Pub Date : 2025-02-08DOI: 10.1016/j.ejor.2025.01.043
Xiaolong Guo, Zenghui Su, Fangkezi Zhou
The retail market is being increasingly invaded by suppliers who are establishing their own direct selling channels, thanks to the rise of e-commerce and internet technology. The cost of direct sales has been identified as a crucial factor in the strategic interaction between suppliers and retailers. However, retailers often struggle to accurately assess this cost due to their decision biases. To address this issue, we propose a strategic mental model to examine how these biases impact supplier encroachment and firms’ performance outcomes. Our analysis reveals that a retailer can benefit from having an underestimation bias. Additionally, the bias of one firm can benefit the other, depending on whether it is an underestimation or overestimation bias. Interestingly, both the supplier and the retailer can earn more when they are biased compared to when neither of them is biased, resulting in a win-win scenario. However, if the supplier fails to recognize the retailer’s bias, the possibility of mutual gains is eliminated. Furthermore, when biases are present, the option of encroachment may backfire for the supplier compared to the scenario without encroachment, and the retailer may be worse off in the sequential encroachment setting compared to the simultaneous setting. Finally, we extend our model by considering imperfect substitutability between the retailer’s and supplier’s products, and the main conclusions remain robust. Our findings suggest that considering these biases alters the nature of strategic dynamics and provides new insights into supplier encroachment and information management.
{"title":"Supplier encroachment with decision biases","authors":"Xiaolong Guo, Zenghui Su, Fangkezi Zhou","doi":"10.1016/j.ejor.2025.01.043","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.043","url":null,"abstract":"The retail market is being increasingly invaded by suppliers who are establishing their own direct selling channels, thanks to the rise of e-commerce and internet technology. The cost of direct sales has been identified as a crucial factor in the strategic interaction between suppliers and retailers. However, retailers often struggle to accurately assess this cost due to their decision biases. To address this issue, we propose a strategic mental model to examine how these biases impact supplier encroachment and firms’ performance outcomes. Our analysis reveals that a retailer can benefit from having an underestimation bias. Additionally, the bias of one firm can benefit the other, depending on whether it is an underestimation or overestimation bias. Interestingly, both the supplier and the retailer can earn more when they are biased compared to when neither of them is biased, resulting in a win-win scenario. However, if the supplier fails to recognize the retailer’s bias, the possibility of mutual gains is eliminated. Furthermore, when biases are present, the option of encroachment may backfire for the supplier compared to the scenario without encroachment, and the retailer may be worse off in the sequential encroachment setting compared to the simultaneous setting. Finally, we extend our model by considering imperfect substitutability between the retailer’s and supplier’s products, and the main conclusions remain robust. Our findings suggest that considering these biases alters the nature of strategic dynamics and provides new insights into supplier encroachment and information management.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"47 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443570","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-08DOI: 10.1016/j.ejor.2025.01.042
Søren Glud Johansen, Anders Thorstenson
In this paper we provide an analysis of an inventory system with compound Poisson demand and a sequential supply system, specifically focusing on non-crossing lead times. The purpose is to facilitate a comparison with order crossing caused by independent lead times. We apply a new approach for modeling exogenous, non-crossing Erlang distributed lead times. It is assumed that the inventory is controlled by continuous review and a base-stock level. Any part of a customer demand which cannot be satisfied immediately from inventory is lost. Set-up costs are negligible and the relevant cost parameters for choosing the best base stock consist of the holding cost rate and a shortage cost per unit lost. We provide a proof that the long-run average total relevant cost is a convex function of the base stock. Hence, the base stock is easy to compute and it is optimal in the case of geometrically distributed customer demand sizes. For demand sizes which are not geometric, we suggest an approximation to specify the base stock. An approximation is also suggested for the case when the lead time is only specified by its mean and standard deviation (SD). Our numerical study shows that the average cost is very sensitive to the SD. This in sharp contrast to the complete insensitivity of SD in case of geometric demand sizes and independent lead times.
