八叉树
有限元法
多重网格法
网格
计算机科学
解算器
拓扑(电路)
自由度(物理和化学)
嵌入
自适应网格优化
边界(拓扑)
分辨率(逻辑)
网格生成
曲面(拓扑)
计算科学
算法
几何学
数学
人工智能
数学分析
偏微分方程
物理
热力学
程序设计语言
组合数学
量子力学
作者
Jun Wu,Christian Dick,Rüdiger Westermann
标识
DOI:10.2312/pe/vriphys/vriphys11/029-038
摘要
Recent work has demonstrated that composite finite-elements provide an effective means for physically based modeling of deformable bodies. In this paper we present a number of highly effective improvements of previous work to allow for a high-performance and high-quality simulation of boundary surfaces of deformable bodies with changing topology, for instance, due to cuts and incisions. Starting at a coarse resolution simulation grid, along a cut we perform an adaptive octree refinement of this grid down to a desired resolution and iteratively pull the fine level finite-element equations to the coarse level. In this way, the fine level dynamics can be approximated with a small number of degrees of freedom at the coarse level. By embedding the hierarchical adaptive composite finite-element scheme into a geometric multigrid solver, and by exploiting the fact that during cutting only a small number of cells are modified in each time step, high update rates can be achieved for high resolution surfaces at very good approximation quality. To construct a high quality surface that is accurately aligned with a cut, we employ the dual-contouring approach on the fine resolution level, and we instantly bind the constructed triangle mesh to the coarse grid via geometric constrains.
科研通智能强力驱动
Strongly Powered by AbleSci AI