Multi-step nonlinear conjugate gradient methods for unconstrained minimization

Conjugate gradient methods are appealing for large scale nonlinear optimization problems, because they avoid the storage of matrices. Recently, seeking fast convergence of these methods, Dai and Liao (Appl. Math. Optim. 43:87–101, 2001) proposed a conjugate gradient method based on the secant condition of quasi-Newton methods, and later Yabe and Takano (Comput. Optim. Appl. 28:203–225, 2004) proposed another conjugate gradient method based on the modified secant condition. In this paper, we make use of a multi-step secant condition given by Ford and Moghrabi (Optim. Methods Softw. 2:357–370, 1993; J. Comput. Appl. Math. 50:305–323, 1994) and propose two new conjugate gradient methods based on this condition. The methods are shown to be globally convergent under certain assumptions. Numerical results are reported.     

Subspace Barzilai-Borwein Gradient Method for Large-Scale Bound Constrained Optimization

An active set subspace Barzilai-Borwein gradient algorithm for large-scale bound constrained optimization is proposed. The active sets are estimated by an identification technique. The search direction consists of two parts: some of the components are simply defined; the other components are determined by the Barzilai-Borwein gradient method. In this work, a nonmonotone line search strategy that guarantees global convergence is used. Preliminary numerical results show that the proposed method is promising, and competitive with the well-known method SPG on a subset of bound constrained problems from CUTEr collection. This work was supported by the 973 project granted 2004CB719402 and the NSF project of China granted 10471036.     

An adaptive memory algorithm for the k-coloring problem

Let G=(V,E) be a graph with vertex set V and edge set E. The k-coloring problem is to assign a color (a number chosen in {1,…,k}) to each vertex of G so that no edge has both endpoints with the same color. The adaptive memory algorithm is a hybrid evolutionary heuristic that uses a central memory. At each iteration, the information contained in the central memory is used for producing an offspring solution which is then possibly improved using a local search algorithm. The so obtained solution is finally used to update the central memory. We describe in this paper an adaptive memory algorithm for the k-coloring problem. Computational experiments give evidence that this new algorithm is competitive with, and simpler and more flexible than, the best known graph coloring algorithms.     

Monotonicity in the stock competition game with consumer search

Anupindi and Bassok investigate the impact of centralizing the stocks in a one-manufacturer-two-retailer supply chain with consumer search. They observed through numerical studies that in the decentralized scenario, the retailers’ order quantities are monotonic in the consumer search probability. This paper provides an analytical justification of this monotonicity.     

Greatest Descent Algorithms in Unconstrained Optimization

We show that, for an unconstrained optimization problem, the long-term optimal trajectory consists of a sequence of greatest descent directions and a Newton method in the final iteration. The greatest descent direction can be computed approximately by using a Levenberg-Marquardt like formula. This implies the view that the Newton method approximates a Levenberg-Marquardt like formula at a finite distance from the minimum point, instead of the standard view that the Levenberg-Marquadt formula is a way to approximate the Newton method. With the insight gained from this analysis, we develop a two-dimensional version of a Levenberg-Marquardt like formula. We make use of the two numerically largest components of the gradient vector to define here new search directions. In this way, we avoid the need of inverting a high-dimensional matrix. This reduces also the storage requirements for the full Hessian matrix in problems with a large number of variables. The author thanks Mark Wu, Professors Sanyang Liu, Junmin Li, Shuisheng Zhou and Feng Ye for support and help in this research as well as the referees for helpful comments.     

POPMUSIC for the point feature label placement problem

Point feature label placement is the problem of placing text labels adjacent to point features on a map so as to maximize legibility. The goal is to choose positions for the labels that do not give rise to label overlaps and that minimize obscuration of features. A practical goal is to minimize the number of overlaps while considering cartographic preferences. This article proposes a new heuristic for solving the point feature label placement problem based on the application of the POPMUSIC frame. Computational experiments show that the proposed heuristic outperformed other recent metaheuristics approaches in the literature. Experiments with problem instances involving up to 10 million points show that the computational time of the proposed heuristic increases almost linearly with the problem size. New problem instances based on real data with more than 13,000 labels are proposed.     

An ATM Switch Capacity Allocation Problem

In this paper, we consider a capacity allocation problem arising in the context of deploying asynchronous transfer mode (ATM) networks. Given types of ATM switches, the problem is to find an optimal allocation of capacity such that the total cost of installed switch modules is minimized, while satisfying all traffic demands. We formulate the problem as an integer programming model, which can be conceptualized as a generalized bin packing problem with capacity constraints. We then develop a surrogate constraint technique of the formulation for enhancing the representation of the model. Guided by the tight lower bound of the surrogate relaxation, we devise an effective tabu search heuristic that provides good quality solutions with guaranteed optimality. We present promising computational results that show the viability of the proposed heuristic and lower bounding methods, and that provide insights into implementation for configuring ATM switch in practice.     

Partial integration of frequency allocation within antenna positioning in GSM mobile networks

In this article we propose to partially integrate the antenna positioning (APP) and frequency allocation problems (FAP). The traditional wireless network design process examines these two major issues sequentially in order to avoid the very high complexity associated with the simultaneous resolution of the two problems. The proposed integration involves the introduction of interference protection guarantees within the APP. It is customary to define such guarantees in an intermediate step and to use them as input to FAP, in order to protect against interference in critical areas. The proposed approach consists of selecting these protections while solving the APP, allowing the optimization procedure to exploit the degrees of freedom that this would offer. Results on two real-life problem instances indicate a significant improvement in interference levels and resource utilization.     

Variable neighborhood search for minimum cost berth allocation

The berth allocation problem is to allocate space along the quayside to incoming ships at a container terminal in order to minimize some objective function. We consider minimization of total costs for waiting and handling as well as earliness or tardiness of completion, for all ships. We assume ships can arrive at any given time, i.e., before or after the berths become available. The resulting problem, which subsumes several previous ones, is expressed as a linear mixed 0–1 program. As it turns out to be too time-consuming for exact solution of instances of realistic size, a Variable Neighborhood Search (VNS) heuristic is proposed, and compared with Multi-Start (MS), a Genetic Search algorithm (GA) and a Memetic Search algorithm (MA). VNS provides optimal solutions for all instances solved to optimality in a previous paper of the first two authors and outperforms MS, MA and GA on large instances.     

Pattern search ranking and selection algorithms for mixed variable simulation-based optimization

The class of generalized pattern search (GPS) algorithms for mixed variable optimization is extended to problems with stochastic objective functions. Because random noise in the objective function makes it more difficult to compare trial points and ascertain which points are truly better than others, replications are needed to generate sufficient statistical power to draw conclusions. Rather than comparing pairs of points, the approach taken here augments pattern search with a ranking and selection (R&S) procedure, which allows for comparing many function values simultaneously. Asymptotic convergence for the algorithm is established, numerical issues are discussed, and performance of the algorithm is studied on a set of test problems.