Asia-Pacific Journal of Operational Research

In any airline’s schedule development process, aircraft rotations must be planned for individual fleet types after fleet assignment. The aircraft rotation plans must conform to stringent maintenance requirements and this problem can be formulated as a periodic routing problem on an Eulerian graph. We analyze the computational complexity of developing maintenance rotations when some overnighting aircraft may not have sufficient time on the ground to complete extended maintenance (referred to as a maintenance infeasibility). The paper also provides a theoretical analysis of heuristics for the aircraft maintenance rotation problem with maintenance infeasibilities.

In this paper, we propose an inexact majorized symmetric Gauss–Seidel (sGS) alternating direction method of multipliers (ADMM) with indefinite proximal terms for multi-block convex composite programming. This method is a specific form of the inexact majorized ADMM which is further proposed to solve a general two-block separable optimization problem. The new methods adopt certain relative error criteria to solve the involving subproblems approximately, and the step-sizes allow to choose in the scope $(0,(1+\sqrt{5})/2)$. Under more general conditions, we establish the global convergence and Q-linear convergence rate of the proposed methods.

In this paper, we consider a class of stochastic variational inequality problems (SVIPs). Different from the classical variational inequality problems, the SVIP contains a mathematical expectation, which may not be evaluated in an explicit form in general. We combine a hybrid Newton method for deterministic cases with an unconstrained optimization reformulation based on the well-known D-gap function and sample average approximation (SAA) techniques to present an SAA-based hybrid Newton method for solving the SVIP. We show that the level sets of the approximation D-gap function are bounded. Furthermore, we prove that the sequence generated by the hybrid Newton method converges to a solution of the SVIP under appropriate conditions, and some numerical experiments are presented to prove the effectiveness and competitiveness of the hybrid Newton method. Finally, we apply this method to solve two specific traffic equilibrium problems.

Online-to-offline (O2O) on-demand services require one-hour delivery and the demands vary substantially within one day. The capacity plans in the O2O industry evolve into three main modes: (i) in-house drivers only; (ii) full-time and part-time crowd-sourcing drivers; (iii) a mix of in-house and crowd-sourcing drivers. For current capacity plans, two issues remain unclear for both academia and industry. First, what is the optimal staffing decision when considering the behaviors of crowd-sourcing drivers. Second, how to choose from different capacity plans to match different operation strategies and market environments. To address these questions, we build an M/M/n queueing model to optimize the staffing decision with the aim of minimizing the total operation costs. Incentive mechanisms for both customers and crowd-sourcing drivers are crafted to improve their loyalty towards the O2O platform, in order to better manage capacity. Moreover, we apply a real dataset from one of the largest O2O platforms in China to verify our model. Our analyses show that adding flexibility — capacity-type flexibility and agent flexibility — to the O2O on-demand logistics system can help control costs and maintain a high service level. Furthermore, conditions in which different capacity plans match with different operation strategies and market environments are proposed.

The price risk of fresh agricultural products has been a significant topic in recent years. Taking the two-level fresh agricultural product supply chain as the research object, this paper studies the optimal ordering and coordination of supply chain based on two-period price, wholesale price and option contract. The optimal order decision of the retailer at the single period price and the optimal decision corresponding to the supplier are obtained when the output of the supplier is uncertain under decentralized decision-making. The range of penalty cost parameter that avoids supplier default is also obtained. The effect of two-period price on the optimal order decision and supply chain profits is discussed when the production yield of the supplier is fixed. Cost-sharing contract is introduced to increase the order quantity and achieve coordination because the option contract cannot completely make the supply chain coordination with two-period price. This paper provides a low-cost approach that can be applied in fresh agricultural supply chain to solve financing and order problems.

The primary purpose of this paper is to aggregate the overall rating based on guests’ online ratings by performing a social choice analysis of online hotel rating. Specifically, we first define the individual preference as the subjective judgement on the important order of the dimensions of online hotel rating, then quantify the individual preference through an analytical approach, the pessimistic and optimistic results of which are balanced by the Hurwicz criterion approach, lastly formulate the social choice result by means of the ideal-point concept. An empirical study using the real data collected from Trip.com is conducted to show the applicability and superiority of our methodology.

The main aim of this paper is to study second-order sensitivity analysis in parametric vector optimization problems. We prove that the proper perturbation maps and the proper efficient solution maps of parametric vector optimization problems are second-order composed proto-differentiable under some appropriate qualification conditions. Some examples are provided to illustrate our results.

We deal with the critical node problem (CNP) in a graph $G$, in which a given number $K$ of nodes are removed to minimize the connectivity of the residual graph in some sense. Several ways to minimize some connectivity measurement have been proposed, including minimizing the connectivity index(MinCI), maximizing the number of components, minimizing the maximal component size. We propose two classes of CNPs by combining the above measurements together. The objective is to minimize the sum of connectivity indexes and the total degrees in the residual graph. The CNP with an upper-bound $M$ on the maximal component size is denoted by MSCID-CS and the one with an extra upper-bound $P$ on the number of components is denoted by MSCID-CSN. They are generalizations of the MinCI, which has been shown NP-hard for general graphs. In particular, we study the case where $G$ is a tree. Two dynamic programming algorithms are proposed to solve the two classes of CNPs. The time complexities of the algorithms for MSCID-CS and MSCID-CSN are $O(n^2{K}^2{M}^2)$ and $O(n^2{K}^2{M}^2{P}^2)$, respectively, where $n$ is the number of nodes in $G$. Computational experiments are presented which show the effectiveness of the algorithms.

In recent times, a variety of nonconvex feasibility problems have been solved empirically by employing Douglas–Rachford (DR) splitting methods. However, this theory is not adequate in explaining the observed success and is more concerned with the local convergence. In this paper, we study the convergence of the DR splitting method for finding a point of intersection of a closed ball and a sphere in the $n$-dimensional Euclidean spaces. Also, we provide the region for the global convergence of the DR splitting method.