摘要
This paper is concerned with a class of convex multivariable nonlinear program problems. By virtue of linearization philosophy, a linearization problem (LP) is constructed and theoretical equivalence between (LP) and the original problem established. Based on relaxation techniques an algorithm for solving (LP) is proposed, which is efficient from a computational viewpoint, since at each iteration the only program that needs to solve is a standard linear program. Furthermore, the optimality criterion is derived. The convergence analysis conducted in this paper indicates that the algorithm guarantees finite ε convergence.
Ｄ ＯＦ Ｎ－ＤＩＭＥＮＳＩＯＮＡＬ ＮＯＮＬＩＮＥＡＲ ＯＰＴＩＭＩＺＡＴＩＯＮ ＰＲＯＢＬＥＭTX＠徐飞＠王浣尘ＩｎｔｒｏｄｕｃｔｉｏｎＩｔｉｓｄｅｓｉｒａｂｌｅｔｏｒｅｐｌａｃｅｎｏｎｌｉｎｅａｒｐｒｏｂｌｅｍｓｗｉｔｈｌｉｎｅａｒａｐｐｒｏｘｉｍａｔ...