{"title":"Non-crossing vs. independent lead times in a lost-sales inventory system with compound Poisson demand","authors":"Søren Glud Johansen, Anders Thorstenson","doi":"10.1016/j.ejor.2025.01.042","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.042","url":null,"abstract":"In this paper we provide an analysis of an inventory system with compound Poisson demand and a sequential supply system, specifically focusing on non-crossing lead times. The purpose is to facilitate a comparison with order crossing caused by independent lead times. We apply a new approach for modeling exogenous, non-crossing Erlang distributed lead times. It is assumed that the inventory is controlled by continuous review and a base-stock level. Any part of a customer demand which cannot be satisfied immediately from inventory is lost. Set-up costs are negligible and the relevant cost parameters for choosing the best base stock consist of the holding cost rate and a shortage cost per unit lost. We provide a proof that the long-run average total relevant cost is a convex function of the base stock. Hence, the base stock is easy to compute and it is optimal in the case of geometrically distributed customer demand sizes. For demand sizes which are not geometric, we suggest an approximation to specify the base stock. An approximation is also suggested for the case when the lead time is only specified by its mean and standard deviation (SD). Our numerical study shows that the average cost is very sensitive to the SD. This in sharp contrast to the complete insensitivity of SD in case of geometric demand sizes and independent lead times.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"65 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143477832","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-08DOI: 10.1016/j.ejor.2025.01.041
Stijn Loeys, Robert N. Boute, Katrien Antonio
We explore the value of using operational sensor data to improve the risk assessment of service contracts that cover all maintenance-related costs during a fixed period. An initial estimate of the contract risk is determined by predicting the maintenance costs via a gradient-boosting machine based on the machine’s and contract’s characteristics observable at the onset of the contract period. We then periodically update this risk assessment based on operational sensor data observed throughout the contract period. These sensor data reveal operational machine usage that drives the maintenance risk. We validate our approach on a portfolio of about 4,000 full-service contracts of industrial equipment and show how dynamic sensor data improves risk differentiation.
{"title":"The use of IoT sensor data to dynamically assess maintenance risk in service contracts","authors":"Stijn Loeys, Robert N. Boute, Katrien Antonio","doi":"10.1016/j.ejor.2025.01.041","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.041","url":null,"abstract":"We explore the value of using operational sensor data to improve the risk assessment of service contracts that cover all maintenance-related costs during a fixed period. An initial estimate of the contract risk is determined by predicting the maintenance costs via a gradient-boosting machine based on the machine’s and contract’s characteristics observable at the onset of the contract period. We then periodically update this risk assessment based on operational sensor data observed throughout the contract period. These sensor data reveal operational machine usage that drives the maintenance risk. We validate our approach on a portfolio of about 4,000 full-service contracts of industrial equipment and show how dynamic sensor data improves risk differentiation.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"24 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443571","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-07DOI: 10.1016/j.ejor.2025.01.040
Nikita Kozodoi, Stefan Lessmann, Morteza Alamgir, Luis Moreira-Matias, Konstantinos Papakonstantinou
Scoring models support decision-making in financial institutions. Their estimation and evaluation rely on labeled data from previously accepted clients. Ignoring rejected applicants with unknown repayment behavior introduces sampling bias, as the available labeled data only partially represents the population of potential borrowers. This paper examines the impact of sampling bias and introduces new methods to mitigate its adverse effect. First, we develop a bias-aware self-labeling algorithm for scorecard training, which debiases the training data by adding selected rejects with an inferred label. Second, we propose a Bayesian framework to address sampling bias in scorecard evaluation. To provide reliable projections of future scorecard performance, we include rejected clients with random pseudo-labels in the test set and use Monte Carlo sampling to estimate the scorecard’s expected performance across label realizations. We conduct extensive experiments using both synthetic and observational data. The observational data includes an unbiased sample of applicants accepted without scoring, representing the true borrower population and facilitating a realistic assessment of reject inference techniques. The results show that our methods outperform established benchmarks in predictive accuracy and profitability. Additional sensitivity analysis clarifies the conditions under which they are most effective. Comparing the relative effectiveness of addressing sampling bias during scorecard training versus evaluation, we find the latter much more promising. For example, we estimate the expected return per dollar issued to increase by up to 2.07 and up to 5.76 percentage points when using bias-aware self-labeling and Bayesian evaluation, respectively.
