A novel global optimization method based on an Augmented Lagrangian framework is introduced for continuous constrained nonlinear optimization problems. At each outer iteration k the method requires the εk-global minimization of the Augmented Lagrangian with simple constraints, where εk → ε. Global convergence to an ε-global minimizer of the original problem is proved. The subproblems are solved using the αBB method. Numerical experiments are presented.

Abstract. Based on work originating in the early 1970s, a number of recent global optimization algorithms have relied on replacing an original nonconvex nonlinear program by convex or linear relaxations. Such linear relaxations can be generated automatically through an automatic differentiation process. This process decomposes the objective and constraints (if any) into convex and nonconvex unary and binary operations. The convex operations can be approximated arbitrarily well by appending additional constraints, while the domain must somehow be subdivided (in an overall branch-and-bound process or in some other local process) to handle nonconvex constraints. In general, a problem can be hard if even a single nonconvex term appears. However, certain nonconvex terms lead to easier-to-solve problems than others. Recently, Neumaier, Lebbah, Michel, ourselves, and others have paved the way to utilizing such techniques in a validated context. In this paper, we present a symbolic preprocessing step that provides a measure of the intrinsic difficulty of a problem. Based on this step, one of two methods can be chosen to relax nonconvex terms. This preprocessing step is similar to a method previously proposed by Epperly and Pistikopoulos [J. Global Optim., 11 (1997), pp. 287–311] for determining subspaces in which to branch, but we present it from a different point of view that is amenable to simplification of the problem presented to the linear programming solver, and within a validated context. Besides an illustrative example, we have implemented general relaxations in a validated context, as well as the preprocessing technique, and we present experiments on a standard test set. Finally, we present conclusions.

solution of quadratic constraint satisfaction problems

Abstract. New theorems of the alternative for polynomial constraints (based on the Positivstellensatz from real algebraic geometry) and for linear constraints (generalizing the transposition theorems of Motzkin and Tucker) are proved. Based on these, two Karush-John optimality conditions – holding without any constraint qualification – are proved for single- or multi-objective constrained optimization problems. The first condition applies to polynomial optimization problems only, and gives for the first time necessary and sufficient global optimality conditions for polynomial problems. The second condition applies to smooth local optimization problems and strengthens known local conditions. If some linear or concave constraints are present, the new version reduces the number of constraints for which a constraint qualification is needed to get the Kuhn-Tucker conditions.

Abstract. In validated branch and bound algorithms for global optimization, upper bounds on the global optimum are obtained by evaluating the objective at an approximate optimizer; the upper bounds are then used to eliminate subregions of the search space. For constrained optimization, in general, a small region must be constructed within which existence of a feasible point can be proven, and an upper bound on the objective over that region is obtained. We had previously proposed a perturbation technique for constructing such a region. In this work, we propose a much simplified and improved technique, based on an orthogonal decomposition of the normal space to the constraints. In purely inequality constrained problems, a point, rather than a region, can be used, and, for equality and inequality constrained problems, the region lies in a smaller-dimensional subspace, giving rise to sharper upper bounds. Numerical experiments on published test sets for global optimization provide evidence of the superiority of the new approach within our GlobSol environment. 1.

Testing is a crucial part of software development in general, and hence also in mathematical programming. Unfortunately, it is often a time consuming and little exciting activity. This naturally motivated us to increase the e ciency in testing solvers for optimization problems and to automatize as much of the procedure as possible. Keywords: test environment, optimization, solver benchmarking, solver comparison The testing procedure typically consists of three basic tasks: a) organize test problem sets, also called test libraries; b) solve selected test problems with selected solvers; c) analyze, check and compare the results. The Test Environment is a graphical user interface (GUI) that enables to manage the tasks a) and b) interactively, and task c) automatically. The Test Environment is particularly designed for users who seek to 1. adjust solver parameters, or 2. compare solvers on single problems, or 3. evaluate solvers on suitable test sets.

Abstract. This paper serves as background information for the Vienna proposal for interval standardization, explaining what is needed in practice to make competent use of the interval arithmetic provided by an implementation of the standard to be. Discussed are methods to improve the quality of interval enclosures of the range of a function over a box, considerations of possible hardware support facilitating the implementation of such methods, and the results of a simple interval challenge that I had posed to the reliable computing mailing list on November 26, 2008. Also given is an example of a bound constrained global optimization problem in 4 variables that has a 2-dimensional continuum of global minimizers. This makes standard branch and bound codes extremely slow, and therefore may serve as a useful degenerate test problem.

This project aims at the development of a flexible modeling system for the specification of models for large-scale numerical work in optimization, data analysis, and partial differential equations. Its input should be provided in a form natural for the working mathematician, while the choice of the numerical solvers and the transformation to the format required by the solvers is done by the interface system. The input format should combine the simplicity of LaTeX source code with the semantic conciseness and modularity of current modeling languages such as AMPL, and it should be as close as possible to the mathematical language people use to explain and communicate their models in publications and lectures. In order that the system is useful for the intended applications, interfaces translating the model formulated in the proposed system into the input required for current state of the art solvers, and into the dominant current modeling languages are needed and shall be provided. Moreover, certain shortcomings of the current generation of modeling languages, such as the lack of support for the correct treatment of uncertainties and rounding errors, shall be overcome. The experience gained in this project will be useful in future work in the more general context

We propose an efficient interval partitioning algorithm to solve the continuous Constraint Satisfaction Problem (CSP). The method comprises a new dynamic tree search management system that also invokes local search in selected subintervals. This approach is compared with two classical tree search techniques and three other interval methods. We study some challenging kinematics problems for testing the algorithm. The goal in solving kinematics problems is to identify all real solutions of the system of equations defining the problem. In other words, it is desired to find all object positions and orientations that satisfy a coupled nonlinear system of equations. The kinematics benchmarks used here arise in industrial applications.

Global optimization is aimed at finding the best solution of a constrained nonlinear optimization problem by performing a complete search over the set of feasible solutions. In contrast with local optimization, a complete search exhaustively checks the entire feasible region. For a comprehensive up-to-date archive of online information on global optimization, see [6]. As surveyed in [7], there are numerous mathematical and engineering problem classes for which a complete search is required. An example is the 300 year old Kepler problem of finding the densest packing of equal spheres in 3-dimensional Euclidean space, for which a computer-assisted proof is proposed by T. C. Hales [4]. The proof consists in reducing the problem to several thousands of linear programs and using interval calculations to ensure rigorous handling of rounding errors when establishing correctness of inequalities. Many other famous difficult optimization problems, such as the traveling salesman problem and the protein folding problem, are global optimization problems. The Global Optimization Toolbox (GOT for short), first released in June 2004 with Maple 9.5, is part of the Maple Professional Toolbox series of add-on products that must be