Numerical optimization
First-order and operator splitting methods for linear, quadratic, and convex optimization.
Applied mathematician · Hong Kong
Optimization methods
built for scale.
I design algorithms and GPU implementations for large-scale linear and quadratic programming, connecting rigorous optimization theory with high-performance scientific computing.
01 Theory, implementation, and numerical experiments
01 / Research
From convergence theory to computational scale.
My doctoral research focuses on scalable algorithms for large optimization models, with an emphasis on operator splitting methods and modern GPU architectures.
GPU computing
High-throughput implementations designed for large sparse instances and stringent tolerances.
Scalable solvers
Open software and careful benchmarking that connect theoretical advances to practical performance.
02 / Publications
Selected work
Linear and quadratic programming, splitting methods, and certified robustness.
-
2026
-
2025
On the Relationships among GPU-Accelerated First-Order Methods for Solving Linear Programming
-
2025
-
2025
Peaceman-Rachford Splitting Method Converges Ergodically for Solving Convex Optimization Problems
-
2024
Collective Certified Robustness against Graph Injection Attacks
03 / Software
Research, made executable.
Open-source Julia and CUDA implementations for large-scale optimization experiments.
Linear programmingJulia · GPU
HPR-LP
A GPU solver for large-scale linear programs, pairing a Halpern–Peaceman–Rachford method with adaptive restart and penalty parameter updates.
- Large sparse LP instances
- Large-scale benchmark suite
- High-accuracy numerical experiments
Quadratic programmingJulia · CUDA
HPR-QP
A dual HPR framework for large-scale convex composite quadratic programming, designed around efficient range-space updates and GPU scalability.
- Restricted Wolfe dual formulation
- Adaptive practical enhancements
- Scalable sparse computation
04 / Background
Mathematics with a computational point of view.
2020—2026
Ph.D. in Applied Mathematics
The Hong Kong Polytechnic University
Department of Applied Mathematics2016—2020
B.Sc. in Information and Computing Science
Hunan University
School of MathematicsMethods & domains
LP · QP · operator splitting · sparse linear algebra · GPU computing