论文速递 | INFORMS Journal on Computing 2月文章精选

编者按

在本系列文章中，我们对顶刊《INFORMS Journal on Computing》于2025年2月份发布文章中进行了精选（共5篇），并总结其基本信息，旨在帮助读者快速洞察行业最新动态。本月IJOC发文聚焦设施选址、运输优化、工具更换及无人机-车辆路径，涵盖启发式/精确算法、分支规则与鲁棒优化。

推荐文章1

● 题目：The Balanced Facility Location Problem: Complexity and Heuristics

平衡设施点位置问题：复杂性和启发式

● 原文链接：https://pubsonline.informs.org/doi/abs/10.1287/ijoc.2024.0693

● 作者：Malena Schmidt，Bismark Singh

● 发布时间：2025-2-14

● 摘要：

A recent work proposes a new quadratic facility location model to address ecological challenges faced by policymakers in Bavaria, Germany. Building on this, we significantly extend our understanding of this new problem. We develop connections to traditional combinatorial optimization models and show that the problem is 𝒩⁢𝒫 hard. We then develop several classes of easy-to-implement heuristics to solve this problem. These are rooted in solving special cases of the generalized quadratic assignment problem as a subproblem; this subproblem is also 𝒩⁢𝒫 hard. On moderate-sized instances from Bavaria—that were previously intractable—our proposed heuristics compute feasible solutions that are 0.5% (on average) improved over the generic solution method in just over a minute (on average), even when the generic solver runs for 20,000 seconds. Larger instances show an improvement of 5% (on average) compared with the generic solution method in an average of 410 seconds.

最近的一项研究提出了一种新的二次设施选址模型，以应对德国巴伐利亚地区决策者所面临的生态挑战。在此基础上，我们显著扩展了对这一新问题的理解。我们将该问题与传统的组合优化模型建立了联系，并证明该问题是 𝒩⁢𝒫难的。随后，我们开发了若干类易于实现的启发式算法来求解这一问题。这些算法基于将广义二次指派问题的特例作为子问题来求解，而该子问题同样是𝒩⁢𝒫难的。

在来自巴伐利亚的中等规模实例中——这些实例此前被认为难以求解——我们提出的启发式算法能够在平均仅需一分钟多的时间内计算出可行解，其结果相比通用求解方法平均提高了 0.5%，而即使通用求解方法运行 20,000 秒也难以达到同样效果。对于更大规模的实例，我们的方法相比通用求解方法平均提高了 5%，而求解时间平均为 410 秒。

推荐文章2

● 题目：Non-Monotonicity of Branching Rules with Respect to Linear Relaxations

分支规则关于线性松弛的非单调性

● 原文链接：https://pubsonline.informs.org/doi/abs/10.1287/ijoc.2024.0709

● 作者：Prachi Shah, Santanu S.Dey，Marco Molinaro

● 发布时间：2025-2-19

● 摘要：

Modern mixed-integer programming solvers use the branch-and-cut framework, where cutting planes are added to improve the tightness of the linear programming (LP) relaxation, with the expectation that the tighter formulation would produce smaller branch-and-bound trees. In this work, we consider the question of whether adding cuts will always lead to smaller trees for a given fixed branching rule. We formally call such a property of a branching rule monotonicity. We prove that any branching rule which exclusively branches on fractional variables in the LP solution is nonmonotonic. Moreover, we present a family of instances where adding a single cut leads to an exponential increase in the size of full strong branching trees, despite improving the LP bound. Finally, we empirically attempt to estimate the prevalence of nonmonotonicity in practice while using full strong branching. We consider randomly generated multidimensional knapsacks tightened by cover cuts as well as instances from the MIPLIB 2017 benchmark set for the computational experiments. Our main insight from these experiments is that if the gap closed by cuts is small, change in tree size is difficult to predict, and often increases, possibly due to inherent nonmonotonicity. However, when a sufficiently large gap is closed, a significant decrease in tree size may be expected.

