Advances in Computational Mathematics ( IF 1.7 ) Pub Date : 2022-05-10 , DOI: 10.1007/s10444-022-09953-3 Dang Van Hieu 1 , Le Dung Muu 2 , Pham Kim Quy 3
The paper introduces an one-step optimization method for solving a monotone equilibrium problem including a Lipschitz-type condition in a Hilbert space. The method uses variable stepsizes and is constructed by the proximal-like mapping associated with the cost bifunction and incorporated with regularization terms. Comparing with the extragradient-like methods, our new method has an elegant and simple structure with a cheap computation over each iteration. By an appropriate choice of stepsizes and regularization parameters, we establish the strong convergence of the iterative sequence generated by the method to a solution of the considered equilibrium problem. We also show the numerical behavior of our new method and illustrate the computational effectiveness of it over other methods via experiments.
中文翻译:
平衡问题的一步优化方法
本文介绍了一种单调平衡问题的单调优化方法,该问题包括希尔伯特空间中的 Lipschitz 型条件。该方法使用可变步长,并由与成本双函数相关联的类似近端的映射构造,并结合正则化项。与类外梯度方法相比,我们的新方法结构优雅简单,每次迭代的计算成本低。通过适当的步骤和正则化参数的选择,我们确定了该方法生成的迭代序列的强收敛,以对考虑的平衡问题的解决方案。我们还展示了我们的新方法的数值行为,并通过实验说明了它相对于其他方法的计算有效性。