Xinyu Lu

Xinyu Lu

逯新宇

Molecular AI · Spectroscopy · Geometric Learning

I build AI systems that infer molecular structure and physics from spectra.

My research combines vibrational spectroscopy, molecular generation, foundation models, and equivariant deep learning to help computers interpret scientific observations and accelerate molecular discovery.

I am currently a Ph.D. student at Xiamen University and Shanghai Innovation Institute, jointly advised by Prof. Bin Ren and Prof. Tong Zhu.

I welcome collaborations at the intersection of molecular AI, spectroscopy, and geometric learning. The best way to reach me is by email.

Selected publications

see all →
Published Co-first author

Vib2Conf: AI-driven discrimination of molecular conformations from vibrational spectra

Xinyu Lu*, De-Yi Lin*, Hao Ma, Tong Zhu, Bin Ren, Guo-Kun Liu
Analytical Chemistry • 2026
Abstract
An AI framework for distinguishing molecular conformations directly from vibrational spectra.
Code
Vibrational Spectroscopy Molecular Conformation Deep Learning
Vib2Conf graphical abstract showing spectral and molecular encoders for conformation discrimination
Preprint

Suiren-1.0 Technical Report: A Family of Molecular Foundation Models

Junyi An, Xinyu Lu, Yun-Fei Shi, Li-Cheng Xu, Nannan Zhang, Chao Qu, , Fenglei Cao
arXiv preprint arXiv:2603.21942 • 2026
Abstract
A family of molecular foundation models bridging three-dimensional conformational geometry and two-dimensional statistical ensemble spaces.
Models
Molecular Foundation Models Equivariant Learning AI for Science
Suiren graphical abstract illustrating molecular representations conformational distributions and ensemble properties
Preprint First author

Vib2Mol: From Vibrational Spectra to Molecular Structures—A Versatile Deep Learning Model

Xinyu Lu, Hao Ma, Hui Li, Jia Li, Tong Zhu, Guokun Liu, Bin Ren
arXiv preprint arXiv:2503.07014 • 2025
Abstract
A unified framework connecting spectrum–structure retrieval and de novo molecular generation across Raman and infrared spectroscopy.
Models
Spectrum-to-Structure Molecular Generation Vibrational Spectroscopy
Vib2Mol graphical abstract showing spectrum retrieval and molecular generation tasks
Preprint

Equivariant Spherical Transformer for Efficient Molecular Modeling

Junyi An, Xinyu Lu, Chao Qu, Yunfei Shi, Peijia Lin, Qianwei Tang, , Yuan Qi
arXiv preprint arXiv:2505.23086 • 2025
Abstract
An expressive and efficient SE(3)-equivariant Transformer for molecular property prediction and atomistic modeling.
Equivariant Learning Molecular Modeling Transformer
EST graphical abstract comparing spherical representations tensor products and the proposed Transformer operation

Career highlights

see CV →