Accelerated Randomized Coordinate Descent for Solving Linear Systems

The randomized coordinate descent (RCD) method is a simple but powerful approach to solving inconsistent linear systems. In order accelerate this approach, the Nesterov accelerated (NARCD) proposed. with momentum (RCDm) proposed by Nicolas Loizou, we will provide new convergence boundary. global rates of two methods are established in our paper. addition, show that RCDm has an rate choosing proper parameter. Finally, numerical experiments, both and NARCD faster than RCD for uniformly distributed data. Moreover, better acceleration effect stochastic gradient method. When correlation matrix A stronger, better.