Asia-Pacific Journal of Operational Research

In this work, we face a variant of the capacitated lot sizing problem. This is a classical problem addressing the issue of aggregating lot sizes for a finite number of discrete periodic demands that need to be satisfied, thus setting up production resources and eventually creating inventories, while minimizing the overall cost. In the proposed variant we take into account lifetime constraints, which model products with maximum fixed shelflives due to several possible reasons, including regulations or technical obsolescence. We propose four formulations, derived from the literature on the classical version of the problem and adapted to the proposed variant. An extensive experimental phase on two datasets from the literature is used to test and compare the performance of the proposed formulations.

Grain price forecasting is significant for all market participants in managing risks and planning operations. However, precise forecasting of price series is difficult because of the inherent stochastic behavior of grain price. In this paper, a novel hybrid stochastic method for grain price forecasting is introduced. The proposed method combines decomposed stochastic time series processes with artificial neural networks. The initial parameters of the hybrid stochastic model are optimized by a random search using a genetic algorithm. The proposed method is finally validated in China’s national grain market and compared with several recent price forecasting models. Results indicate that the proposed hybrid stochastic method provides a satisfactory forecasting performance in grain price series.

In this paper, we consider the online single machine scheduling problem to minimize the maximum starting time of the jobs. For the non-preemptive model, we show that there is no determined or randomized online algorithm with a bounded competitive ratio. For the preemption-resume model, we show that the well-known SRPT rule yields an optimal schedule. For the preemption-restart model, we show that any determined online algorithm has a competitive ratio of at least 2 and present an online algorithm with the best-possible competitive ratio of 2.

The “Dynamic-network” version of cost efficiency measurement in Data Envelopment Analysis (DEA) is proposed in this paper. The classical DEA models ignore operations of individual processes within a system; moreover, they compute efficiency at the same time. Therefore, we suggest a relational model to estimate cost efficiency in static network structures. Also, we incorporate the dynamic effect in network structures. The proposed models here evaluate the overall efficiency over the whole periods and indicate it as a weighted average of period efficiencies. The main advantage revealed in this study is recognition of: which divisions at what periods caused the inefficiency of the system, the internal activities of the system over time, considered; moreover, the results obtained here is applicable in, improving the performance of the system. A case study of Iranian banking industry is used to show the applicability of the approach.

We introduce and study the rescheduling on a single machine to minimize the maximum lateness under the sequence disruptions of original jobs. In the problem, each original job $J_j$ has a constraint disruption on its sequence respect to an original optimal schedule $\pi$, i.e., $D_j(\pi )\le k_j$. That is, if $J_j$ is the $x$th job in $\pi$, then it is required that the position index $y$ of $J_j$ in a schedule for all jobs satisfies $|y-x|\le k_j$. By introducing the positive sequence disruption $D_j^{+}(\pi )$ and the negative sequence disruption $D_j^{-}(\pi )$, three problems are considered in this paper: problem (P1) is $1|D_j(\pi )\le k_j|L_{\mathrm{\text{max}}}$, problem (P2) is $1|D_j^{+}(\pi )\le k_j|L_{\mathrm{\text{max}}}$, and problem (P3) is $1|D_j^{+}(\pi )\le k_j,D_j^{-}(\pi )=0|L_{\mathrm{\text{max}}}$. We show that the three problems are equivalent and can be solved in $O(nlogn)$ time. Then we study an extension of problem (P2): $1|\text{prec},{\sigma }^{-1}(j)\le k_j|f_{\mathrm{\text{max}}}$. We show that the extended problem can be solved in $O(n^2)$ time.

We propose an efficient algorithm for computing the equilibrium of a capital asset pricing model with heterogeneous investors and short-sale constraints. We show that the equilibrium prices of the risky assets in the model are proportional to the Lagrangian multipliers of an equivalent dual formulation of the problem. Based on this observation, we derive sufficient conditions to guarantee the existence and uniqueness of equilibrium and prove the convergence of the algorithm. Numerical examples are also provided to illustrate the algorithm.

A firm sets up his facilities including manufacturing/remanufacturing plants and distribution/collection centers, incorporating an existing closed-loop supply chain (CLSC) network. The entering firm has to compete with the existing firms in the existing network. The entering firm behaves as the leader of a Stackelberg game while the existing firms in the existing network are followers. We assume that the entering firm can anticipate the existing firms’ reaction to his potential location decision before choosing his optimal policy. We use a CLSC network equilibrium model in which the decision makers are faced with multiple objectives to capture the existing firms’ reaction. A mathematical programming model with equilibrium constraints is developed for this competitive CLSC network design problem by taking into account the market competition existing in the decentralized CLSC network. A solution method is developed by integrating Genetic algorithm with an inexact logarithmic-quadratic proximal augmented Lagrangian method. Finally, numerical examples and the related results are studied for illustration purpose.

In the traditional DEA model, each DMU maximizes its efficiency with the most favorable weights. This leads to flexibility and unreality of input and output weights. Subsequently, it is unfair to compare and rank the efficiencies of different DMUs obtained on the basis of these weights. In this paper, we propose a novel approach to determine a common set of weights with more consensus to evaluate and rank the performance of all DMUs by weighting the rescaled weights based on the degree of consensus, where the weights obtained from DEA are rescaled for comparison among DMUs. Moreover, to overcome the non-uniqueness of the weights, a novel secondary goal is developed based on the agreement between self-evaluation and peer-evaluation. In addition, the restriction of weights is taken into account to avoid trivial weights. Finally, an example of 14 international passenger airlines is used to illustrate the performance and credibility of our proposed method.