Application of the lifting wavelet-like transform to fast multipole methods
Ming-Sheng Chen, Xianliang Wu, Wei E. I. Sha, Zhixiang Huang · 2008
In this paper, a similar but novel method presented in [3, 4], lifting wavelet-like transform (LWLT), is applied to the FMM method to sparsify the aggregation and disaggregation matrix. By compressing the elements in those two matrices in time, the method proposed can further speed up the MVM and save much memory when the FMM is used. Numerical results for different shaped three-dimensional objects are considered. Compared with the FMM technique, the application of the LWLT to the FMM can accelerate the MVM by factor of two with lower memory consumed.