Local MLACA for fast solution of large-scale scattering problems based on H-matrix structure

In this paper, the local multilevel adaptive cross approximation (LMLACA) based on multilevel partition with H-matrix structure is developed. This structure can reduce the number of leaves related to the sub-block impedance matrix. In order to reduce the CPU memory and time further, variable tolerance of ACA algorithm based on physical mechanism is adopted. It is decided according to the distance between the two interaction sub-blocks. Numerical Results demonstrate the efficiency of LMLACA.