EURO Journal on Computational Optimization最新文献

In civil aircraft, two partially redundant hydraulic circuits typically power various systems. During assembly, a critical phase involves simultaneously rinsing and purging these hydraulic circuits using loops. Precedence constraints are necessary to prevent the recontamination of already rinsed loops, leading to increased rinsing time. This paper presents this problem as a unique instance of the Resource Constrained Parallel Machine Scheduling Problem, where each circuit represents a machine, pipe loops to be rinsed represent jobs, and machines share a hydraulic power source. For two dedicated processors and a single resource, an optimal schedule minimizing the makespan can be generated in polynomial time. However, due to the requirement of rinsing certain pipe loops on a circuit before others, there are precedence constraints between some jobs within the same circuit. By employing a reduction of the 3-partition problem, we demonstrate that this situation results in a problem that is NP-hard in the strong sense. We evaluate several Mixed-Integer Linear Programming and Constraint Programming formulations of the problem, using Cplex, CPO, Gurobi, and Z3, against several proposed heuristics. Given that the size of the instances we need to solve exceeds what can be solved in acceptable time by solvers, we propose a heuristic and compare its performance with the optimum.

Due to the continued success of machine learning and deep learning in particular, supervised classification problems are ubiquitous in numerous scientific fields. Training these models typically involves the minimization of the empirical risk over large data sets along with a possibly non-differentiable regularization. In this paper, we introduce a stochastic gradient method for the considered classification problem. To control the variance of the objective's gradients, we use an automatic sample size selection along with a variable metric to precondition the stochastic gradient directions. Further, we utilize a non-monotone line search to automatize step size selection. Convergence results are provided for both convex and non-convex objective functions. Extensive numerical experiments verify that the suggested approach performs on par with state-of-the-art methods for training both statistical models for binary classification and artificial neural networks for multi-class image classification. The code is publicly available at https://github.com/koblererich/lisavm.

We develop a Design and Analysis of the Computer Experiments (DACE) approach to the stochastic unit commitment problem for power systems with significant renewable integration. For this purpose, we use a two-stage stochastic programming formulation of the stochastic unit commitment-economic dispatch problem. Typically, a sample average approximation of the true problem is solved using a cutting plane method (such as the L-shaped method) or scenario decomposition (such as Progressive Hedging) algorithms. However, when the number of scenarios increases, these solution methods become computationally prohibitive. To address this challenge, we develop a novel DACE approach that exploits the structure of the first-stage unit commitment decision space in a design of experiments, uses features based upon solar generation, and trains a multivariate adaptive regression splines model to approximate the second stage of the stochastic unit commitment-economic dispatch problem. We conduct experiments on two modified IEEE-57 and IEEE-118 test systems and assess the quality of the solutions obtained from both the DACE and the L-shaped methods in a replicated procedure. The results obtained from this approach attest to the significant improvement in the computational performance of the DACE approach over the traditional L-shaped method.

We introduce a heuristic rule for calculating the stepsize in the subgradient method for unconstrained convex nonsmooth optimization which, unlike the classic approach, is based on retaining some information from previous iteration. The rule is inspired by the well known two-point stepsize by Barzilai and Borwein (BB) [6] for smooth optimization and it coincides with (BB) in case the function to be minimised is convex quadratic.

Under the use of appropriate safeguards we demonstrate that the method terminates at a point that satisfies an approximate optimality condition.

The proposed approach is tested in the framework of Lagrangian relaxation for integer linear programming where the Lagrangian dual requires maximization of a concave and nonsmooth (piecewise affine) function. In particular we focus on the relaxation of the Minimum Spanning Tree problem with Conflicting Edge Pairs (MSTC). Comparison with classic subgradient method is presented. The results on some widely used academic test problems are provided too.

We revisit a formulation for the simple plant facility location and p-median problems introduced by Cornuéjols, Nemhauser and Wolsey (1980). Despite being the smallest known formulation regarding the number of variables, this formulation is barely used or cited in the literature. Here, we reintroduce the formulation for the p-median problem from a different perspective, resulting from the intersection of a selection problem with an additional family of optimality constraints to define the costs correctly. An alternative proof that the linear relaxation of the formulation is equivalent to the linear relaxation of the well-known classical formulation is provided. By exploring the optimality constraints we discuss approaches to derive bounds for large-size instances. These approaches are based on relaxations obtained by eliminating optimality constraints and can be seen as a simple matheuristic to solve large size instances. In particular, we characterize relaxations which provide the optimal solution, and therefore, can be seen as new formulations for the p-median problem. Computational tests are reported showing that the renewed formulation can be used efficiently to solve p-median instances.

Given that about half of the produced energy in the world is consumed in industries, there has been an increasing concern about optimizing energy consumption in manufacturing sectors. As one of the most effective ways, proper production scheduling to reduce energy consumption is of crucial importance among researchers and manufacturers. This paper addresses an unrelated parallel machine energy-efficient scheduling problem with sequence-dependent setup times by considering different energy consumption tariffs. The setup times are studied in two modes: disjointed from/jointed to processing time. For each one of these problems, two mixed-integer linear programming models have been formulated. The presented models for the problem with setup time disjointed from processing time can solve up to 16 machines and 45 jobs. In contrast, this capability is changed to 20 machines and 40 jobs for processing time jointed to the setup time problem. Furthermore, a fix and relax heuristic algorithm is presented for large-size instances, which can solve instances of up to 20 machines and 100 jobs for each of the two considered problems.

We investigate a location-allocation-routing problem where trucks deliver goods from a central production facility to a set of warehouses with fixed locations and known demands. Due to limited capacities congestion occurs and results in queueing problems. The location of the center is determined to maximize the utilization of the given resources (measured in throughput) and the minimal number of trucks is determined to satisfy the overall demand generated by the warehouses. Main results for this integrated decision problem on strategic and tactical/operational level are: (i) The location decision is reduced to a standard Weber problem with weighted distances. (ii) The joint decision for location and fleet size is separable. (iii) The location of the center is robust against perturbations of several system parameters on the operational/tactical level. Additionally, we consider minimization of travel times as optimization target. By numerical examples we demonstrate the consequences of neglecting available information on long-term (rough) demand structure.

Price differentiation is a common strategy in many markets. In this paper, we study a static multiproduct price optimization problem with demand given by a discrete mixed multinomial logit model. By considering a mixed logit model that includes customer specific variables and parameters in the utility specification, our pricing problem reflects well the discrete choice models used in practice. To solve this pricing problem, we design an efficient iterative optimization algorithm that asymptotically converges to the optimal solution. To this end, a linear optimization (LO) problem is formulated, based on the trust-region approach, to find a “good” feasible solution and approximate the problem from below. A convex optimization problem is designed using a convexification technique to approximate the optimization problem from above. Then, using a branching method, we tighten the optimality gap. The effectiveness of our algorithm is illustrated on several cases, and compared against solvers and existing state-of-the-art methods in the literature.