. Two trust--region interior--point algorithms for the solution of minimization problems with simple bounds are analyzed and tested. The algorithms scale the local model in a way similar to Coleman and Li [1]. The first algorithm is more usual in that the trust region and the local quadratic model are consistently scaled. The second algorithm proposed here uses an unscaled trust region. A global convergence result for these algorithms is given and dogleg and conjugate--gradient algorithms to compute trial steps are introduced. Some numerical examples that show the advantages of the second algorithm are presented. Keywords. trust--region methods, interior--point algorithms, Dikin--Karmarkar ellipsoid, Coleman and Li affine scaling, simple bounds. AMS subject classification. 49M37, 90C20, 90C30 1. Introduction. In this note we consider the box--constrained minimization problem minimize f(x) subject to a x b; (1) where x 2 IR n , a 2 (IR [ f\Gamma1g) n , b 2 (IR [ f+1g) n and...

In this paper a family of trust-region interior-point SQP algorithms for the solution of a class of minimization problems with nonlinear equality constraints and simple bounds on some of the variables is described and analyzed. Such nonlinear programs arise e.g. from the discretization of optimal control problems. The algorithms treat states and controls as independent variables. They are designed to take advantage of the structure of the problem. In particular they do not rely on matrix factorizations of the linearized constraints, but use solutions of the linearized state equation and the adjoint equation. They are well suited for large scale problems arising from optimal control problems governed by partial differential equations. The algorithms keep strict feasibility with respect to the bound constraints by using an affine scaling method proposed for a different class of problems by Coleman and Li and they exploit trust--region techniques for equality-constrained optimizatio...

This paper describes an active-set algorithm for large-scale nonlinear programming based on the successive linear programming method proposed by Fletcher and Sainz de la Maza [10]. The step computation is performed in two stages. In the first stage a linear program is solved to estimate the active set at the solution. The linear program is obtained by making a linear approximation to the ` 1 penalty function inside a trust region. In the second stage, an equality constrained quadratic program (EQP) is solved involving only those constraints that are active at the solution of the linear program.

We present an introduction to a new class of derivative free methods for unconstrained optimization. We start by discussing the motivation for such methods and why they are in high demand by practitioners. We then review the past developments in this field, before introducing the features that characterize the newer algorithms. In the context of a trust region framework, we focus on techniques that ensure a suitable "geometric quality" of the considered models. We then outline the class of algorithms based on these techniques, as well as their respective merits. We finally conclude the paper with a discussion of open questions and perspectives. 1 Motivation In this paper, we consider the problem of minimizing a nonlinear smooth objective function of several variables when the derivatives of the objective function are unavailable and when no constraints are specified on the problem's variables. More formally, we consider the problem min x2R n f(x); where we assume that f i...

Direct boundary value problem methods in combination with SQP iteration have proved to be very successful in solving nonlinear optimal control problems. Such methods use parameterized control functions, discretize the state differential equations by, e.g., multiple shooting or collocation, and treat the discretized boundary value problem as an equality constraint in a large, nonlinear, constrained optimization problem. In real-life applications several thousand variables may appear in the NLP. Solution by standard techniques is therefore impractical. This dissertation develops a general concept for a class of structured direct SQP methods based on a decoupling strategy. A careful choice of the discretization reveals an inherent multistage block structure of the QP subproblems. We present a recursive solution algorithm for the associated KKT systems which makes full use of this sparse structure, and propose a structure-preserving primal-dual interior point method for treating the genera...

The approximate minimization of a quadratic function within an ellipsoidal trust region is an important subproblem for many nonlinear programming methods. When the number of variables is large, the most widely-used strategy is to trace the path of conjugate gradient iterates either to convergence or until it reaches the trust-region boundary. In this paper, we investigate ways of continuing the process once the boundary has been encountered. The key is to observe that the trust-region problem within the currently generated Krylov subspace has very special structure which enables it to be solved very efficiently. We compare the new strategy with existing methods. The resulting software package is available as HSL VF05 within the Harwell Subroutine Library. 1 Department for Computation and Information, Rutherford Appleton Laboratory, Chilton, Oxfordshire, OX11 0QX, England, EU Email : n.gould@rl.ac.uk 2 Current reports available by anonymous ftp from joyous-gard.cc.rl.ac.uk (internet ...

We propose an algorithm for nonlinear optimization that employs both trust region techniques and line searches. Unlike traditional trust region methods, our algorithm does not resolve the subproblem if the trial step results in an increase in the objective function, but instead performs a backtracking line search from the failed point. Backtracking can be done along a straight line or along a curved path. We show that the new algorithm preserves the strong convergence properties of trust region methods. Numerical results are also presented.

Discrete-time optimal control (DTOC) problems are large-scale optimization problems with a dynamic structure. In previous work this structure has been exploited to provide very fast and efficient local procedures. Two examples are the differential dynamic programming algorithm (DDP) and the stagewise Newton procedure -- both require only O(N) operations per iteration, where N is the number of timesteps. Both exhibit a quadratic convergence rate. However, most algorithms in this category do not have a satisfactory global convergence strategy. The most popular global strategy is shifting: this sometimes works poorly due to the lack of automatic adjustment to the shifting element. In this paper we propose a method that incorporates the trust region idea with the local stagewise Newton's method. This method possesses advantages of both the trust region idea and the stagewise Newton's method, i.e., our proposed method has strong global and local convergence properties yet remains economical...

We consider the problem of computing the solution of large-scale discrete ill-posed problems when there is noise in the data. These problems arise in important areas such as seismic inversion, medical imaging and signal processing. We pose the problem as a quadratically constrained least squares problem and develop a method for the solution of such problem. Our method does not require factorization of the coefficient matrix, it has very low storage requirements and handles the high degree of singularities arising in discrete ill-posed problems. We present numerical results on test problems and an application of the method to a practical problem with real data.

bundle filter method for nonsmooth nonlinear