Summary
BrightEST:
I focused on high-performance computing and large-scale C++ system optimization for computational geometry (EDA). I designed and implemented an intra-node parallel algorithm for polygon boolean operations using a quadtree-based approach, reducing wall-time from 4.13 seconds to 0.35 seconds on 32 threads, achieving an approximately 11.8× speedup with 36% parallel efficiency. I also implemented the Hopkins method for computational lithography using OpenMP to further leverage multi-core architectures.
KLA:
I designed a sweep-line–based algorithm combined with a self-balancing interval tree and a Red-Black tree forest to efficiently count overlapping polygons in design files. The solution achieved an optimal time complexity of ~ O(N(k+logN)), where N is the total number of rectangles and k is the number of reported polygons, significantly improving scalability for large layouts. On the image processing side, I implemented a CUDA-based bilateral filter for noise reduction, leveraging shared memory to improve performance. I also implemented a C++ algorithm for detecting large-scale defects at the micrometer level, successfully addressing a critical customer escalation from Samsung. Additionally, I enhanced image stacking throughput using CUDA, enabling effective noise reduction.
PhD:
I am a contributor to the GAMER, a hybrid GPU/OpenMP/MPI parallelization finite-volume program solving hydrodynamic equations. Specifically, I designed a new scheme for conversion between primitive and conserved variables in the special relativistic hydro solver. The code adopting this scheme can handle critical problems involving a Lorentz factor as high as 10^6 and optimally avoid catastrophic cancellation, especially in non- and ultra-relativistic limits. The results have been published in MNRAS 2021 Vol. 504, pp. 3298-3315. If that wasn't enough, I have designed another new numerical scheme that adaptively and locally reduces the so-called min-mod coefficient in the finite-volume method to promote numerical stability. The new approach was soon merged by the research teams headed by Prof. Hsi-Yu Schive and Prof. Kuo-Chun Pan, who served at National Taiwan and Tsing-Hua University, respectively.
Read List
2026
- C++17: The Complete Guide, Nicolai M. Josuttis
2025
- C++ Move Semantics: The Complete Guide, Nicolai M. Josuttis (study note)
- Inside The C++ Object Model, Stanley B. Lippman
2024
- C++ Primer (3rd ed.), Lippman & Lajoie (study note)
Notes
Sr. C++ SW at BrightEST Corp., Taiwan
Ph.D. in Computational Astrophysics at Taiwan University