GAMES Webinar 2019 – 87期（物理模拟专题课程） | 何小伟（中科院软件所），王鑫磊（浙江大学）
【GAMES Webinar 2019-87期】
报告题目：A Tutorial on How to Simulate Versatile Physical Materials with Particle-based Methods
Particle-based methods have become an important modeling tool for computer graphics. Compared to other methods, particle-based methods are easy to implement, but powerful enough to create a variety of dynamic scenarios, thus has been widely used in both films and video games. In this talk, I will describe the main techniques and fundamental theories necessary to learn and simulate physical materials with particle-based methods. I will show how to simulate versatile materials ranging from incompressible fluids to elastoplastic solids and granular matters. I will conclude my presentation by outlining the challenges in developing high-fidelity particle-based methods.
Xiaowei He is currently an associate professor at the Institute of Software, Chinese Academy of Sciences. He received both his BS and MS degrees from Peking University, and his Ph.D. from Institute of Software, Chinese Academy of Sciences. His research interests are mainly focused on computer graphics, computational physics, smoothed particle hydrodynamics, peridynamics and nolocal theory. He has published several papers in international journals/conferences including TOG, TVCG, SCA, etc. He is currently doing research on how to apply machine learning to help improve both the performance and accuracy over traditional numerical solvers.
报告题目：GPU Optimizations of Material Point Method and Collision Detection
The material point method has been a research focus in computer graphics for some time. It has the advantages of handling complex topology changes, multi-material and multiphase coupling, (self) collisions, etc., in an elegant way. However, its generalization is usually hindered by time step restrictions and heavy computation workloads. In this talk, we start by designing an efficient pipeline of the material point method for the GPU. Several major technical challenges will be discussed, and an optimized parallel implementation will be presented. Additionally, more GPU optimization patterns and strategies will be briefly discussed surrounding the broad-phase collision detection.
Xinlei Wang is a fourth-year Ph.D. student at Zhejiang University, advised by Prof. Min Tang. He obtained his bachelor degree in Huazhong University of Science and Technology (HUST), and is currently visiting University of Pennsylvania under the supervision of Prof. Chenfanfu Jiang. His research interests lie in computer graphics including physics-based animations, and high performance parallel computing.
GAMES主页的“使用教程”中有 “如何观看GAMES Webinar直播？”及“如何加入GAMES微信群？”的信息；