Pub Date : 2025-12-26DOI: 10.1016/j.orp.2025.100374
Balázs Szabó , Zsolt Nemeskéri
This paper presents a hybrid continuous–discrete framework for optimizing production, pricing, quality, and learning investments under organizational learning, defined as efficiency gains through accumulated experience. Extending prior work on price-quality dynamics and production optimization, the model integrates real-time operational decisions with periodic strategic planning to address dynamic market challenges. It incorporates a stochastic initial learning rate to capture uncertainty in learning processes. In contrast to conventional discrete models that may exhibit qualitatively different dynamics (Vörös 2021), the hybrid framework is constructed to converge rigorously to the continuous-time optimum, ensuring consistent decision-making across planning horizons. Monte Carlo simulations provide insights into production increases, gradual price reductions, and quality improvements, with profit gains up to 14.7% in scenarios with high learning and quality sensitivity. The framework offers actionable guidance for industries like automotive, electronics, aerospace, and pharmaceuticals, enhancing cost efficiency and competitiveness through integrated strategic and operational optimization, though without explicit competitive interactions.
{"title":"Integrated production and quality dynamics under organizational learning: A hybrid continuous–discrete framework","authors":"Balázs Szabó , Zsolt Nemeskéri","doi":"10.1016/j.orp.2025.100374","DOIUrl":"10.1016/j.orp.2025.100374","url":null,"abstract":"<div><div>This paper presents a hybrid continuous–discrete framework for optimizing production, pricing, quality, and learning investments under organizational learning, defined as efficiency gains through accumulated experience. Extending prior work on price-quality dynamics and production optimization, the model integrates real-time operational decisions with periodic strategic planning to address dynamic market challenges. It incorporates a stochastic initial learning rate to capture uncertainty in learning processes. In contrast to conventional discrete models that may exhibit qualitatively different dynamics (Vörös 2021), the hybrid framework is constructed to converge rigorously to the continuous-time optimum, ensuring consistent decision-making across planning horizons. Monte Carlo simulations provide insights into production increases, gradual price reductions, and quality improvements, with profit gains up to 14.7% in scenarios with high learning and quality sensitivity. The framework offers actionable guidance for industries like automotive, electronics, aerospace, and pharmaceuticals, enhancing cost efficiency and competitiveness through integrated strategic and operational optimization, though without explicit competitive interactions.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"16 ","pages":"Article 100374"},"PeriodicalIF":3.7,"publicationDate":"2025-12-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145897988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-20DOI: 10.1016/j.orp.2025.100375
Hassan Mohammad , Sulaiman Mohammed Ibrahim , Hayoung Choi , Rabiu Bashir Yunus
This paper presents a novel diagonal Hestenes–Stiefel conjugate gradient (DHS-CG) algorithm for solving large-scale unconstrained optimization problems. Building on the approaches presented in Dong et al. (2015) and Mohammad and Santos (2018), the proposed algorithm integrates a computationally efficient diagonal Hessian approximation into the HS-CG scheme. The algorithm constructs search directions that enjoy sufficient descent, trust region, and conjugacy properties without imposing restrictive line search conditions, thus ensuring efficiency robustness under Wolfe and Armijo-type line search strategies. We establish the global convergence and iterative complexity of the proposed algorithm under standard bounds and Lipschitz gradient assumptions. Extensive numerical experiments on benchmark problems, including large-scale cases with up to 500,000 variables, demonstrated that DHS-CG consistently outperformed state-of-the-art CG variants, such as CG_DESCENT and classical Hestenes–Stiefel (HS). In both low and high dimensional settings, the proposed algorithm achieves faster convergence, fewer function evaluations, and lower CPU time, highlighting its suitability for high performance scientific and engineering computations.
