Optimal supplier choice with discounting

This paper investigates a model for pricing the demand for a set of goods when suppliers operate discount schedules based on total business value. We formulate the buyers's decision problem as a mixed binary integer program, which is a generalization of the capacitated facility location problem (CFLP). A branch and bound (BnB) procedure using Lagrangean relaxation and subgradient optimization is developed for solving large-scale problems that can arise when suppliers’ discount schedules contain multiple price breaks. Results of computer trials on specially adapted large benchmark instances of the CFLP confirm that a sub-gradient optimization procedure based on Shor and Zhurbenko's r-algorithm, which employs a space dilation in the direction of the difference between two successive subgradients, can be used efficiently for solving the dual problem at any node of the BnB tree.     

Optimizing the periodic pick-up of raw materials for a manufacturer of auto parts

We describe a solution procedure for a special case of the periodic vehicle routing problem (PVRP). Operation managers at an auto parts manufacturer in the north of Spain described the optimization problem to the authors. The manufacturer must pick up parts (raw material) from geographically dispersed locations. The parts are picked up periodically at scheduled times. The problem consists of assigning a pickup schedule to each of its supplier’s locations and also establishing daily routes in order to minimize total transportation costs. The time horizon under consideration may be as long as 90 days. The resulting PVRP is such that the critical decision is the assignment of locations to schedules, because once this is done, the daily routing of vehicles is relatively straightforward. Through extensive computational experiments, we show that the metaheuristic procedure described in this paper is capable of finding high-quality solutions within a reasonable amount of computer time. Our main contribution is the development of a procedure that is more effective at handling PVRP instances with long planning horizons when compared to those proposed in the literature.     

An aggregate stochastic programming model for air traffic flow management

In this paper, we present an aggregate mathematical model for air traffic flow management (ATFM), a problem of great concern both in Europe and in the United States. The model extends previous approaches by simultaneously taking into account three important issues: (i) the model explicitly incorporates uncertainty in the airport capacities; (ii) it also considers the trade-off between airport arrivals and departures, which is a crucial issue in any hub airport; and (iii) it takes into account the interactions between different hubs.The level of aggregation proposed for the mathematical model allows us to solve realistic size instances with a commercial solver on a PC. Moreover it allows us to compute solutions which are perfectly consistent with the Collaborative Decision-Making (CDM) procedure in ATFM, widely adopted in the USA and which is currently receiving a lot of attention in Europe. In fact, the proposed model suggests the number of flights that should be delayed, a decision that belongs to the ATFM Authority, rather than assigning delays to individual aircraft.     

A fast solution method for the time-dependent orienteering problem

This paper introduces a fast solution procedure to solve 100-node instances of the time-dependent orienteering problem (TD-OP) within a few seconds of computation time. Orienteering problems occur in logistic situations were an optimal combination of locations needs to be selected and the routing between the selected locations needs to be optimized. In the time-dependent variant, the travel time between two locations depends on the departure time at the first location. Next to a mathematical formulation of the TD-OP, the main contribution of this paper is the design of a fast and effective algorithm to tackle this problem. This algorithm combines the principles of an ant colony system (ACS) with a time-dependent local search procedure equipped with a local evaluation metric. Additionally, realistic benchmark instances with varying size and properties are constructed. The average score gap with the known optimal solution on these test instances is only 1.4% with an average computation time of 0.5 seconds. An extensive sensitivity analysis shows that the performance of the algorithm is insensitive to small changes in its parameter settings.     