现代混合整数规划求解器采用分支切割框架，通过添加割平面来提高线性规划（LP）松弛的紧密度，期望更紧的模型能生成更小的分支定界树。本文探讨在固定分支规则下，添加割平面是否总能缩小搜索树规模。我们正式将分支规则的这种性质定义为单调性（monotonicity）。我们证明，任何仅基于LP解中分数变量进行分支的规则均具有非单调性（nonmonotonic）。进一步，我们构建了一类实例：即使添加单个割平面能够提升LP下界，却会导致完全强分支树的规模呈指数级增长。最后，我们通过实验评估完全强分支规则在实践中的非单调性表现。计算实验使用覆盖割收紧的随机生成多维背包问题以及MIPLIB 2017基准测试集实例。实验的核心发现是：当割平面封闭的间隙较小时，树规模变化难以预测且常出现增大，这可能源于固有的非单调性；但当封闭的间隙足够大时，树规模通常会有显著缩减。

推荐文章3

● 题目：Iterated Inside Out: A New Exact Algorithm for the Transportation Problem

向内迭代：求解运输问题的一种新的精确算法

● 原文链接：https://pubsonline.informs.org/doi/abs/10.1287/ijoc.2024.0642

● 作者：Roberto Bargetto, Federico Della Croce,Rosario Scatamacchia

● 发布时间：2025-2-20

● 摘要：

We propose a novel exact algorithm for the transportation problem, one of the paradigmatic network optimization problems. The algorithm, called Iterated Inside Out, requires as input a basic feasible solution and is composed of two main phases that are iteratively repeated until an optimal basic feasible solution is computed. In the first “inside” phase, the algorithm progressively improves upon a given basic solution by increasing the value of several nonbasic variables with negative reduced cost. This phase typically outputs a nonbasic feasible solution interior to the constraint set polytope. The second “out” phase moves in the opposite direction by iteratively setting to zero several variables until a new improved basic feasible solution is reached. Extensive computational tests show that the proposed approach strongly outperforms all versions of network and linear programming algorithms available in the commercial solvers CPLEX and Gurobi and other exact algorithms available in the literature.

我们提出了一种用于运输问题的新型精确算法，该问题是典型的网络优化问题之一。该算法称为 "Iterated Inside Out"，需要基本可行解决方案作为输入，它由两个主要阶段组成，这两个阶段会迭代重复，直到计算出最理想的基本可行解决方案。在第一个 “inside” 阶段，该算法通过增加几个非基本变量的值来逐步改进给定的基本解，同时成本为负降低。该阶段通常会生成一个处于约束集合多面体内部的非基可行解。第二个 “out” 阶段向相反方向移动，将几个变量迭代设置为零，直到达到新的改进的基本可行解决方案。广泛的计算测试表明，所提出的方法明显优于商业求解器 CPLEX 和 Gurobi 中可用的所有版本的网络和线性规划算法，以及文献中可用的其他精确算法。

推荐文章4

● 题目：A Computational Study of the Tool Replacement Problem

工具更换问题的计算研究

● 原文链接：https://pubsonline.informs.org/doi/abs/10.1287/ijoc.2023.0474

● 作者：Yuzhuo Qiu, Mikhail Cherniavskii, Boris Goldengorin，Panos M.Pardalos

● 发布时间：2025-2-25

● 摘要：

In the Tool Replacement Problem (TRP) for the given sequence of jobs, we consider a discretized interval where each point in time corresponds to a specific job and its collection of tools sufficient to complete that job. A passive interval w.r.t. a specific tool is an interval where that tool is not used at any point within that interval but is used at the boundary points in time. The TRP aims to find a loading schedule of tools (tool switches) that minimizes the total number of tool loadings in the magazine. Based on the concept of a passive interval, we introduce our reformulation of the TRP as follows. The minimum total number of tool loadings (switches) in the TRP is equal to the difference between the total number of tools for all scheduled jobs with tool repetitions and the maximum total number of passive intervals. We solve the TRP to optimality by designing and implementing two algorithms: one for finding the optimal objective function value (Insertion Greedy Algorithm (IGA)) and the other (To Full Magazine (ToFullMag) algorithm) for finding an optimal solution, that is, an optimal sequence of tool loadings. We apply our reformulation of the TRP to design the IGAfull algorithm starting with IGA and continuing with ToFullMag. The  IGAfull achieves the best possible running time and thus settles the computational complexity of TRP. We prove that IGAfull outperforms the most popular Keep Tool Needed Soonest (KTNS) algorithm by at least an order of magnitude in terms of CPU time. Moreover, after replacing the KTNS algorithm by IGAfull within the state-of-the-art Hybrid Genetic Searches heuristic for solving the job Scheduling and tool Switching Problem (SSP), our computational study shows the reduction of CPU times by at least an order of magnitude for medium- and large-scale SSP data sets.

