Mirjam Dür

Copositive Programming – a Survey

Copositive programming is a relatively young field in mathematical optimization. It can be seen as a generalization of semidefinite programming, since it means optimizing over the cone of so called copositive matrices. Like semidefinite programming, it has proved particularly useful in combinatorial and quadratic optimization. The purpose of this survey is to introduce the field to interested readers in the optimization community who wish to get an understanding of the basic concepts and recent developments in copositive programming, including modeling issues and applications, the connection to semidefinite programming and sum-of-squares approaches, as well as algorithmic solution approaches for copositive programs.

On Copositive Programming and Standard Quadratic Optimization Problems

A standard quadratic problem consists of finding global maximizers of a quadratic form over the standard simplex. In this paper, the usual semidefinite programming relaxation is strengthened by replacing the cone of positive semidefinite matrices by the cone of completely positive matrices (the positive semidefinite matrices which allow a factorization FFT where F is some non-negative matrix). The dual of this cone is the cone of copositive matrices (i.e., those matrices which yield a non-negative quadratic form on the positive orthant). This conic formulation allows us to employ primal-dual affine-scaling directions. Furthermore, these approaches are combined with an evolutionary dynamics algorithm which generates primal-feasible paths along which the objective is monotonically improved until a local solution is reached. In particular, the primal-dual affine scaling directions are used to escape from local maxima encountered during the evolutionary dynamics phase.

An Adaptive Linear Approximation Algorithm for Copositive Programs

We study linear optimization problems over the cone of copositive matrices. These problems appear in nonconvex quadratic and binary optimization; for instance, the maximum clique problem and other combinatorial problems can be reformulated as such problems. We present new polyhedral inner and outer approximations of the copositive cone which we show to be exact in the limit. In contrast to previous approximation schemes, our approximation is not necessarily uniform for the whole cone but can be guided adaptively through the objective function, yielding a good approximation in those parts of the cone that are relevant for the optimization and only a coarse approximation in those parts that are not. Using these approximations, we derive an adaptive linear approximation algorithm for copositive programs. Numerical experiments show that our algorithm gives very good results for certain nonconvex quadratic problems.

Interior points of the completely positive cone.

A matrix A is called completely positive if it can be decomposed as A = BB^T with an entrywise nonnegative matrix B. The set of all such matrices is a convex cone. We provide a characterization of the interior of this cone as well as of its dual.

Solving sum-of-ratios fractional programs using efficient points

Constrained maximization of a sum of p1 ratios is a difficult nonconvex optimization problem (even if all functions involved are linear) with many applications in management sciences. In this paper, we first give a brief introductory survey of this problem. Then we propose a general branch-and-bound algorithm which uses rectangular partitions in the Euclidean space of dimension p. Theoretically, this algorithm is applicable under very general assumptions. Practically, we give an efficient implementation for fine fractions. Here the bounding procedures use dual constructions and the calculation of efficient points of a corresponding multiple-objective optimization problem. Finally, we present some promising numerical results

Lagrange duality and partitioning techniques in nonconvex global optimization

It is shown that, for very general classes of nonconvex global optimization problems, the duality gap obtained by solving a corresponding Lagrangian dual in reduced to zero in the limit when combined with suitably refined partitioning of the feasible set. A similar result holds for partly convex problems where exhaustive partitioning is applied only in the space of nonconvex variables. Applications include branch-and-bound approaches for linearly constrained problems where convex envelopes can be computed, certain generalized bilinear problems, linearly constrained optimization of the sum of ratios of affine functions, and concave minimization under reverse convex constraints.

An improved algorithm to test copositivity

Copositivity plays a role in combinatorial and nonconvex quadratic optimization. However, testing copositivity of a given matrix is a co-NP-complete problem. We improve a previously given branch-and-bound type algorithm for testing copositivity and discuss its behavior in particular for the maximum clique problem. Numerical experiments indicate that the speedup is considerable.

Depth-first simplicial partition for copositivity detection, with an application to MaxClique

Detection of copositivity plays an important role in combinatorial and quadratic optimization. Recently, an algorithm for copositivity detection by simplicial partition has been proposed. In this paper, we develop an improved depth-first simplicial partition algorithm which reduces memory requirements significantly and therefore enables copositivity checks of much larger matrices – of size up to a few thousands instead of a few hundreds. The algorithm has been investigated experimentally on a number of MaxClique problems as well as on generated random problems. We present numerical results showing that the algorithm is much faster than a recently published linear algebraic algorithm for copositivity detection based on traditional ideas – checking properties of principal sub-matrices. We also show that the algorithm works very well for solving MaxClique problems through copositivity checks.

Dual bounding procedures lead to convergent Branch–and–Bound algorithms

Abstract.Branch–and–Bound methods with dual bounding procedures have recently been used to solve several continuous global optimization problems. We improve results on their convergence theory and give a condition that enables us to detect infeasible partition sets from the dual optimal value.

LINEAR-TIME COMPLETE POSITIVITY DETECTION AND DECOMPOSITION OF SPARSE MATRICES ∗

A matrix