I am a master’s student at the School of Mathematical Sciences, Peking University. I work on optimization and efficient training of large language models, with a particular interest in optimization for machine learning and machine learning for optimization.
I am advised by Prof. Kun Yuan at CMLR, Peking University. Previously, I received my B.S. in Mathematics and Applied Mathematics from Sichuan University in 2026.
A reference for what optimization algorithms can achieve.
An interactive reference for lower and upper complexity bounds for optimization algorithms, with assumptions, paper references, and expert validation. The current release covers decentralized optimization.
A collaborative agent system for mathematical research, from literature synthesis and theoretical reasoning to experiments.
Randomized subspaces reduce the memory cost of bilevel optimization while retaining convergence guarantees.
A unified view of decentralized Muon, connecting communication design with convergence and structural limitations.
How perturbations propagate through forward and backward passes, and when stochastic gradient descent still converges.
* Equal contribution.
M.Sc. in Big Data
School of Mathematical Sciences
B.S. in Mathematics and Applied Mathematics
School of Mathematics
GPA: 3.90 / 4.0 · Rank: 1 / 40
Mathematical formalization and theorem proving in Lean 4. Outstanding Participant Award and Special Prize.
Working with Prof. Kun Yuan on randomized subspace methods, compositional error propagation, and OptBound.
Research with Prof. Ju Ren at Tsinghua University on optimization formulations and convergence in RLHF.
Course notes on sets, Lebesgue measure, measurable functions, integration, differentiation, and Lp spaces.
Mathematics · Course notes · In Chinese