Speaker:Tan Tiow Seng, National University of Singapore

Date:Sept. 3, 2019

Time:2:00pm-3:30pm

Location:Room 310, Third Floor, Office Building, Software Campus

Abstract:

This talk is about the use of parallel computational power of GPU to speedup geometric computation. In particular, computation of quality triangulations that are often used in various engineering and scientific applications such as finite element methods, interpolation etc. We show how one should better utilize the GPU for solving computational geometry problems, like generating quality triangulations that are not known to present themselves readily for a GPU computation. Lessons learnt here may be of independent interest for one to study other seemingly non-parallelizable problems.

Bibliography:

Chen ZH and Tan TS, “Computing Three-dimensional Constrained Delaunay Refinement Using the GPU”, the 28th International Conference on Parallel Architectures and Compilation Techniques, Seattle, WA, USA, 2019.

Bio:

Dr Tan Tiow Seng has been the Director of International Education Programs, School of Computing, National University of Singapore since 2018. Before that, he has been the Deputy Director (Education) of NUS (Suzhou) Research Institute for 7 years. He obtained his PhD degree from the University of Illinois, Urbana-Champaign, USA. Dr Tan specializes in algorithm designs, in particular, geometric algorithms, with applications to interactive graphics, visualization and games. His research group has released a number of GPU software with source codes on geometric computation for free download. These are the only ones available that are robust, accurate and efficient till date. Dr Tan holds fives US/Singapore patents, and is a two-time recipient of the NUS Teaching Excellence Award.

For more information, please visit:

http://www.view.sdu.edu.cn/info/1020/122148.htm

Edited by:Su Chang