Subsampling bias and the best-discrepancy systematic cross validation

Statistical machine learning models should be evaluated and validated before putting to work.Conventional k-fold Monte Carlo cross-validation(MCCV)procedure uses a pseudo-random sequence to partition instances into k subsets,which usually causes subsampling bias,inflates generalization errors and jeopardizes the reliability and effectiveness of cross-validation.Based on ordered systematic sampling theory in statistics and low-discrepancy sequence theory in number theory,we propose a new k-fold cross-validation procedure by replacing a pseudo-random sequence with a best-discrepancy sequence,which ensures low subsampling bias and leads to more precise expected-prediction-error(EPE)estimates.Experiments with 156 benchmark datasets and three classifiers(logistic regression,decision tree and na?ve bayes)show that in general,our cross-validation procedure can extrude subsampling bias in the MCCV by lowering the EPE around 7.18%and the variances around 26.73%.In comparison,the stratified MCCV can reduce the EPE and variances of the MCCV around 1.58%and 11.85%,respectively.The leave-one-out(LOO)can lower the EPE around 2.50%but its variances are much higher than the any other cross-validation(CV)procedure.The computational time of our cross-validation procedure is just 8.64%of the MCCV,8.67%of the stratified MCCV and 16.72%of the LOO.Experiments also show that our approach is more beneficial for datasets characterized by relatively small size and large aspect ratio.This makes our approach particularly pertinent when solving bioscience classification problems.Our proposed systematic subsampling technique could be generalized to other machine learning algorithms that involve random subsampling mechanism.     

A new bilevel formulation for the vehicle routing problem and a solution method using a genetic algorithm

The Vehicle Routing Problem (VRP) is one of the most well studied problems in operations research, both in real life problems and for scientific research purposes. During the last 50 years a number of different formulations have been proposed, together with an even greater number of algorithms for the solution of the problem. In this paper, the VRP is formulated as a problem of two decision levels. In the first level, the decision maker assigns customers to the vehicles checking the feasibility of the constructed routes (vehicle capacity constraints) and without taking into account the sequence by which the vehicles will visit the customers. In the second level, the decision maker finds the optimal routes of these assignments. The decision maker of the first level, once the cost of each routing has been calculated in the second level, estimates which assignment is the better one to choose. Based on this formulation, a bilevel genetic algorithm is proposed. In the first level of the proposed algorithm, a genetic algorithm is used for calculating the population of the most promising assignments of customers to vehicles. In the second level of the proposed algorithm, a Traveling Salesman Problem (TSP) is solved, independently for each member of the population and for each assignment to vehicles. The algorithm was tested on two sets of benchmark instances and gave very satisfactory results. In both sets of instances the average quality is less than 1%. More specifically in the set with the 14 classic instances proposed by Christofides, the quality is 0.479% and in the second set with the 20 large scale vehicle routing problems, the quality is 0.826%. The algorithm is ranked in the tenth place among the 36 most known and effective algorithms in the literature for the first set of instances and in the sixth place among the 16 algorithms for the second set of instances. The computational time of the algorithm is decreased significantly compared to other heuristic and metaheuristic algorithms due to the fact that the Expanding Neighborhood Search Strategy is used.     

Approximating Promethee II’s net flow scores by piecewise linear value functions

Promethee II is a prominent method for multi-criteria decision aid (MCDA) that builds a complete ranking on a set of potential actions by assigning each of them a so-called net flow score. However, to calculate these scores, each pair of actions has to be compared, causing the computational load to increase quadratically with the number of actions, eventually leading to prohibitive execution times for large decision problems. For some problems, however, a trade-off between the ranking’s accuracy and the required evaluation time may be acceptable. Therefore, we propose a piecewise linear model that approximates Promethee II’s net flow scores and reduces the computational complexity (with respect to the number of actions) from quadratic to linear at the cost of some wrongly ranked actions. Simulations on artificial problem instances allow us to quantify this time/quality trade-off and to provide probabilistic bounds on the problem size above which our model satisfyingly approximates Promethee II’s rankings. They show, for instance, that for decision problems of 10,000 actions evaluated on 7 criteria, the Pearson correlation coefficient between the original scores and our approximation is of at least 0.97. When put in balance with computation times that are more than 7000 times faster than for the Promethee II model, the proposed approximation model represents an interesting alternative for large problem instances.     