{"title":"Fighting sampling bias: A framework for training and evaluating credit scoring models","authors":"Nikita Kozodoi, Stefan Lessmann, Morteza Alamgir, Luis Moreira-Matias, Konstantinos Papakonstantinou","doi":"10.1016/j.ejor.2025.01.040","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.040","url":null,"abstract":"Scoring models support decision-making in financial institutions. Their estimation and evaluation rely on labeled data from previously accepted clients. Ignoring rejected applicants with unknown repayment behavior introduces sampling bias, as the available labeled data only partially represents the population of potential borrowers. This paper examines the impact of sampling bias and introduces new methods to mitigate its adverse effect. First, we develop a bias-aware self-labeling algorithm for scorecard training, which debiases the training data by adding selected rejects with an inferred label. Second, we propose a Bayesian framework to address sampling bias in scorecard evaluation. To provide reliable projections of future scorecard performance, we include rejected clients with random pseudo-labels in the test set and use Monte Carlo sampling to estimate the scorecard’s expected performance across label realizations. We conduct extensive experiments using both synthetic and observational data. The observational data includes an unbiased sample of applicants accepted without scoring, representing the true borrower population and facilitating a realistic assessment of reject inference techniques. The results show that our methods outperform established benchmarks in predictive accuracy and profitability. Additional sensitivity analysis clarifies the conditions under which they are most effective. Comparing the relative effectiveness of addressing sampling bias during scorecard training versus evaluation, we find the latter much more promising. For example, we estimate the expected return per dollar issued to increase by up to 2.07 and up to 5.76 percentage points when using bias-aware self-labeling and Bayesian evaluation, respectively.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"11 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443572","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-07DOI: 10.1016/j.ejor.2025.01.034
Alessandro Agnetis, Jean-Charles Billaut, Michael Pinedo, Dvir Shabtay
This paper presents an overview of scheduling research done over the last half century. The main focus is on what is typically referred to as machine scheduling. The first section describes the general framework for machine scheduling models and introduces the notation. The second section discusses the basic deterministic machine scheduling models, including single machine, parallel machines, flow shops, job shops, and open shops. The third section describes more elaborate models, including multi-objective and multi-agent scheduling models, scheduling with controllable processing times, scheduling with rejection, just-in-time scheduling, scheduling with due date assignments, time-dependent scheduling, and scheduling with batching and setups. The two subsequent sections consider scheduling under uncertainty; section four goes into online and robust scheduling and section five covers stochastic scheduling models. The next section describes a variety of important scheduling applications, including applications in manufacturing, in services, and in information processing. The last section presents the main conclusions and discusses future research directions.
{"title":"Fifty years of research in scheduling — Theory and applications","authors":"Alessandro Agnetis, Jean-Charles Billaut, Michael Pinedo, Dvir Shabtay","doi":"10.1016/j.ejor.2025.01.034","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.034","url":null,"abstract":"This paper presents an overview of scheduling research done over the last half century. The main focus is on what is typically referred to as machine scheduling. The first section describes the general framework for machine scheduling models and introduces the notation. The second section discusses the basic deterministic machine scheduling models, including single machine, parallel machines, flow shops, job shops, and open shops. The third section describes more elaborate models, including multi-objective and multi-agent scheduling models, scheduling with controllable processing times, scheduling with rejection, just-in-time scheduling, scheduling with due date assignments, time-dependent scheduling, and scheduling with batching and setups. The two subsequent sections consider scheduling under uncertainty; section four goes into online and robust scheduling and section five covers stochastic scheduling models. The next section describes a variety of important scheduling applications, including applications in manufacturing, in services, and in information processing. The last section presents the main conclusions and discusses future research directions.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"138 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143583106","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-06DOI: 10.1016/j.ejor.2025.01.039
Yuyu Chen , Taizhong Hu , Ruodu Wang , Zhenfeng Zou
We study stochastic dominance between portfolios of independent and identically distributed (iid) extremely heavy-tailed (i.e., infinite-mean) Pareto random variables. With the notion of majorization order, we show that a more diversified portfolio of iid extremely heavy-tailed Pareto random variables is larger in the sense of first-order stochastic dominance. This result is further generalized for Pareto random variables caused by triggering events, random variables with tails being Pareto, bounded Pareto random variables, and positively dependent Pareto random variables. These results provide an important implication in investment: Diversification of extremely heavy-tailed Pareto profits uniformly increases investors’ profitability, leading to a diversification benefit. Remarkably, different from the finite-mean setting, such a diversification benefit does not depend on the decision maker’s risk aversion.
