Zhichao Jia

I am a PhD student at Operations Research, Georgia Institute of Technology, advised by Guanghui (George) Lan. Previously, I received my master's degree from Applied Mathematics and Statistics, Johns Hopkins University, where my master's thesis was advised by Benjamin Grimmer.

Email  /  CV  /  LinkedIn  /  Google Scholar

profile photo
Education

PhD student in Operations Research at Georgia Institute of Technology (2023.08-present).

Master of Science in Engineering in Applied Mathematics and Statistics at Johns Hopkins University (2021.08-2022.12) [Thesis].

Bachelor of Science in Information and Computing Science at Sun Yat-Sen University (2017.09-2021.06) [Thesis].

Research Interests

I am interested in optimization. My current research is about continuous optimization, including convex/nonconvex, smooth/nonsmooth, unconstrained/constrained and deterministic/stochastic optimization, focusing on design and analysis of first-order/zeroth-order algorithms, with applications in solving modern data science and machine learning problems. My previous research also includes exploring multiple applications of numerical methods.

Academic Work
First-Order Methods for Lipschitz Constrained Optimization [ArXiv]
Take a look at the Goldstein stationarity for constrained problems!
Advisor: Benjamin Grimmer
First-Order Methods for Nonsmooth Weakly Convex Constrained Optimization [ArXiv]
More than KKT stationarity, let us see Fritz John! Look at our SCAD objects with interesting non-Slater properties, when acting as nonsmooth nonconvex functional constraints. Also see our non-Lipschitz analysis for the switching subgradient method.
Advisor: Benjamin Grimmer
The Zeroth-Order SPSA method for Constrained Stochastic Optimization [ArXiv]
See Simultaneous Perturbation Stochastic Approximation (SPSA) here! Take a look at our designs on the SPSA dealing with functional inequality constraints.
Advisor: James Spall
Teammate: Ziyi Wei
Stochastic First-Order Methods with Stochastic Polyak Stepsizes
Coming soon!
Advisor: Nicolas Loizou
On a Tailored Fast and Flexible Holomorphic Embedding Method for Time-Varying Control Systems
Advisor: Tao Wang
Teammates: Yuning Ding, Zi'ang Fang, Yusi Zhang
Implementation and Analyzing of the Augmented Lagrangian Method
Advisor: Jovan Ilic
Teammates: Zhaoxuan Hu, Jiagan Ma
Patent

Mechanical Arm Motion Planning Method based on Fast and Flexible Full-Pure Embedding Thought [Website]
with Tao Wang, Yuning Ding, Zi'ang Fang    China Patent 202110154750.4    Issued 02/04/2021