科研之友
微信
新浪微博
Facebook
分享链接
摘要
This paper develops a projection-based splitting method for solving the structured variational inequality (SVI). The method is a two-stage method which consists of a prediction step and a correction step and which fully exploits the favorable structure of the problem. One important benefit of our method is that the prediction step can be implemented simultaneously and the correction step requires tiny computational cost. Thus, it gains eligibility for solving large-scale problems. We prove the global convergence of the new method under some mild assumptions. In addition, we formulate the generalized split equality problem and network resource allocation problem as special cases of the SVI and gainfully employ our new method to these problems.
