中文摘要：

针对时域有限差分（FDTD）算法的计算机实现困难，提出时间与空间容错概念，并详细分析了FDTD算法的容错特性，以及影响容错FDTD执行性能与计算精度的几个关键因素。在容错实现中，本文提出容错写盘时忽略理想匹配层（PML）吸收边界，并利用内存映射技术实现电／磁场值数组，然后将其应用到FDTD算法的时间与空间容错中。数值模拟实例和相关的性能比较结果验证了方法的有效性。

英文摘要：

To aim at the difficulty of the implementaion in computers for finite-difference time-domain （FDTD） algorithm, the concepts of time and space fault-tolerance are presented here.Furthermore, the fault-tolerance characteristic of FDTD algorithm is analyzed in detail, and some key factors are discussed which influence the running performance and computation precision in the fault-tolerant FDTD algorithm. With the implementaion of the fault-tolerance,the perfectly matched layer （PML） absorbing boundary is ignored, and the data arrays of electromagnetic field are gained by using memory-mapped technique, which is applied to time and space fault-tolerance for FDTD algorithm successfully. The computed examples are given, and various comparisons are made to prove the feasibility of this approach. The results show that the proposed method is effective.