陈天宇
Tianyu Chen
北京航空航天大学 · 计算机学院
助理研究员 / 卓越百人博士后
博士导师:李建欣 教授
合作导师:郑志明 院士
School of Computer Science and Engineering
Beihang University
Assistant Researcher / Postdoctoral Fellow
PhD Advisor: Prof. Jianxin Li
Co-advisor: Acad. Zhiming Zheng
Tianyu Chen
研究方向为科学智算(AI for Science),关注如何用机器学习方法求解偏微分方程(PDE),探索物理机理嵌入的基础模型构建方法。目前聚焦于神经算子、科学计算基础模型的理论与工程实践。同时负责国家科学智算共性技术平台 CNAI4S 的建设与运维,服务用户超 19 万。
My research focuses on AI for Science (AI4S), specifically on solving partial differential equations (PDEs) with machine learning and building physics-informed foundation models. I work on neural operators, scientific computing foundation models, and lead the development of CNAI4S, a national platform for AI-driven scientific computing serving 190K+ users.
智算星座计划The Constellation Program
《尚书》:璇玑玉衡,以齐七政 璇玑玉衡,以齐七政Book of Documents: observe Xuan, Ji and Yuheng to set the Seven Regulators in order.
to Polaris Alkaid
摇光
Mizar
开阳
玉衡Alioth Megrez
天权
天玑Phecda 天璇OmniArch Dubhe
天枢
科学智算:新一代计算型人工智能
AI-enabled Scientific Computing: the next generation of Computation-Native AI
学答案(基座模型)learn the answer (foundation model)
智算天璇 OmniArch
学习方程通用解,
规模效应扩展,突破数值方法极限。
Learns general solutions, scaling up past the limits of numerical methods.
学方法(混合求解)learn the method (hybrid solving)
智算玉衡 Alioth
学习求解方法,
数值混合求解,精度对标工业。
Learns solution methods, hybrid with numerics, industrial-grade accuracy.
学判断(关联发现)learn the judgment (association discovery)
智算天玑 Phecda
学习决策判断,
自主闭环引擎,最优配置推荐。
Learns to decide, a self-evolving engine recommending the best configuration.
AI-Native 求解器 = 学答案(基座模型) + 学方法(混合求解) + 学判断(关联发现)
AI-Native solver = learn the answer (foundation model) + learn the method (hybrid solving) + learn the judgment (association discovery)
— 2026 —
— 2025 —
— 2022 – 2023 —
— 2021 —