A note on DEA vs principal component analysis: An improvement to Joe Zhu's approach

This paper considers a previous article published by Zhu in the European Journal of Operational Research which describes a joint use of data envelopment analysis (DEA) and principal component analysis (PCA) in ranking of decision making units (DMUs). In Zhu's empirical study, DEA and PCA yield a consistent ranking. However, this paper finds that in certain instances, DEA and PCA may yield inconsistent rankings. The PCA procedure adopted by Zhu is slightly modified in this article by incorporating other important features of ranking that Zhu has not considered. Numerical results reveal that both approaches show a consistency in ranking with DEA when the data set has a small number of efficient units. But, when a majority of the DMUs in the sample are efficient, only the modified approach produces consistent ranking with DEA.     

A heuristic for BILP problems: The Single Source Capacitated Facility Location Problem

In the Single Source Capacitated Facility Location Problem (SSCFLP) each customer has to be assigned to one facility that supplies its whole demand. The total demand of customers assigned to each facility cannot exceed its capacity. An opening cost is associated with each facility, and is paid if at least one customer is assigned to it. The objective is to minimize the total cost of opening the facilities and supply all the customers. In this paper we extend the Kernel Search heuristic framework to general Binary Integer Linear Programming (BILP) problems, and apply it to the SSCFLP. The heuristic is based on the solution to optimality of a sequence of subproblems, where each subproblem is restricted to a subset of the decision variables. The subsets of decision variables are constructed starting from the optimal values of the linear relaxation. Variants based on variable fixing are proposed to improve the efficiency of the Kernel Search framework. The algorithms are tested on benchmark instances and new very large-scale test problems. Computational results demonstrate the effectiveness of the approach. The Kernel Search algorithm outperforms the best heuristics for the SSCFLP available in the literature. It found the optimal solution for 165 out of the 170 instances with a proven optimum. The error achieved in the remaining instances is negligible. Moreover, it achieved, on 100 new very large-scale instances, an average gap equal to 0.64% computed with respect to a lower bound or the optimum, when available. The variants based on variable fixing improved the efficiency of the algorithm with minor deteriorations of the solution quality.     

Pareto ant colony optimization with ILP preprocessing in multiobjective project portfolio selection

One of the most important, common and critical management issues lies in determining the “best” project portfolio out of a given set of investment proposals. As this decision process usually involves the pursuit of multiple objectives amid a lack of a priori preference information, its quality can be improved by implementing a two-phase procedure that first identifies the solution space of all efficient (i.e., Pareto-optimal) portfolios and then allows an interactive exploration of that space. However, determining the solution space is not trivial because brute-force complete enumeration only solves small instances and the underlying NP-hard problem becomes increasingly demanding as the number of projects grows. While meta-heuristics in general provide an attractive compromise between the computational effort necessary and the quality of an approximated solution space, Pareto ant colony optimization (P-ACO) has been shown to perform particularly well for this class of problems. In this paper, the beneficial effect of P-ACO’s core function (i.e., the learning feature) is substantiated by means of a numerical example based on real world data. Furthermore, the original P-ACO approach is supplemented by an integer linear programming (ILP) preprocessing procedure that identifies several efficient portfolio solutions within a few seconds and correspondingly initializes the pheromone trails before running P-ACO. This extension favors a larger exploration of the search space at the beginning of the search and does so at a low cost.     

Heuristic,meta-heuristic and hyper-heuristic approaches for fresh produce inventory control and shelf space allocation

The allocation of fresh produce to shelf space represents a new decision support research area which is motivated by the desire of many retailers to improve their service due to the increasing demand for fresh food. However, automated decision making for fresh produce allocation is challenging because of the very short lifetime of fresh products. This paper considers a recently proposed practical model for the problem which is motivated by our collaboration with Tesco. Moreover, the paper investigates heuristic and meta-heuristic approaches as alternatives for the generalized reduced gradient algorithm, which becomes inefficient when the problem size becomes larger. A simpler single-item inventory problem is firstly studied and solved by a polynomial time bounded procedure. Several dynamic greedy heuristics are then developed for the multi-item problem based on the procedure for the single-item inventory problem. Experimental results show that these greedy heuristics are much more efficient and provide competitive results when compared to those of a multi-start generalized reduced gradient algorithm. In order to further improve the solution, we investigated simulated annealing, a greedy randomized adaptive search procedure and three types of hyper-heuristics. Their performance is tested and compared on a set of problem instances which are made publicly available for the research community.     