{"title":"A diagonal Hestenes–Stiefel conjugate gradient algorithm with iterative complexity analysis and its application in robotic model","authors":"Hassan Mohammad , Sulaiman Mohammed Ibrahim , Hayoung Choi , Rabiu Bashir Yunus","doi":"10.1016/j.orp.2025.100375","DOIUrl":"10.1016/j.orp.2025.100375","url":null,"abstract":"<div><div>This paper presents a novel diagonal Hestenes–Stiefel conjugate gradient (DHS-CG) algorithm for solving large-scale unconstrained optimization problems. Building on the approaches presented in Dong et al. (2015) and Mohammad and Santos (2018), the proposed algorithm integrates a computationally efficient diagonal Hessian approximation into the HS-CG scheme. The algorithm constructs search directions that enjoy sufficient descent, trust region, and conjugacy properties without imposing restrictive line search conditions, thus ensuring efficiency robustness under Wolfe and Armijo-type line search strategies. We establish the global convergence and iterative complexity of the proposed algorithm under standard bounds and Lipschitz gradient assumptions. Extensive numerical experiments on benchmark problems, including large-scale cases with up to 500,000 variables, demonstrated that DHS-CG consistently outperformed state-of-the-art CG variants, such as CG_DESCENT and classical Hestenes–Stiefel (HS). In both low and high dimensional settings, the proposed algorithm achieves faster convergence, fewer function evaluations, and lower CPU time, highlighting its suitability for high performance scientific and engineering computations.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"16 ","pages":"Article 100375"},"PeriodicalIF":3.7,"publicationDate":"2025-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145897989","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01DOI: 10.1016/j.orp.2025.100369
Ayoub Abusalih, Zeyu Liu
The freight transportation sector is critical to the economic prosperity in the US, but it also constitutes a major source of carbon emissions. Although intermodal transportation has been shown to increase operating efficiency and reduce carbon emissions, research on infrastructural support for intermodal transportation is still insufficient. In this study, we establish a mixed integer programming model to jointly optimize strategic infrastructure development decisions and freight transportation decisions over a long horizon. Our model features a mixture of traditional single-mode facilities and hybrid hubs that facilitate rail–water transportation integration. A branch-and-cut decomposition algorithm is developed to solve large-scale problems. We collect real-world freight, infrastructure, and operations data to conduct computational studies on the model performance and algorithm efficiency. We provide insights for practitioners to address infrastructure planning and budgetary concerns. A case study using the established model at the national scale shows that well-optimized transportation infrastructure investment could have over 300% return during a 25-year horizon. In addition, fully capitalizing on the maturing clean vehicle technologies could reduce carbon emissions by 73.56 million tons at an annual rate.
{"title":"A cost and emission optimization framework for strategic intermodal freight transportation infrastructure development","authors":"Ayoub Abusalih, Zeyu Liu","doi":"10.1016/j.orp.2025.100369","DOIUrl":"10.1016/j.orp.2025.100369","url":null,"abstract":"<div><div>The freight transportation sector is critical to the economic prosperity in the US, but it also constitutes a major source of carbon emissions. Although intermodal transportation has been shown to increase operating efficiency and reduce carbon emissions, research on infrastructural support for intermodal transportation is still insufficient. In this study, we establish a mixed integer programming model to jointly optimize strategic infrastructure development decisions and freight transportation decisions over a long horizon. Our model features a mixture of traditional single-mode facilities and hybrid hubs that facilitate rail–water transportation integration. A branch-and-cut decomposition algorithm is developed to solve large-scale problems. We collect real-world freight, infrastructure, and operations data to conduct computational studies on the model performance and algorithm efficiency. We provide insights for practitioners to address infrastructure planning and budgetary concerns. A case study using the established model at the national scale shows that well-optimized transportation infrastructure investment could have over 300% return during a 25-year horizon. In addition, fully capitalizing on the maturing clean vehicle technologies could reduce carbon emissions by 73.56 million tons at an annual rate.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100369"},"PeriodicalIF":3.7,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145690179","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01DOI: 10.1016/j.orp.2025.100370
Xue Huang , Hongyu He , Hong-Bin Bei , Yanzhi Zhao , Ning Wang , Yu Chang
In this article, we investigate the resource allocations group-scheduling with position-based learning effects. Under a single-machine, the purpose is to determine an optimal group sequence, job sequence within each group, and convex resource allocations (i.e., second partial derivatives of resources are not negative) assigned to the jobs. For the total resource consumption minimization with limited makespan constraint, we certify that the problem is polynomially solvable for some special situations. For the general situation, we establish a heuristic and a branch-and-bound algorithm. Computation experiments are given to test the effectiveness of solution algorithms. The proposed model can be probably applied to green manufacturing scenarios, supporting sustainable production by considering controllable processing time.
