Yuxi Hong, Postdoc Researcher @ Berkeley Lab
[yuxihong at lbl.gov] [Google Scholar] [CV]
[yuxihong at lbl.gov] [Google Scholar] [CV]
I'm a postdoc researcher at Performance and Algorithms Research Group at Berkelely Lab. I'm working with Aydin Bluluc on sparse matrix computation. I design scalable and parallel matrix algorithms which lies in the heart of scientific computation and machine learning workload. I exploit the sparsity and structure of scientific data to reduce the computation time, thus accelerate scientific discovery. I am a member of Sparsitute. I have published more than 15 peer-reviewed papers in top-tier journals and conferences, including SC, ISC, IJHPCA. One of my paper is recongnized as a 2023 Gordon Bell Prize finalist.
I will join Intelligent Systems Engineering Department at Indiana University Bloomington as an assistant professor this fall!
I’ll be presenting my latest work at SC24 in Atlanta on Wednesday, November 20, 2024, from 11:00 am to 11:30 am EST, Location B311 at the convention center. [Twitter] [SC24 Session][Paper Link]
One paper is published on IJHPCA. [DOI]
One paper is accepted by SC 24. [Arxiv]
I will attend Sparsitute 2023 all-hands meeting at Chicago, IL.
[Full News]
I am dedicated to advancing high-performance matrix computation algorithms, focusing on a "sparse" perspective to support future scientific research. In scientific computing, it is well known that matrices often exhibit sparse patterns. However, there is significant untapped potential to fully exploit these sparse patterns and achieve better performance through algorithm-application-architecture co-design. My approach is sparsity-aware, as it leverages the inherent sparse structures in the matrices. My research goals are to:
1. Develop new data structure to represent matrices based on the sparsity pattern.
2. Design parallel and scalable sparsity-aware matrix computation algorithms and runtime systems to support high-level algebraic operations.
3. Optimize the performance of sparsity-aware matrix computation algorithms on emerging hardware accelerators.
Yuxi Hong is a postdoctoral research fellow in the Performance and Algorithms group of the Computer Science Department at Lawrence Berkeley National Laboratory. He obtained his Ph.D. in Computer Science at King Abdullah University of Science and Technology (KAUST). He received an MS degree in Electronics Engineering from Tsinghua University and a BSc from Tsinghua University. His current research interests include HPC, Numerical Linear Algebra, GPU programming, sparse computation, low rank methods and efficient Machine Learning/ Deep learning. My PhD advisors are David Keyes, Hatem Ltaief and Matteo Ravasi.