本文用多尺度投影方法求解离散的DSM问题,与传统全投影方法相比,减少了内积计算个数,保持了最优收敛率.最后,算例说明了算法的有效性.
In this paper, we develop a fast multiscale projection method to solve the discretization of the DSM problem. Compared with the traditionM projection technique, we keep the optimal convergence rate, but less than the number of inner products calculation. Finally, numerical experiments are given to illustrate the efficiency of the method.