Analysis of a chance-constrained new product risk model with multiple customer classes

We consider the non-convex problem of minimizing a linear deterministic cost objective subject to a probabilistic requirement on a nonlinear multivariate stochastic expression attaining, or exceeding a given threshold. The stochastic expression represents the output of a noisy system featuring the product of mutually-independent, uniform random parameters each raised to a linear function of one of the decision vector’s constituent variables. We prove a connection to (i) the probability measure on the superposition of a finite collection of uncorrelated exponential random variables, and (ii) an entropy-like affine function. Then, we determine special cases for which the optimal solution exists in closed-form, or is accessible via sequential linear programming. These special cases inspire the design of a gradient-based heuristic procedure that guarantees a feasible solution for instances failing to meet any of the special case conditions. The application motivating our study is a consumer goods firm seeking to cost-effectively manage a certain aspect of its new product risk. We test our heuristic on a real problem and compare its overall performance to that of an asymptotically optimal Monte-Carlo-based method called sample average approximation. Numerical experimentation on synthetic problem instances sheds light on the interplay between the optimal cost and various parameters including the probabilistic requirement and the required threshold.     

A Beam Search approach for the optimization version of the Car Sequencing Problem

The Car Sequencing Problem (CSP) is a feasibility problem that has attracted the attention of the Constraint Programming community for a number of years now. In this paper, a new version (opt-CSP) that extends the original problem is defined, converting this into an optimization problem in which the goal is to satisfy the typical hard constraints. This paper presents a solution procedure for opt-CSP using Beam Search. Computational results are presented using public instances that verify the goodness of the procedure and demonstrate its excellent performance in obtaining feasible solutions for the majority of instances while satisfying the new constraints.     

Multiobjective scatter search for a commercial territory design problem

In this paper, a multiobjective scatter search procedure for a bi-objective territory design problem is proposed. A?territory design problem consists of partitioning a set of basic units into larger groups that are suitable with respect to some specific planning criteria. These groups must be compact, connected, and balanced with respect to the number of customers and sales volume. The bi-objective commercial territory design problem belongs to the class of NP-hard problems. Previous work showed that large instances of the problem addressed in this work are practically intractable even for the single-objective version. Therefore, the use of heuristic methods is the best alternative for obtaining approximate efficient solutions for relatively large instances. The proposed scatter search-based framework contains a diversification generation module based on a greedy randomized adaptive search procedure, an improvement module based on a relinked local search strategy, and a combination module based on a solution to an assignment problem. The proposed metaheuristic is evaluated over a variety of instances taken from literature. This includes a comparison with two of the most successful multiobjective heuristics from literature such as the Scatter Tabu Search Procedure for Multiobjective Optimization (SSPMO) by Molina et al. (INFORMS J. Comput. 19(1):91?C100, 2007), and the Non-dominated Sorting Genetic Algorithm (NSGA-II) by Deb et?al. (Parallel problem solving from nature ?C PPSN VI, Lecture notes in computer science, vol. 1917, Springer, Berlin, pp.?849?C858, 2000). Experimental work reveals that the proposed procedure consistently outperforms both heuristics, SSPMO and NSGA-II, on all instances tested.     

An Empirical Study of MAX-2-SAT Phase Transitions

The decision version of the maximum satisfiability problem (MAX-SAT) is stated as follows: Given a set S of propositional clauses and an integer g, decide if there exists a truth assignment that falsifies at most g clauses in S, where g is called the allowance for false clauses. We conduct an extensive experiment on over a million of random instances of 2-SAT and identify statistically the relationship between g, n (number of variables) and m (number of clauses). In our experiment, we apply an efficient decision procedure based on the branch-and-bound method. The statistical data of the experiment confirm not only the “scaling window” of MAX-2-SAT discovered by Chayes, Kim and Borgs, but also the recent results of Coppersmith et al. While there is no easy-hard-easy pattern for the complexity of 2-SAT at the phase transition, we show that there is such a pattern for the decision problem of MAX-2-SAT associated with the phase transition. We also identify that the hardest problems are among those with high allowance for false clauses but low number of clauses.     