{"title":"Diversification for infinite-mean Pareto models without risk aversion","authors":"Yuyu Chen , Taizhong Hu , Ruodu Wang , Zhenfeng Zou","doi":"10.1016/j.ejor.2025.01.039","DOIUrl":"10.1016/j.ejor.2025.01.039","url":null,"abstract":"<div><div>We study stochastic dominance between portfolios of independent and identically distributed (iid) extremely heavy-tailed (i.e., infinite-mean) Pareto random variables. With the notion of majorization order, we show that a more diversified portfolio of iid extremely heavy-tailed Pareto random variables is larger in the sense of first-order stochastic dominance. This result is further generalized for Pareto random variables caused by triggering events, random variables with tails being Pareto, bounded Pareto random variables, and positively dependent Pareto random variables. These results provide an important implication in investment: Diversification of extremely heavy-tailed Pareto profits uniformly increases investors’ profitability, leading to a diversification benefit. Remarkably, different from the finite-mean setting, such a diversification benefit does not depend on the decision maker’s risk aversion.</div></div>","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"323 1","pages":"Pages 341-350"},"PeriodicalIF":6.0,"publicationDate":"2025-02-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143436944","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-05DOI: 10.1016/j.ejor.2025.01.028
Michael Simonis, Stefan Nickel
This paper discusses an iterative simulation–optimization approach to estimate high-quality solutions for the multi-level capacitated lot-sizing problem with linked lot sizes and backorders (MLCLSP-L-B) based on probabilistic demand. It presents the application of the Generalized Uncertainty Framework (GUF) to the MLCLSP-L-B. The research provides an exact mathematical problem formulation and a variable neighborhood search (VNS) algorithm for the GUF. The evaluation procedure uses anonymized real-world data of multi-level pharmaceutical tablets manufacturing processes. It compares the GUF against a two-stage stochastic programming (SP) approach from the literature regarding manufacturing costs and customer service levels. Finally, planning rules and managerial insights are given for the tablets manufacturing processes.
{"title":"A simulation–optimization approach for capacitated lot-sizing in a multi-level pharmaceutical tablets manufacturing process","authors":"Michael Simonis, Stefan Nickel","doi":"10.1016/j.ejor.2025.01.028","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.028","url":null,"abstract":"This paper discusses an iterative simulation–optimization approach to estimate high-quality solutions for the multi-level capacitated lot-sizing problem with linked lot sizes and backorders (MLCLSP-L-B) based on probabilistic demand. It presents the application of the Generalized Uncertainty Framework (GUF) to the MLCLSP-L-B. The research provides an exact mathematical problem formulation and a variable neighborhood search (VNS) algorithm for the GUF. The evaluation procedure uses anonymized real-world data of multi-level pharmaceutical tablets manufacturing processes. It compares the GUF against a two-stage stochastic programming (SP) approach from the literature regarding manufacturing costs and customer service levels. Finally, planning rules and managerial insights are given for the tablets manufacturing processes.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"81 4 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443575","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-05DOI: 10.1016/j.ejor.2025.01.033
Martin Albrecht
This paper proposes a new mechanism for coordinating master planning in a buyer–supplier supply chain. Parties take different roles in the mechanism: there is an informed party (IP) and a reporting party (RP). The mechanism consists of three steps. First, the RP defines a lump sum payment, which he will receive if a supply proposal that deviates from the default is implemented. Second, parties generate new supply proposals based on mathematical programming models. The RP communicates his profit changes for the new proposals to the IP. Third, the IP decides whether to implement one of the new proposals and to pay to the RP the lump sum minus the RP’s local profit change. We show that the mechanism is strategyproof, budget-balanced and individually rational. If the RP has uniformly distributed prior knowledge about the actual surplus from coordination, at least 3/4 of the surplus will be realized on the average. Our computational tests suggest that the coordination mechanism is effective and versatile. It identifies near optimal solutions for various master planning models and outperforms an existing approach from literature.
{"title":"Coordination of master planning in supply chains","authors":"Martin Albrecht","doi":"10.1016/j.ejor.2025.01.033","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.033","url":null,"abstract":"This paper proposes a new mechanism for coordinating master planning in a buyer–supplier supply chain. Parties take different roles in the mechanism: there is an informed party (IP) and a reporting party (RP). The mechanism consists of three steps. First, the RP defines a lump sum payment, which he will receive if a supply proposal that deviates from the default is implemented. Second, parties generate new supply proposals based on mathematical programming models. The RP communicates his profit changes for the new proposals to the IP. Third, the IP decides whether to implement one of the new proposals and to pay to the RP the lump sum minus the RP’s local profit change. We show that the mechanism is strategyproof, budget-balanced and individually rational. If the RP has uniformly distributed prior knowledge about the actual surplus from coordination, at least 3/4 of the surplus will be realized on the average. Our computational tests suggest that the coordination mechanism is effective and versatile. It identifies near optimal solutions for various master planning models and outperforms an existing approach from literature.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"63 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143385309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-05DOI: 10.1016/j.ejor.2025.01.038
Nils Boysen, Dirk Briskorn, Benoit Montreuil, Lennart Zey
The Physical Internet (PI or π) is a design metaphor that applies the digital internet as an archetype to rethink freight logistics and transportation in a more sustainable, interoperable, and cooperative way. Analogously to the protocols of the digital internet, freight should be encapsulated into standardized π-containers and transported through an open network of cooperating π-hubs. Despite the inspiring analogy, parallels are not absolute. Digital data packages can be duplicated without cost, can be (re-)sent anywhere in short time, generate no return flows, and their routing decisions have to be taken in microseconds. In this paper, we focus on a characteristic of the digital internet that can be emulated by the PI but has received less attention yet: split transports. Analogously to the internet protocol, which forwards each data packet individually according to dynamically adapted routing tables, larger shipments can be split into multiple smaller π-containers, so that different containers with the same destination may be routed via different paths. To evaluate whether these split transports significantly promote the aims of the PI, we derive a basic scheduling problem, called the π-transportation problem, where a given set of (either split or unsplit) shipments aim to travel along a linear transport corridor with given π-hubs. Based on this problem, we derive analytical and computational results to quantify the impact of split transports on the success of the PI concept. Our results suggest that split transports are not among the important features when redesigning the transportation sector.
