3-T heat conduct equation including electron, ion and photon (radiation) temperatures can be used to approximately describe the energy broadcast across multimedia for radial flow dynamics and discover the energy swapping among photon,electron and ion. Owing to the strong nonlinear diffusion coefficients and energy swapping coefficients and strong discontinuous coefficients across media interfaces,this equation is difficult to be solved with high numerical resolution. Based on the parallel adaptive multigrid software framework UG on 2-D unstructured grid, this paper successfully solved such equation with high resolution by combining the finite volume implicit discretization scheme and parallel adaptive multigrid algorithm, and gained much significant results.
时间相关偏微分方程隐式离散后,通常需要求解一个稀疏线性代数方程组序列.利用序列中相邻方程组性质的差异性与相似性,自适应地选取预条件子,提升方程组序列的并行求解效率,从而缩短总体求解时间,是一个值得研究的问题.本文针对科学与工程计算中广泛使用的代数多重网格(AMG)预条件子,设计了方程组序列相关的自适应预条件策略.通过惯性约束聚变(ICF)的辐射流体力学数值模拟典型应用,验证了该策略的有效性.测试结果表明,在某高性能计算机的3125个CPU核上,自适应预条件策略可将并行效率从47%提升到61%,将模拟总时间从19.7 h降为14.5 h.
