Linear,quadratic, and bilinear programming approaches to the linear complementarity problem

Traditional pivoting procedures for solving the linear complementarity problem can only guarantee convergence for problems having well defined structures. Recently, optimization procedures based on linear, quadratic, and bilinear programming have been developed to extend the class of problems that can be solved efficiently. These latter approaches are the focus of this paper.The strengths and weaknesses of each of the approaches are discussed. The linear programming approach, advanced by Mangasarian, is the most efficient once an appropriate objective function is found. This requires the solution of a system of linear and bilinear equations that is easily solvable only in some cases. Extensions to this approach, due to Shiau, show some promise but are still limited to special cases of the general problem. The quadratic programming approaches discussed here are restricted to specialized procedures for the complementarity problem. One, proposed by Cheng, is based on the levitin-Poljak gradient projection method, and the other, due to Cirina, is based on Karush-Kuhn-Tucker theory. Both are only successful on some problems. The two bilinear programming algorithms discussed are the most general. For any problem, they are guaranteed to find at least one solution or conclude that none exist. One is a specialization of the recent biconvex programming algorithm of Al-Khayyal and Falk and the other is an entirely new implicit enumeration procedure.     

A new algorithm for solving the general quadratic programming problem

For the general quadratic programming problem (including an equivalent form of the linear complementarity problem) a new solution method of branch and bound type is proposed. The branching procedure uses a well-known simplicial subdivision and the bound estimation is performed by solving certain linear programs.     

Semidefinite programming approach for the quadratic assignment problem with a sparse graph

The matching problem between two adjacency matrices can be formulated as the NP-hard quadratic assignment problem (QAP). Previous work on semidefinite programming (SDP) relaxations to the QAP have produced solutions that are often tight in practice, but such SDPs typically scale badly, involving matrix variables of dimension \(n^2\) where n is the number of nodes. To achieve a speed up, we propose a further relaxation of the SDP involving a number of positive semidefinite matrices of dimension \(\mathcal {O}(n)\) no greater than the number of edges in one of the graphs. The relaxation can be further strengthened by considering cliques in the graph, instead of edges. The dual problem of this novel relaxation has a natural three-block structure that can be solved via a convergent Alternating Direction Method of Multipliers in a distributed manner, where the most expensive step per iteration is computing the eigendecomposition of matrices of dimension \(\mathcal {O}(n)\). The new SDP relaxation produces strong bounds on quadratic assignment problems where one of the graphs is sparse with reduced computational complexity and running times, and can be used in the context of nuclear magnetic resonance spectroscopy to tackle the assignment problem.     

Descent approaches for quadratic bilevel programming

The bilevel programming problem involves two optimization problems where the data of the first one is implicitly determined by the solution of the second. In this paper, we introduce two descent methods for a special instance of bilevel programs where the inner problem is strictly convex quadratic. The first algorithm is based on pivot steps and may not guarantee local optimality. A modified steepest descent algorithm is presented to overcome this drawback. New rules for computing exact stepsizes are introduced and a hybrid approach that combines both strategies is discussed. It is proved that checking local optimality in bilevel programming is a NP-hard problem.Support of this work has been provided by INIC (Portugal) under Contract 89/EXA/5, by FCAR (Québec), and by NSERC and DND-ARP (Canada).     

Linear programming insights into solvable cases of the quadratic assignment problem

The quadratic assignment problem is an NP-hard discrete optimization program that has been extensively studied for over 50 years. It has a variety of applications in many fields, but has proven itself extremely challenging to solve. As a result, an area of research has been to identify special cases which admit efficient solution strategies. This paper examines four such cases, and shows how each can be explained in terms of the dual region to the continuous relaxation of a classical linear reformulation of the problem known as the level-1 RLT representation. The explanations allow for simplifications and/or generalizations of the conditions defining the special cases.     

Parametric linear programming techniques for the indefnite quadratic programming problem

In this paper, we discuss a multiparametric technique for findinga global minimum for an indefinite quadratic programming problembased on the spectral decomposition of the matrix of the quadraticform. Special attention is given to the case where this matrixhas only rank 1, where the multiparametric linear program turnsout to be a single-parameter linear program. An extension ofthe traditional linear parametric procedure is introduced whichin general solves this problem efficiently. However, an exampleis presented showing that this technique may take an exponentialnumber of steps.     

On active-set methods for the quadratic programming problem

The active-set Newton method developed earlier by the authors for mixed complementarity problems is applied to solving the quadratic programming problem with a positive definite matrix of the objective function. A theoretical justification is given to the fact that the method is guaranteed to find the exact solution in a finite number of steps. Numerical results indicate that this approach is competitive with other available methods for quadratic programming problems.     

Partial Lagrangian relaxation for general quadratic programming

We give a complete characterization of constant quadratic functions over an affine variety. This result is used to convexify the objective function of a general quadratic programming problem (Pb) which contains linear equality constraints. Thanks to this convexification, we show that one can express as a semidefinite program the dual of the partial Lagrangian relaxation of (Pb) where the linear constraints are not relaxed. We apply these results by comparing two semidefinite relaxations made from two sets of null quadratic functions over an affine variety.      

