Research Assistant
Current
University Of Kentucky, Department Of Physics
Calculate hadron structure with lattice QCD.High performance computing• Programmed million-by-million matrices operations across 1000+ CPU cores on top 10 super computers • Used C and C++, with MPI and OpenMP.• Adopted multiple numerical techniques such as Arnoldi iteration and conjugate gradient method.• Developed advanced debug skill and “sense of troubleshooting”.Statistical modeling of large data• Used C++, Python and Scheme.• Results helped PI applying for major funding (220k+) and computing resources (~100 million hours).• Applied statistical method such as jackknife and correlated 𝜒2 fitting.• Developed exotic fitting algorithms to solve a major puzzle in the field. (We were the first and only group that have a solution to the said puzzle.)Visualization of large data• Created data visualization framework to quickly compare or to get a sense of the data. • Results helped PI applying for major funding and computing resources.• Developed visualization framework to let data go through complex processing pipelines. Dozens of figures could be computed and generated with 100 lines of code, harnessing the power of super computers or consumer multi-core CPUs.Helped create new Scheme data processing frameworkCreated Python data processing frameworkMisc.• Wrote a program to automatically extract data from plot images. Results used in a research paper.• Wrote a web spider to scrap data from favorite social network website.• Wrote a Mac app to bind any application to any global hotkey, to control a Mac by using only the keyboard.• Wrote an app to import BibTeX data into Microsoft Word’s citation database.Teaching• Taught undergraduate-level physics courses for 4 years with very positive student evaluations (“I would take his section again”, “very helpful and gave thorough explanations”).• Acted as “super-TA”, and helped training new teaching assistants for the department.