A NEW TRUST-REGION ALGORITHM FOR NONLINEAR CONSTRAINED OPTIMIZATION

We propose a new trust region algorithm for nonlinear constrained optimization problems. In each iteration of our algorithm, the trial step is computed by minimizing a quadratic approximation to the augmented Lagrange function in the trust region. The augmented Lagrange function is also used as a merit function to decide whether the trial step should be accepted. Our method extends the traditional trust region approach by combining a filter technique into the rules for accepting trial steps so that a trial step could still be accepted even when it is rejected by the traditional rule based on merit function reduction. An estimate of the Lagrange multiplier is updated at each iteration, and the penalty parameter is updated to force sufficient reduction in the norm of the constraint violations. Active set technique is used to handle the inequality constraints. Numerical results for a set of constrained problems from the CUTEr collection are also reported.     

SEQUENTIAL QUADRATIC PROGRAMMING METHODS FOR OPTIMAL CONTROL PROBLEMS WITH STATE CONSTRAINTS

A Kind of direct methods is presented for the solution of optimal control problems with state constraints.These methods are sequential quadratic programming methods.At every iteration a quadratic programming which is obtained by quadratic approximation to Lagrangian function and Linear approximations to constraints is solved to get a search direction for a merit function.The merit function is formulated by augmenting the Lagrangian funetion with a penalty term.A line search is carried out along the search direction to determine a step length such that the merit function is decreased.The methods presented in this paper include continuous sequential quadratic programming methods and discreate sequential quadrade programming methods.     

COMPOSITE-STEP LIKE FILTER METHODS FOR EQUALITY CONSTRAINT PROBLEMS

In a composite-step approach, a step sk is computed as the sum of two components vk and hk. The normal component vk, which is called the vertical step, aims to improve the linearized feasibility, while the tangential component hk, which is also called horizontal step, concentrates on reducing a model of the merit functions. As a filter method, it reduces both the infeasibility and the objective function. This is the same property of these two methods. In this paper, one concerns the composite-step like filter approach. That is, a step is tangential component hk if the infeasibility is reduced. Or else, sk is a compositestep composed of normal component Vk and tangential component hk.     

线性互补问题的一种新Lagrange乘子法

A new multiplier method for solving the linear complementarity problem LCP(q, M) is proposed. Based on the Lagrangian of LCP(q,M) introduced here, we construct a new differentiable merit function θ(x,λ) which containing a multiplier vector λ and satisfying θ(x,λ) ≥ 0 and θ(x,λ) = 0 if and if only x solves LCP(q,M). A simple damped Newton-type algorithm which based on the merit function θ(x,λ) is presented. The main feature of the method is that the multiplier self-adjusting step accelerates the local convergence rate without losing global convergence. When M is the P-matrix, the sequence {θ(x^k,λ^k)}where {(x^k,λ^k)} generated by the algorithm is globally linearly convergent to zero and convergent in finite number of iterations if the solution is nondegenerate. Numerical results suggest that the method is high efficient and promising.     

基于增广Lagrange函数的RQP方法

Recursive quadratic programming is a family of techniques developd by Bartholomew-Biggs and other authors for solving nonlinear programming problems.This paperdescribes a new method for constrained optimization which obtains its search di-rections from a quadratic programming subproblem based on the well-known aug-mented Lagrangian function.It avoids the penalty parameter to tend to infinity.We employ the Fletcher‘s exact penalty function as a merit function and the use of an approximate directional derivative of the function that avoids the need toevaluate the second order derivatives of the problem functions.We prove that thealgorithm possesses global and superlinear convergence properties.At the sametime, numerical results are reported.     

A MONOTONE COMPACT IMPLICIT SCHEME FOR NONLINEAR REACTION-DIFFUSION EQUATIONS

A monotone compact implicit finite difference scheme with fourth-order accuracy in space and second-order in time is proposed for solving nonlinear reaction-diffusion equations. An accelerated monotone iterative method for the resulting discrete problem is presented. The sequence of iteration converges monotonically to the unique solution of the discrete problem, and the convergence rate is either quadratic or nearly quadratic, depending on the property of the nonlinear reaction. The numerical results illustrate the high accuracy of the proposed scheme and the rapid convergence rate of.the iteration.     

