一类多线程并行四面体网格优化算法

提出了一类多线程并行、组合了光滑化与局部重连操作的四面体网格优化算法。采用传统的基于拓扑的数据分解策略实现了并行光滑化算法,利用图染色算法将待光滑化的点分解成多个独立点集。同时提出将一类基于几何的数据分解策略应用于局部重连操作的并行化,在每个局部重连操作涉及的几何区域中定义一个特征点,然后沿希尔伯特(Hilbert)曲线对特征点进行排序,曲线的均匀分解对应局部重连操作在各线程的分配。这一分配策略的优点是使并行执行局部重连操作时重连区域相互干涉的情形极少出现。因此,当干涉情形出现时,可选择放弃产生干涉的操作,并行优化效率和效果并无明显的负面影响。最后,数值实验验证了本文算法的效率和有效性。

A multi-threaded parallel algorithm for quality improvement of tetrahedral meshes

A multi-threaded parallel version of a sequential quality improvement algorithm for tetrahedral meshes is proposed.The sequential algorithm combines mesh smoothing operations and local reconnection operations;thus,different techniques are developed to parallelize both operations.The parallelization of mesh smoothing operations is based on an existing data decomposition technique,which colors the dual graph of the mesh to subdivide mesh points into a few independent sets.The parallelization of local reconnection operation is based on a new data decomposition technique,which defines a feature point in the interior of each local reconnection operation,and sorts the features points along a Hilbert curve.The decomposition of this Hilbert curve in order could result in a required distribution of local reconnection operations on active threads.This distribution features its ability to ensure that the regions of local operations concurrently executed in different threads are usually separated far away,so that the possibility of interference is reduced to a very low level.As a result,even if interference happens,we could simply withdraw those local reconnection operations that cause interference,and it was observed that no obvious negative impact was introduced by this simple strategy.Finally,numerical experiments are presented to verify the effectiveness and efficiency of the proposed algorithm.