{"title":"Group-scheduling with simultaneous learning effects and convex resource allocations","authors":"Xue Huang , Hongyu He , Hong-Bin Bei , Yanzhi Zhao , Ning Wang , Yu Chang","doi":"10.1016/j.orp.2025.100370","DOIUrl":"10.1016/j.orp.2025.100370","url":null,"abstract":"<div><div>In this article, we investigate the resource allocations group-scheduling with position-based learning effects. Under a single-machine, the purpose is to determine an optimal group sequence, job sequence within each group, and convex resource allocations (i.e., second partial derivatives of resources are not negative) assigned to the jobs. For the total resource consumption minimization with limited makespan constraint, we certify that the problem is polynomially solvable for some special situations. For the general situation, we establish a heuristic and a branch-and-bound algorithm. Computation experiments are given to test the effectiveness of solution algorithms. The proposed model can be probably applied to green manufacturing scenarios, supporting sustainable production by considering controllable processing time.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100370"},"PeriodicalIF":3.7,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145747419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01DOI: 10.1016/j.orp.2025.100367
Ritsaart Bergsma , Corné de Ruijt , Sandjai Bhulai
This systematic review investigates the applications of machine learning (ML) in inventory control, analyzing 122 articles to provide a comprehensive overview of the state of the art and identify future research directions. The study proposes a typology to classify the integration of ML into the inventory optimization framework, distinguishing three primary approaches: (1) separate estimation and optimization, where ML is applied to demand forecasting before optimization, (2) static ML-integrated optimization, where ML is directly embedded into optimization models, and (3) dynamic ML-integrated optimization, where reinforcement learning (RL) is employed to derive optimal inventory policies. The findings highlight that while RL applications are gaining prominence, significant research gaps remain, particularly in scaling algorithms to real-world problems, handling large action spaces, and developing RL algorithms that are tailored to inventory control. The review also assesses the operational dynamics of inventory systems addressed in the literature, such as single/multi-item models, lead time assumptions, and echelon structures. Underexplored areas include stochastic lead times, complementary items, quantity discounts, product obsolescence, and multi-echelon networks. The study concludes by outlining key research gaps and offering directions for future research to advance the integration of ML in inventory control.
{"title":"A systematic review of machine learning approaches in inventory control optimization","authors":"Ritsaart Bergsma , Corné de Ruijt , Sandjai Bhulai","doi":"10.1016/j.orp.2025.100367","DOIUrl":"10.1016/j.orp.2025.100367","url":null,"abstract":"<div><div>This systematic review investigates the applications of machine learning (ML) in inventory control, analyzing 122 articles to provide a comprehensive overview of the state of the art and identify future research directions. The study proposes a typology to classify the integration of ML into the inventory optimization framework, distinguishing three primary approaches: (1) separate estimation and optimization, where ML is applied to demand forecasting before optimization, (2) static ML-integrated optimization, where ML is directly embedded into optimization models, and (3) dynamic ML-integrated optimization, where reinforcement learning (RL) is employed to derive optimal inventory policies. The findings highlight that while RL applications are gaining prominence, significant research gaps remain, particularly in scaling algorithms to real-world problems, handling large action spaces, and developing RL algorithms that are tailored to inventory control. The review also assesses the operational dynamics of inventory systems addressed in the literature, such as single/multi-item models, lead time assumptions, and echelon structures. Underexplored areas include stochastic lead times, complementary items, quantity discounts, product obsolescence, and multi-echelon networks. The study concludes by outlining key research gaps and offering directions for future research to advance the integration of ML in inventory control.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100367"},"PeriodicalIF":3.7,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145623416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01DOI: 10.1016/j.orp.2025.100366
Yun Gu
With the rapid development of global trade and cross-border e-commerce, optimizing cross-border multi-level warehouse networks has become a critical challenge to enhance supply chain efficiency and reduce operational costs. Traditional logistics planning methods struggle to address complex multi-level network structures, heterogeneous big data, and multi-dimensional influencing factors. This study proposes a mixed-integer linear programming model based on real-world operational requirements to optimize the layout of cross-border multi-level warehouse networks. The model integrates transportation costs, warehousing costs, tariff costs, and service lead time as key considerations. Through the incorporation of heuristic constraints and relaxation strategies, the model significantly improves computational efficiency and stability. Experimental results using real data from a cross-border e-commerce enterprise demonstrate that compared to existing solutions, the MILP model reduces total costs by 20.7 %, outperforms heuristic algorithms by >8 %, achieves faster computational speed, and maintains stable results. Furthermore, in 16 perturbation experiments, the model retained optimal solutions in 15 instances, showcasing strong robustness. This research provides critical theoretical and practical guidance for the scientific planning of cross-border logistics networks.
