 2022.

My research focuses on theoretical analysis and applications of graph neural networks and flow matching, especially for fine-grained molecular modeling. You can see my papers on Google Scholar or download my CV.

If you want to reach me, email me at godxie.steven@gmail.com.

News

• [2026/07] We released an efficient PE calculation method for directed graphs.
• [2026/01] We released a theoretic and experimental review of PE on graphs with GNN [KDD'2026]

Education & Experience

• Shanghai AI Lab — Assistant Researcher, AI for Science Center (2025 — Present)
• Fudan University — Ph.D., Computer Science (Sept 2026 - Current)
• ETH Zurich — M.Sc., Computer Science (2022 - 2025)
• University of Edinburgh — B.Eng., Electronics and Computer Science (2019 - 2022)

Publications 3🥇 1🥈 0🥉

^‡ indicates equal contribution.

Graph Neural Networks

• Gauge-Invariant Learnable Spectral Positional Encodings for Directed Graphs via Hermitian Block Krylov Subspaces
  Jiaqing Xie, Yuxin Wang
  Arxiv
  Paper · Code
• GeoBA: Geometry-Biased Attention for Molecular Property Prediction
  Jiaqing Xie, Ziheng Chi, Zhuo Yang
  Under Review
  Paper · Code
• Benchmarking Positional Encodings for GNNs and Graph Transformers
  Florian Grotschla^‡, Jiaqing Xie^‡ and Roger Wattenhofer
  SIGKDD 2026
  Paper · Code

Flow Matching

• SpectraFlow: Physics-guided Flow Matching for IR-Raman Spectral Interconversion
  Jiaqing Xie, Zhuo Yang, Ben Gao, Shuaike Shen, Tianfan Fu, Yuqiang Li
  To be released
  Paper · Code

Molecular Dynamics

• More coming soon.

Personal Info

• Physical: height 183 cm, weight 80 kg.
• Running (PB): 50 m 6.5 s; 1000 m 3′22″; 5000 m 27′30″; 10000 m 55′30″.
• Weiqi (Go): Chinese Weiqi Association amateur 1 dan (2008).
• Singing: 300+ fans.
• Zhihu: 2800 followers — P等于NP.
• Codeforces: 1200.

Resources

• Marx — PDF

Academic Services (Reviewer)

• 2025: ACL, CVPR, EMNLP, ICONIP, NeurIPS
• 2024: ACML, NeurIPS