MERIT FUNCTION AND GLOBAL ALGORITHMFOR BOX CONSTRAINED VARIATIONALINEQUALITIES

The authors consider optimization methods for box constrained variational inequalities. First, the authors study the KKT-conditions problem based on the original problem. A merit function for the KKT-conditions problem is proposed, and some desirable properties of the merit function are obtained. Through the merit function, the original problem is reformulated as minimization with simple constraints. Then, the authors show that any stationary point of the optimization problem is a solution of the original problem. Finally, a descent algorithm is presented for the optimization problem, and global convergence is shown.

New smooth gap function for box constrained variational inequalities

A new smooth gap function for the box constrained variational inequality problem （VIP） is proposed based on an integral global optimality condition. The smooth gap function is simple and has some good differentiable properties. The box constrained VIP can be reformulated as a differentiable optimization problem by the proposed smooth gap function. The conditions, under which any stationary point of the optimization problem is the solution to the box constrained VIP, are discussed. A simple frictional contact problem is analyzed to show the applications of the smooth gap function. Finally, the numerical experiments confirm the good theoretical properties of the method.

Monotone projected gradient methods for large-scale box-constrained quadratic programming

Inspired by the success of the projected Barzilai-Borwein (PBB) method for largescale box-constrained quadratic programming, we propose and analyze the monotone projected gradient methods in this paper. We show by experiments and analyses that for the new methods,it is generally a bad option to compute steplengths based on the negative gradients. Thus in our algorithms, some continuous or discontinuous projected gradients are used instead to compute the steplengths. Numerical experiments on a wide variety of test problems are presented, indicating that the new methods usually outperform the PBB method.

非线性l_1问题的一种解法

本文对非线性l1问题minx∈RnF(x)=∑mi=1|fi(x)|,从理论上研究了F(x)的下降方向、最优解与某种盒式约束最小二乘问题的最优解之间的关系,进而构造了一个非线性l1问题的下降算法,并证明了该算法的收敛性.数值例子说明所给的非线性l1问题的下降算法是有效的.

箱式约束变分不等式的一类新光滑gap函数

针对箱式约束变分不等式问题,利用一类积分型全局最优性条件,提出了一个新光滑gap函数.该光滑gap函数形式简单且具有较好的性质.利用该gap函数,箱式约束变分不等式可转化为等价光滑优化问题进行求解.进一步地,讨论了可保证等价光滑优化问题的任意聚点为箱式约束变分不等式问题解的条件.以一个简单的摩擦接触问题为例阐释了该方法的应用.最后,利用标准的变分不等式考题验证了方法的有效性.

框式约束凸二次规划问题的内点算法

目前已经有许多关于凸二次规划问题的研究,如文[1][2][5]等,文章对文[1]所给的原始-对偶内点算法理论上的某些缺陷加以更正,给出了框式约束凸二次规划问题的一个修正原始-对偶内点算法并进行了证明.

箱约束变分不等式的一种非精确半光滑算法

为提高求解箱约束变分不等式问题的效率,文章在一个互补函数的基础上,将原问题转化为与之等价的方程组,给出一种非精确半光滑算法。在该算法的每步迭代中,相应的线性方程组都采用非精确求解方法。算法的全局收敛性被证明,数值试验表明,算法对求解该类问题稳定可靠。

对Frank-Wolfe算法在图像恢复中最小二乘问题的研究

在图像恢复问题中经常需要求解一个带箱约束的最小二乘问题。传统上,该问题通常先采用最速下降法求解一个无约束的最小二乘问题,然后将解投影到箱式约束中。这样一种途径得到的解是次优的。Frank-Wolfe算法是一个经典的求解带约束问题的迭代算法,其收敛速度为O(1/k)。本文采用该算法来解决图像恢复中的带箱式约束的最小二乘问题。数值结果表明,采用Frank-Wolfe算法得到的恢复图像要优于最速下降法。

一类箱约束线性最小二乘问题牛顿型算法的有限步终止性

在适当条件下,建立了一类箱约束线性最小二乘问题牛顿型算法的有限步终止性。