A SMOOTHING QP-FREE INFEASIBLE METHOD FOR NONLINEAR INEQUALITY CONSTRAINED OPTIMIZATION

In this paper,a smoothing QP-free infeasible method is proposed for nonlinear inequality constrained optimization problems.This iterative method is based on the solution of nonlinear equations which is obtained by the multipliers and the smoothing Fisher-Burmeister function for the KKT first-order optimality conditions.Comparing with other QP-free methods, this method does not request the strict feasibility of iteration.In particular,this method is implementable and globally convergent without assuming the strict complementarity condition and the isolatedness of accumulation points.Furthermore,the gradients of active constraints are not requested to be linearly independent.Preliminary numerical results indicate that this smoothing QP-free infeasible method is quite promising.     

Properties of a family of merit functions and a merit function method for the NCP

A family of merit functions are proposed, which are the generalization of several existing merit functions. A number of favorable properties of the proposed merit functions are established. By using these properties, a merit function method for solving nonlinear complementarity problem is investigated, and the global convergence of the proposed algorithm is proved under some standard assumptions. Some preliminary numerical results are given.     

带非局部阻尼项的粘弹性方程解的衰减估计

In this paper,we consider the following viscoelastic equation u tt- △u +∫t 0 g(t-s)△u(s)ds + a(x)u t + u |u|r = 0 with initial condition and Dirichlet boundary condition.The decay property of the energy function closely depends on the properties of the relaxation function g(t) at infinity.In the previous works of [3,7,11],it was required that the relaxation function g(t) decay exponentially or polynomially as t → +∞.In the recent work of Messaoudi [12,13],it was shown that the energy decays at a similar rate of decay of the relaxation function,which is not necessarily dacaying in a polynomial or exponential fashion.Motivated by [12,13],under some assumptions on g(x),a(x) and r,and by introducing a new perturbed energy,we also prove the similar results for the above equation.     

$D^2(\Omega,\rho)$的再生核与再生核诱导的度量

An important property of the reproducing kernel of D^2（Ω, ρ） is obtained and the reproducing kernels for D^2（Ω, ρ） are calculated when Ω = Bn× Bn and ρ are some special functions. A reproducing kernel is used to construct a semi-positive definite matrix and a distance function defined on Ω×Ω. An inequality is obtained about the distance function and the pseudodistance induced by the matrix.     

垂直线性互补问题的一种光滑算法

文中给出了垂直线性互补问题的一个新的光滑价值函数,不同于光滑化方法中的价值函数,它不包含任何必须趋向零的参数,因此算法中不涉及参数调整步骤,而且具有良好的强制性.基此价值函数,提出了求解垂直线性互补问题的一种阻尼Newton类算法,并证明了该算法对竖块P0+R0矩阵的垂直线性互补问题具有全局收敛性;当解满足相当于BD-正则条件时,算法具有局部二次收敛性;在不增加额外校正步骤(算法的每个迭代步只求解一个Newton方程)的情形下,算法对竖块P-矩阵垂直线性互补问题(无须假设严格互补),具有有限步收敛性.数值实验结果令人满意.     

箱约束变分不等式的一个简单光滑价值函数和阻尼牛顿法

通过引入中间值函数的一类光滑价值函数,构造了箱约束变分不等式的一种新的光滑价值函数,该函数形式简单且具有良好的微分性质.基于此给出了求解箱约束变分不等式的一种阻尼牛顿算法,在较弱的条件下,证明了算法的全局收敛性和局部超线性收敛率,以及对线性箱约束变分不等式的有限步收敛性.数值实验结果表明了算法可靠有效的实用性能.     

求解极小极大问题的非单调过滤算法

提出一种改进的求解极小极大问题的信赖域滤子方法,利用SQP子问题来求一个试探步,尾服用滤子来衡量是否接受试探步,避免了罚函数的使用;并且借用已有文献的思想, 使用了Lagrange函数作为效益函数和非单调技术,在适当的条件下,分析了算法的全局和局部收敛性,并进行了数值实验.     

A globally convergent BFGS method for nonlinear monotone equations without any merit functions

Since 1965, there has been significant progress in the theoretical study on quasi-Newton methods for solving nonlinear equations, especially in the local convergence analysis. However, the study on global convergence of quasi-Newton methods is relatively fewer, especially for the BFGS method. To ensure global convergence, some merit function such as the squared norm merit function is typically used. In this paper, we propose an algorithm for solving nonlinear monotone equations, which combines the BFGS method and the hyperplane projection method. We also prove that the proposed BFGS method converges globally if the equation is monotone and Lipschitz continuous without differentiability requirement on the equation, which makes it possible to solve some nonsmooth equations. An attractive property of the proposed method is that its global convergence is independent of any merit function.We also report some numerical results to show efficiency of the proposed method.