在给定作业序列的工具更换问题（TRP）中，我们将时间离散化，其中每个时间点对应于一个具体的作业及其所需完成该作业的工具集合。关于特定工具的“被动区间”是指该工具在该区间内未被使用，但在区间的边界时间点上被使用。TRP 的目标是找到一个工具装载（工具切换）调度，以最小化工具装载（切换）的总次数。基于被动区间的概念，我们引入了 TRP 的新表述。TRP 的最小工具装载（切换）次数等于所有已调度作业中工具总使用次数与所有被动区间总数之间的差值。我们通过设计并实现两个算法来求解 TRP 达到最优解：一个用于寻找最优目标函数值的插入式贪婪算法（IGA），另一个是用于寻找最优装载序列的 To Full Magazine 算法（ToFullMag）。我们应用对 TRP 的重新表述来设计IGAfull算法，首先采用 IGA，然后使用 ToFullMag。 IGAfull实现了最佳的运行时间，从而解决了 TRP 的计算复杂性问题。我们证明IGAfull至少在 CPU 时间方面优于当前最流行的 Keep Tool Needed Soonest（KTNS）算法一个数量级。此外，将 KTNS 替换为IGAfull并结合当前最先进的混合遗传搜索算法（Hybrid Genetic Searches）来解决作业调度与工具切换问题（SSP），我们的计算实验表明，在中、大型 SSP 数据集上，CPU 时间减少了至少一个数量级。

推荐文章5

● 题目：A Branch-and-Price Algorithm for Robust Drone-Vehicle Routing Problem with Time Windows

带时间窗的鲁棒无人机-车辆路径问题的分支定价算法

● 原文链接：https://pubsonline.informs.org/doi/abs/10.1287/ijoc.2023.0484

● 作者：Jaegwan Joo, Chunamok Lee

● 发布时间：2025-2-28

● 摘要：

This paper considers a cooperative routing problem in which trucks and multiple drones serve a set of customers collaboratively. A truck can operate as a drone station, dispatching and collecting multiple drones for nearby customers to overcome the drone’s short operation range. Each customer has a time window, so either a truck or a drone must serve the customer within the time window. The travel time uncertainties of the truck and drone are addressed by adopting the robust optimization approach. We first present a compact mathematical formulation for the problem. Then, we develop a decomposition approach based on the branch-and-price framework. After defining extended variables for trucks and drones separately, we decompose the column generation subproblem into two optimization problems, resulting in a two-phase column generation algorithm. We also develop a heuristic algorithm based on the proposed column generation scheme for larger instances. The results of numerical experiments, including real-life benchmark instances, show that the proposed algorithm outperforms the state-of-the-art mixed-integer programming solver.

本文考虑了一个协同路径规划问题，其中卡车和多架无人机协同为一组客户提供服务。卡车可以作为无人机的移动平台，负责部署和回收多个无人机，从而克服无人机续航范围有限的问题。每个客户都有一个服务时间窗口，因此卡车或无人机必须在该时间窗口内完成服务。为应对卡车和无人机行驶时间的不确定性，本文采用了鲁棒优化方法。我们首先为该问题提出了紧凑的数学模型。随后，基于分支定价框架，设计了一种分解求解方法。通过分别为卡车和无人机定义扩展变量，我们将列生成子问题分解为两个优化子问题，形成一个双阶段列生成算法。此外，我们还基于该列生成方案设计了一种启发式算法，以处理大规模实例。包括真实基准实例在内的数值实验结果表明，所提出的算法优于最先进的混合整数规划求解器。