利用对偶理论将正定式几何规划转化为带有非负约束和线性等式约束下的非线性凸规划,并且将简约梯度算法与共轭梯度算法恰当结合,应用于求解约束正定式几何规划的对偶问题,构造出了求解几何规划的一个有效算法,并在Armijo步长搜索和适当的条件下证明了该算法的收敛性。
The posynomial geometric programming is transferred into a nolinear convex progamming with contraints of linear equality by duality principle. Integrating the reduced gradient algorithm with the conjugate gradient algorithm,we solve the dual problem of posynomial geometric program with eon-straints and construct an effective algorithm for geometric programming. The convergent properties of the algorithm are discussed under Armijo step search and appropriate conditions.