{"title":"Cross-border multi-level warehouse network optimization: Modeling and application based on mixed-integer linear programming","authors":"Yun Gu","doi":"10.1016/j.orp.2025.100366","DOIUrl":"10.1016/j.orp.2025.100366","url":null,"abstract":"<div><div>With the rapid development of global trade and cross-border e-commerce, optimizing cross-border multi-level warehouse networks has become a critical challenge to enhance supply chain efficiency and reduce operational costs. Traditional logistics planning methods struggle to address complex multi-level network structures, heterogeneous big data, and multi-dimensional influencing factors. This study proposes a mixed-integer linear programming model based on real-world operational requirements to optimize the layout of cross-border multi-level warehouse networks. The model integrates transportation costs, warehousing costs, tariff costs, and service lead time as key considerations. Through the incorporation of heuristic constraints and relaxation strategies, the model significantly improves computational efficiency and stability. Experimental results using real data from a cross-border e-commerce enterprise demonstrate that compared to existing solutions, the MILP model reduces total costs by 20.7 %, outperforms heuristic algorithms by >8 %, achieves faster computational speed, and maintains stable results. Furthermore, in 16 perturbation experiments, the model retained optimal solutions in 15 instances, showcasing strong robustness. This research provides critical theoretical and practical guidance for the scientific planning of cross-border logistics networks.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100366"},"PeriodicalIF":3.7,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145623417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01DOI: 10.1016/j.orp.2025.100368
Joonrak Kim , Seunghoon Lee
This study develops a robust optimization framework for closed-loop supply chain (CLSC) planning that explicitly accounts for uncertainty in the quality of recycled and remanufactured inputs. While such materials are critical for sustainability, their variable quality poses risks to production feasibility and supply reliability. To address this challenge, we propose an ordering-proportion-based robust model that distributes uncertainty across sourcing proportions and leverages the Bertsimas–Sim budget of uncertainty to balance conservatism and flexibility. A reformulation ensures tractability and preserves robust feasibility. Computational experiments demonstrate that the proposed model reduces shortages and stabilizes performance under independently realized uncertainties, while quantity-based robust models are more effective when uncertainties are correlated. Additional scalability tests confirm that the model remains computationally tractable for medium-sized networks. The findings highlight practical implications for managers, showing how proportion-based sourcing improves resilience, supports reliable demand fulfillment, and strengthens sustainability in CLSCs facing quality risks.
{"title":"Robust optimization model for closed-loop supply chain planning with collected material quality uncertainty","authors":"Joonrak Kim , Seunghoon Lee","doi":"10.1016/j.orp.2025.100368","DOIUrl":"10.1016/j.orp.2025.100368","url":null,"abstract":"<div><div>This study develops a robust optimization framework for closed-loop supply chain (CLSC) planning that explicitly accounts for uncertainty in the quality of recycled and remanufactured inputs. While such materials are critical for sustainability, their variable quality poses risks to production feasibility and supply reliability. To address this challenge, we propose an ordering-proportion-based robust model that distributes uncertainty across sourcing proportions and leverages the Bertsimas–Sim budget of uncertainty to balance conservatism and flexibility. A reformulation ensures tractability and preserves robust feasibility. Computational experiments demonstrate that the proposed model reduces shortages and stabilizes performance under independently realized uncertainties, while quantity-based robust models are more effective when uncertainties are correlated. Additional scalability tests confirm that the model remains computationally tractable for medium-sized networks. The findings highlight practical implications for managers, showing how proportion-based sourcing improves resilience, supports reliable demand fulfillment, and strengthens sustainability in CLSCs facing quality risks.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100368"},"PeriodicalIF":3.7,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145623418","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-19DOI: 10.1016/j.orp.2025.100363
Luigi Pescio, Marta Ribeiro, Bruno F. Santos
