Two-Dimensional Matrix Characteristics Method Based on Krylov Subspace and Domain Decomposition Theories

A matrix characteristics method is proposed based on Krylov subspace and domain decomposition theories to improve the efficiency of the conventional method of characteristics（MOC）.The new numerical method,with the form of linear algebraic equation systems,is derived from the MOC basic equations and solved efficiently by the generalized minimal residual（GMRES） method.The spatial non-overlapping domain decomposition method with multiprocessing parallel technology is then adopted to reduce the computational burden for large-scale problems.By inheriting the geometrical treatment manners of the conventional MOC solver AutoMOC,the theories mentioned above are implemented.Numerical results demonstrate that the matrix characteristics method with satisfactory accuracy is more efficient than the conventional MOC,and has good flexibility in complex geometry and strong scattering problems.