Contributed as second author and co-first author, under the supervision of Prof. Weiyi Shang (UWaterloo), Prof. Jinfu Chen (WHU), and Zishuo Ding (HKUST(GZ)), with two papers submitted to a CCF-A conference.
Developed novel methods and LLVM pass-based analysis for detecting numerical errors, focusing on improving the efficiency and accuracy of error detection processes.
Our approach showed strong alignment (correlation over 0.8) with high-precision programs and required only about 1/1000 of the time needed by high-precision programs.