Flight and maintenance scheduling pose conflicting objectives: while maintenance is vital for ensuring aircraft airworthiness, it comes at the cost of taking aircraft out of operation. In current operations, airlines manually handle tail assignment and maintenance task scheduling separately, missing an opportunity to strike a better balance. This division leads to wasted maintenance resources, restricted fleet availability for schedule flexibility, inconsistent planning, and neglect of schedule resilience. This study presents a novel approach that integrates tail assignment and maintenance scheduling into a unified decision-support framework. An integer program, tailored to meet airline-specific requirements and constraints, is combined with an innovative time-space network (TSN). The TSN incorporates two distinct spaces for maintenance and network activities. The primary objective is to generate feasible plans that increase schedule efficiency (i.e., no cancellations, high fleet availability, high fleet health, and optimal use of maintenance resources) and schedule stability (i.e., limited number of late arrival disruptions during operations) the day before operation. Additionally, this framework addresses overlooked aspects in the literature: it treats maintenance tasks as variable interval activities based on aircraft-specific needs, departing from the traditional fixed interval approach. The performance of the framework is tested with real-data provided by a major European single hub-to-spoke airline, with a heterogeneous fleet of over 50 wide-body aircraft. Historical data from arrival delays is used to create robust buffers that mitigate delay propagation. A 17% reduction in maintenance time was achieved compared to the airline’s current plans, resulting in a 10% increase in fleet availability on the day of operations. This improvement is attributed to higher labour and task interval utilization, indicating the framework’s superior efficiency in scheduling maintenance tasks. Lastly, the framework produced plans more resilient to arrival delays, reducing the number of disruptions and delay propagation over 40%. This framework can be used as a decision-support tool for airlines, enabling the creation of schedules that are both robust against delays and optimized for fleet utilization.
{"title":"Unified tail assignment and maintenance task scheduling: A decision support framework for improved efficiency and stability","authors":"Luigi Pescio, Marta Ribeiro, Bruno F. Santos","doi":"10.1016/j.orp.2025.100363","DOIUrl":"10.1016/j.orp.2025.100363","url":null,"abstract":"<div><div>Flight and maintenance scheduling pose conflicting objectives: while maintenance is vital for ensuring aircraft airworthiness, it comes at the cost of taking aircraft out of operation. In current operations, airlines manually handle tail assignment and maintenance task scheduling separately, missing an opportunity to strike a better balance. This division leads to wasted maintenance resources, restricted fleet availability for schedule flexibility, inconsistent planning, and neglect of schedule resilience. This study presents a novel approach that integrates tail assignment and maintenance scheduling into a unified decision-support framework. An integer program, tailored to meet airline-specific requirements and constraints, is combined with an innovative time-space network (TSN). The TSN incorporates two distinct spaces for maintenance and network activities. The primary objective is to generate feasible plans that increase schedule efficiency (i.e., no cancellations, high fleet availability, high fleet health, and optimal use of maintenance resources) and schedule stability (i.e., limited number of late arrival disruptions during operations) the day before operation. Additionally, this framework addresses overlooked aspects in the literature: it treats maintenance tasks as variable interval activities based on aircraft-specific needs, departing from the traditional fixed interval approach. The performance of the framework is tested with real-data provided by a major European single hub-to-spoke airline, with a heterogeneous fleet of over 50 wide-body aircraft. Historical data from arrival delays is used to create robust buffers that mitigate delay propagation. A 17% reduction in maintenance time was achieved compared to the airline’s current plans, resulting in a 10% increase in fleet availability on the day of operations. This improvement is attributed to higher labour and task interval utilization, indicating the framework’s superior efficiency in scheduling maintenance tasks. Lastly, the framework produced plans more resilient to arrival delays, reducing the number of disruptions and delay propagation over 40%. This framework can be used as a decision-support tool for airlines, enabling the creation of schedules that are both robust against delays and optimized for fleet utilization.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100363"},"PeriodicalIF":3.7,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145579142","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, we address the problem of Column Generation (CG) for routing problems using Reinforcement Learning (RL). Specifically, we use a RL model based on the attention-mechanism architecture to find the columns with most negative reduced cost in the Pricing Problem (PP). Unlike previous Machine Learning (ML) applications for CG, our model deploys an end-to-end mechanism that independently solves the pricing problem without the help of any heuristic. We consider a variant of Vehicle Routing Problem (VRP) as a case study for our method. Through a series of experiments comparing our approach with a Dynamic Programming (DP)-based heuristic for solving the PP, we demonstrate that the proposed method obtains solutions for the linear relaxation up to a reasonable objective gap and significantly faster than the DP-based heuristic for the PP.