     

MERIT FUNCTION AND GLOBAL ALGORITHM FOR BOX CONSTRAINED VARIATIONAL INEQUALITIES

1 IntroductionWe consider tlie variational inequality problelll, deuoted by VIP(X, F), wliicli is to find avector x* E X such thatF(X*)"(X -- X-) 2 0, VX E X, (1)where F: R" - R" is any vector-valued f11uction and X is a uonelllpty subset of R'.This problem has important applicatiolls. in equilibriun1 modeIs arising in fields such asecououtics, transportatioll scieuce alld operations research. See [1]. There exist mauy lllethodsfor solviug tlie variational li1equality problem VIP(X. …     

A globally convergent Newton method for solving strongly monotone variational inequalities

Variational inequality problems have been used to formulate and study equilibrium problems, which arise in many fields including economics, operations research and regional sciences. For solving variational inequality problems, various iterative methods such as projection methods and the nonlinear Jacobi method have been developed. These methods are convergent to a solution under certain conditions, but their rates of convergence are typically linear. In this paper we propose to modify the Newton method for variational inequality problems by using a certain differentiable merit function to determine a suitable step length. The purpose of introducing this merit function is to provide some measure of the discrepancy between the solution and the current iterate. It is then shown that, under the strong monotonicity assumption, the method is globally convergent and, under some additional assumptions, the rate of convergence is quadratic. Limited computational experience indicates the high efficiency of the proposed method.     

无正则性条件下的一个信赖域方法的全局收敛性

1.引 言考虑下列等式约束最优化问题:min f(x)x∈Rn (1.1)s.t.C(x)=0其中f:Rn→R,C（x）=（c1（x）,C2（x）,…,Cm（x））T,Ci:Rn→R,（i=1,…,m）.我们假设f（x）,Ci（x）（i=1,2,…,m）是连续可微函数.令g（x）=　f（x）,A（x）=　C（x）T.为了方便,我们通常用 Ck,fk,gk,Ak分别表示 C（xk）,f（xk）,g（xk）A（xk）. SQP方法是一迭代方法.在 xk点,通过解下列子问题来得到搜索方向 dk     

等式约束非凸优化问题的修正牛顿算法

本文设计了一个新的求解等式约束非凸优化问题的修正牛顿算法.利用修正的拉格朗日函数,通过求解线性方程组获得搜索方向,利用价值函数的线性近似模型确定步长.在没有非奇异性假设的条件下,证明了算法的全局收敛性.数值结果表明,算法是有效的.     

一个解不等式约束非线性规划问题的有效方法

本文提出了一个解不等式约束非线性规划问题有效方法．在这个方法中,考虑解一个等价Ｋｕｈｎ－Ｔｕｃｋｅｒ条件的非线性方程组．这个方程组中ＮＣＰ函数的使用消去了对应于不等式约束的Ｌａｇｒａｎｇｅ乘子的非负性．截断牛顿方法被用来解这个非线性方程组．为了保证全局收敛性,一个强健的损失函数被选为寻查函数,同时方法中插入修正最速下降方向．本文证明了方法的分Ｑ－二阶收敛性,同时指出新方法可以有效地解稀疏大规模非线性规划问题。     

Globally Convergent Interior-Point Algorithm for Nonlinear Programming

This paper presents a primal-dual interior-point algorithm for solving general constrained nonlinear programming problems. The inequality constraints are incorporated into the objective function by means of a logarithmic barrier function. Also, satisfaction of the equality constraints is enforced through the use of an adaptive quadratic penalty function. The penalty parameter is determined using a strategy that ensures a descent property for a merit function. Global convergence of the algorithm is achieved through the monotonic decrease of a merit function. Finally, extensive computational results show that the algorithm can solve large and difficult problems in an efficient and robust way.Communicated by L. C. W. DixonThe research reported in this paper was done while the first author was at Imperial College. The authors gratefully acknowledge constructive comments from Professor L. C. W. Dixon and an anonymous referee. They are also grateful to Dr. Stanislav Zakovic for helpful suggestions and comments. Financial support was provided by EPSRC Grants M16016 and GR/G51377/01.