Time-Division Parallel FDTD Algorithm

Abstract

We propose a novel and efficient algorithm to parallelize the finite-difference time-domain method, where the observation period is divided into an arbitrary number of subsections, whose computation is distributed to corresponding computer nodes. The proposed algorithm roughly reduces the computational time to an nth fraction of the conventional algorithm, where n is the number of nodes for parallel computing, thus verifying its efficiency improvement.

Optimal Parallel Algorithm of Fast Inverse Laplace Transform for Electromagnetic Analysis

“…Table II compares the computational times of parallel IFFT (inverse fast Fourier transform) [23] and FILT for varying numbers of nodes. The frequency-domain or CFD data were transformed into time-domain, and the data length was 2 15 . It can be observed that for a single or few nodes, IFFT is faster than FILT.…”

“…We recently developed a novel technique for conducting both time-and frequency-domain analyses; specifically, we combined FILT with the finite-difference complexfrequency-domain (FDCFD) method [12] for the near-field analysis of plasmonic objects. We also developed a time-division algorithm for the finite-difference time-domain (FDTD) technique [15]. This algorithm combines the FDTD [16], FDCFD, and FILT methods.…”

Reference Solutions for Time Domain Electromagnetic Solvers

“…By truncating the infinite series, the final expression can be obtained as, where k is the truncation number. Here, the accuracy of f ec can be controlled by an approximation parameter α [14], [15]. The instantaneous field distribution at an observation time t can be accurately and efficiently solved using the summation in Eq.…”

“…Our precise time domain electromagnetic responses can be obtained using the following procedure: rigorous solutions or highly accurate numerical results of electromagnetic waves are computed in the complex frequency domain. The waves in the complex frequency domain are numerically transformed into time domain using fast inversion of Laplace transform (FILT) [14], [15]. The instantaneous value is easily and efficiently obtained.…”

Parallel Implenetation of the GPR Techniques for Detecting and Mapping Ancient Buildings by Using CUDA

“…In the next step, the information used in the computations of electromagnetic fields is transformed. In the FDTD algorithm operations performed on the GPU, so there is no data transfer process because all calculations are performed on the GPU [16][17][18][19].…”