{"title":"Reinforcement learning for solving the pricing problem in column generation for routing","authors":"Abdo Abouelrous , Laurens Bliek , Adriana F. Gabor , Yaoxin Wu , Yingqian Zhang","doi":"10.1016/j.orp.2025.100364","DOIUrl":"10.1016/j.orp.2025.100364","url":null,"abstract":"<div><div>In this paper, we address the problem of Column Generation (CG) for routing problems using Reinforcement Learning (RL). Specifically, we use a RL model based on the attention-mechanism architecture to find the columns with most negative reduced cost in the Pricing Problem (PP). Unlike previous Machine Learning (ML) applications for CG, our model deploys an end-to-end mechanism that independently solves the pricing problem without the help of any heuristic. We consider a variant of Vehicle Routing Problem (VRP) as a case study for our method. Through a series of experiments comparing our approach with a Dynamic Programming (DP)-based heuristic for solving the PP, we demonstrate that the proposed method obtains solutions for the linear relaxation up to a reasonable objective gap and significantly faster than the DP-based heuristic for the PP.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100364"},"PeriodicalIF":3.7,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145525286","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-12DOI: 10.1016/j.orp.2025.100365
Hui Yu , Yu Gong , Xiaoli Yan
The newsvendor problem with periodic demand (PFNV) is a common and significant challenge in practice, where traditional methods such as optimization, statistical analysis, and artificial intelligence often struggle to balance effectiveness and operability. We propose the Probability-based Adaptive Strategy (PAS) for the PFNV problem, which formulates decisions through the dual reference points and probabilities. The decision-making process comprises four steps that simulate human behavior based on bounded rationality. The design of reference points is data-driven, using either a linear method or a multi-armed bandit (MAB), while probability calculation is guided by an optimization objective that reflects human regret psychology. The final decision is made through either a random sampling (RS) or an expectation construction (EC) scheme. Experiments with both simulated and real-world data show that PAS effectively captures periodic trends in both stable and volatile datasets. The PAS combining classification, MAB, and the EC scheme performs better in average cost in most cases, while other variants exhibit different characteristics under varying conditions. Compared with several benchmarks, PAS demonstrates potential for cost optimization in certain scenarios while maintaining both operability and interpretability.
{"title":"A Probabilistic and adaptive strategy for the newsvendor problem with periodic demand","authors":"Hui Yu , Yu Gong , Xiaoli Yan","doi":"10.1016/j.orp.2025.100365","DOIUrl":"10.1016/j.orp.2025.100365","url":null,"abstract":"<div><div>The newsvendor problem with periodic demand (PFNV) is a common and significant challenge in practice, where traditional methods such as optimization, statistical analysis, and artificial intelligence often struggle to balance effectiveness and operability. We propose the Probability-based Adaptive Strategy (PAS) for the PFNV problem, which formulates decisions through the dual reference points and probabilities. The decision-making process comprises four steps that simulate human behavior based on bounded rationality. The design of reference points is data-driven, using either a linear method or a multi-armed bandit (MAB), while probability calculation is guided by an optimization objective that reflects human regret psychology. The final decision is made through either a random sampling (RS) or an expectation construction (EC) scheme. Experiments with both simulated and real-world data show that PAS effectively captures periodic trends in both stable and volatile datasets. The PAS combining classification, MAB, and the EC scheme performs better in average cost in most cases, while other variants exhibit different characteristics under varying conditions. Compared with several benchmarks, PAS demonstrates potential for cost optimization in certain scenarios while maintaining both operability and interpretability.</div></div>","PeriodicalId":38055,"journal":{"name":"Operations Research Perspectives","volume":"15 ","pages":"Article 100365"},"PeriodicalIF":3.7,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145579141","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号