A parallel branch and bound algorithm for solving large asymmetric traveling salesman problems

A parallel branch and bound algorithm that solves the asymmetric traveling salesman problem to optimality is described. The algorithm uses an assignment problem based lower bounding technique, subtour elimination branching rules, and a subtour patching algorithm as an upper bounding procedure. The algorithm is organized around a data flow framework for parallel branch and bound. The algorithm begins by converting the cost matrix to a sparser version in such a fashion as to retain the optimality of the final solution. Computational results are presented for three different classes of problem instances: (1) matrix elements drawn from a uniform distribution of integers for instances of size 250 to 10 000 cities, (2) instances of size 250 to 1000 cities that concentrate small elements in the upper left portion of the cost matrix, and (3) instances of size 300 to 3000 cities that are designed to confound neighborhood search heuristics.     

Optimally solving the alternative subgraphs assembly line balancing problem

Assembly line balancing problems (ALBP) consist of distributing the total workload for manufacturing any unit of the products to be assembled among the work stations along a manufacturing line as used in the automotive or the electronics industries. Usually, it is assumed that the production process is fixed, i.e., has been determined in a preceding planning step. However, this sequential planning approach is often suboptimal because the efficiency of the production process can not be evaluated definitely without knowing the distribution of work. Instead, both decisions should be taken simultaneously. This has led to the Alternative Subgraphs ALBP. We give an alternative representation of the problem, formulate an improved mixed-integer program and propose a solution approach based on SALOME, an effective branch-and-bound procedure for the well-known Simple ALBP. Computational experiments indicate that the proposed procedure is successful in finding optimal solutions for small- and medium-sized problem instances and rather good heuristic solutions for large-scaled instances.     

装备维修器材生产路径双目标问题优化决策方法研究

在装备维修器材供应保障中,针对精确保障背景下部队用户对器材保障精度的要求,构建了最小化总成本和最大化订单精准执行率的双目标优化决策模型。在ε-约束法框架内,开发可生成近似Pareto前沿的两阶迭代启发式算法,并采用模糊逻辑决策法选择符合决策者偏好的折中最优解。随机实例测试结果表明所提出的模型和算法可以很好地应用在双目标优化问题的研究中,并在求解不同规模实例时表现出优异的性能。     

Branch-and-price approach for the multi-skill project scheduling problem

This work introduces a procedure to solve the multi-skill project scheduling problem (MSPSP) (Néron and Baptista, International symposium on combinatorial, optimization (CO’2002), 2002). The MSPSP mixes both the classical resource constrained project scheduling problem and the multi-purpose machine model. The aim is to find a schedule that minimizes the completion time (makespan) of a project, composed of a set of activities. In addition, precedence relations and resources constraints are considered. In this problem, resources are staff members that master several skills. Thus, a given number of workers must be assigned to perform each skill required by an activity. Practical applications include the construction of buildings, as well as production and software development planning. We present a column generation approach embedded within a branch-and-price (B&P) procedure that considers a given activity and time-based decomposition approach. Obtained results show that the proposed B&P procedure is able to reach optimal solutions for several small and medium sized instances in an acceptable computational time. Furthermore, some previously open instances were optimally solved.     

Semi-Lagrangian relaxation applied to?the?uncapacitated facility location problem

We show how the performance of general purpose Mixed Integer Programming (MIP) solvers, can be enhanced by using the Semi-Lagrangian Relaxation (SLR) method. To illustrate this procedure we perform computational experiments on large-scale instances of the Uncapacitated Facility Location (UFL) problems with unknown optimal values. CPLEX solves 3 out of the 36 instances. By combining CPLEX with SLR, we manage to solve 18 out of the 36 instances and improve the best known lower bound for the other instances. The key point has been that, on average, the SLR approach, has reduced by more than 90% the total number of relevant UFL variables.