Large-scale nonseparable convex optimizaton: Smooth case

There have been considerable researches for solving large-scale separable convex optimization problems. In this paper the authors present a method for large-scale nonseparable smooth convex optimization problems with block-angular linear constraints. One of them is occurred in reconfiguration of the virtual path network which finds the routing path and assigns the bandwidth of the path for each traffic class in ATM (Asynchronous Transfer Mode ) network. The solution is approximated by solving a sequence of the block-angular structured separable quadratic programming problems. Bundle-based decomposition method is applied to each large-scale separable quadratic programming problem. The authors implement the methods and present some computational experiences. [In Korean.]