In this article, we introduce a general splittingmethod with linearization to solve the split feasibility problem and propose a way of selecting the stepsizes such that the implementation of the method does not need ...
详细信息
In this article, we introduce a general splittingmethod with linearization to solve the split feasibility problem and propose a way of selecting the stepsizes such that the implementation of the method does not need any prior information about the operator norm. We present the constant and adaptive relaxation parameters, and the latter is "optimal" in theory. These ways of selecting stepsizes and relaxation parameters are also practised to the relaxed splitting method with linearization where the two closed convex sets are both level sets of convex functions. The weak convergence of two proposed methods is established under standard conditions and the linear convergence of the general splittingmethod with linearization is analyzed. The numerical examples are presented to illustrate the advantage of our methods by comparing with other methods.
暂无评论