Stochastic graph partitioning: quadratic versus SOCP formulations

We consider a variant of the graph partitioning problem involving knapsack constraints with Gaussian random coefficients. In this new variant, under this assumption of probability distribution, the problem can be traditionally formulated as a binary SOCP for which the continuous relaxation is convex. In this paper, we reformulate the problem as a binary quadratic constrained program for which the continuous relaxation is not necessarily convex. We propose several linearization techniques for latter: the classical linearization proposed by Fortet (Trabajos de Estadistica 11(2):111–118, 1960) and the linearization proposed by Sherali and Smith (Optim Lett 1(1):33–47, 2007). In addition to the basic implementation of the latter, we propose an improvement which includes, in the computation, constraints coming from the SOCP formulation. Numerical results show that an improvement of Sherali–Smith’s linearization outperforms largely the binary SOCP program and the classical linearization when investigated in a branch-and-bound approach.     

A splitting method for quadratic programming problem

1. IntroductionThe quadratic programming (QP) problem is the most simple one in nonlinear pro-gramming and plays a very important role in optimization theory and applications.It is well known that matriX splitting teChniques are widely used for solving large-scalelinear system of equations very successfully. These algorithms generate an infinite sequence,in contrast to the direct algorithms which terminate in a finite number of steps. However,iterative algorithms are considerable simpler tha…     

Unconstrained quadratic bivalent programming problem

A branching and pruning algorithm is proposed to minimize an unconstrained quadratic function of 0–1 variables. A local minimizing point is defined; and a necessary and sufficient condition for such a point is identified. The algorithm generates all such local minimizing points. The computational experience with the algorithm, which is encouraging, is also given.     

An efficient memetic algorithm for the graph partitioning problem

We consider the component testing problem of a device that is designed to perform a mission consisting of a random sequence of phases with random durations. Testing is done at the component level to attain desired levels of mission reliability at minimum cost. The components fail exponentially where the failure rate depends on the phase of the mission. The reliability structure of the device involves a series connection of nonidentical components with different failure characteristics. The optimal component testing problem is formulated as a semi-infinite linear program. We present an algorithmic procedure to compute optimal test times based on the column generation technique, and illustrate it with numerical examples.     

Linear and quadratic programming in oriented matroids

We prove constructively duality theorems of linear and quadratic programming in the combinatorial setting of oriented matroids. One version of our algorithm for linear programing has the interesting feature of maintaining feasibility. The development of the quadratic programming duality result suggests the study of properties of square matrices such as symmetry and positive semi-definiteness in the context of oriented matroids.     

On practical conditions for the existence and uniqueness of solutions to the general equality quadratic programming problem

We present practical conditions under which the existence and uniqueness of a finite solution to a given equality quadratic program may be examined. Different sets of conditions allow this examination to take place when null-space, range-space or Lagrangian methods are used to find stationary points for the quadratic program.This research supported in part by the Natural Sciences and Engineering Research Council, Canada.     

A computational method for the indefinite quadratic programming problem

We present an algorithm for the quadratic programming problem of determining a local minimum of ?(x)=$\text{1}\text{2}$x^TQx+c^Tx such that A^Tx?b where Q ymmetric matrix which may not be positive definite. Our method combines the active constraint strategy of Murray with the Bunch-Kaufman algorithm for the stable decomposition of a symmetric matrix. Under the active constraint strategy one solves a sequence of equality constrained problems, the equality constraints being chosen from the inequality constraints defining the original problem. The sequence is chosen so that ?(x) continues to decrease and x remains feasible. Each equality constrained subproblem requires the solution of a linear system with the projected Hessian matrix, which is symmetric but not necessarily positive definite. The Bunch-Kaufman algorithm computes a decomposition which facilitates the stable determination of the solution to the linear system. The heart of this paper is a set of algorithms for updating the decomposition as the method progresses through the sequence of equality constrained problems. The algorithm has been implemented in a FORTRAN program, and a numerical example is given.     

Formulations and valid inequalities for the node capacitated graph partitioning problem

We investigate the problem of partitioning the nodes of a graph under capacity restriction on the sum of the node weights in each subset of the partition. The objective is to minimize the sum of the costs of the edges between the subsets of the partition. This problem has a variety of applications, for instance in the design of electronic circuits and devices. We present alternative integer programming formulations for this problem and discuss the links between these formulations. Having chosen to work in the space of edges of the multicut, we investigate the convex hull of incidence vectors of feasible multicuts. In particular, several classes of inequalities are introduced, and their strength and robustness are analyzed as various problem parameters change.     

Performance of a genetic algorithm for the graph partitioning problem

The performance of the genetic algorithm (GA) for the graph partitioning problem (GPP) is investigated by comparison with standard heuristics on well-known benchmark graphs. In general, there is a case where a practical performance of a conventional genetic approach, which performs only simple operations without a local search strategy, is not sufficient. However, it is known that a combination of GA and local search can produce better solutions. From this practice, we incorporate a simple local search algorithm into the GA. In particular, the search ability of the GA is compared with standard heuristics such as multistart local search and simulated annealing, which use the same neighborhood structure of the simple local search, for solving the GPP. Experimental results show that the GA performs better than its competitors.