An Extended ADMM Algorithm Based on the Prediction-correction Framework
In order to reduce the dimension of large-scale linear constraints separable convex optimi-zation problem,improve the convergence speed of the problem,an extended symmetric alternating di-rection multiplier method(ADMM)with adjacent term is proposed based on the prediction-correction framework(EPSADMM).The algorithm introduces the adjacent term with parameter into the each sub-problem and corrects the sequence by using the new correction matrix,and relaxes the range of correction step.Numerical experiments show that the EPSADMM algorithm has faster convergence rates than other algorithms based on ADMM.
万方数据
维普