{"title":"The [formula omitted]-transportation problem: On the value of split transports for the Physical Internet concept","authors":"Nils Boysen, Dirk Briskorn, Benoit Montreuil, Lennart Zey","doi":"10.1016/j.ejor.2025.01.038","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.038","url":null,"abstract":"The Physical Internet (PI or <mml:math altimg=\"si55.svg\" display=\"inline\"><mml:mi>π</mml:mi></mml:math>) is a design metaphor that applies the digital internet as an archetype to rethink freight logistics and transportation in a more sustainable, interoperable, and cooperative way. Analogously to the protocols of the digital internet, freight should be encapsulated into standardized <mml:math altimg=\"si55.svg\" display=\"inline\"><mml:mi>π</mml:mi></mml:math>-containers and transported through an open network of cooperating <mml:math altimg=\"si55.svg\" display=\"inline\"><mml:mi>π</mml:mi></mml:math>-hubs. Despite the inspiring analogy, parallels are not absolute. Digital data packages can be duplicated without cost, can be (re-)sent anywhere in short time, generate no return flows, and their routing decisions have to be taken in microseconds. In this paper, we focus on a characteristic of the digital internet that can be emulated by the PI but has received less attention yet: split transports. Analogously to the internet protocol, which forwards each data packet individually according to dynamically adapted routing tables, larger shipments can be split into multiple smaller <mml:math altimg=\"si55.svg\" display=\"inline\"><mml:mi>π</mml:mi></mml:math>-containers, so that different containers with the same destination may be routed via different paths. To evaluate whether these split transports significantly promote the aims of the PI, we derive a basic scheduling problem, called the <mml:math altimg=\"si55.svg\" display=\"inline\"><mml:mi>π</mml:mi></mml:math>-transportation problem, where a given set of (either split or unsplit) shipments aim to travel along a linear transport corridor with given <mml:math altimg=\"si55.svg\" display=\"inline\"><mml:mi>π</mml:mi></mml:math>-hubs. Based on this problem, we derive analytical and computational results to quantify the impact of split transports on the success of the PI concept. Our results suggest that split transports are not among the important features when redesigning the transportation sector.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"14 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443573","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-05DOI: 10.1016/j.ejor.2025.01.037
Rodrigo e Alvim Alexandre, Marcelo D. Fragoso, Virgílio J.M. Ferreira Filho, Edilson F. Arruda
Although there are techniques to address large scale Markov decision processes (MDP), a computationally adequate solution of the so-called curse of dimensionality still eludes, in many aspects, a satisfactory treatment. In this paper, we advance in this issue by introducing a novel multi-subset partitioning scheme to allow for a distributed evaluation of the MDP, aiming to accelerate convergence and enable distributed policy improvement across the state space, whereby the value function and the policy improvement step can be performed independently, one subset at a time. The scheme’s innovation hinges on a design that induces communication properties that allow us to evaluate time aggregated trajectories via absorption analysis, thereby limiting the computational effort. The paper introduces and proves the convergence of a class of distributed time aggregation algorithms that combine the partitioning scheme with two-phase time aggregation to distribute the computations and accelerate convergence. In addition, we make use of Foster’s sufficient conditions for stochastic stability to develop a new theoretical result which underpins a partition design that guarantees that large regions of the state space are rarely visited and have a marginal effect on the system’s performance. This enables the design of approximate algorithms to find near-optimal solutions to large scale systems by focusing on the most visited regions of the state space. We validate the approach in a series of experiments featuring production and inventory and queuing applications. The results highlight the potential of the proposed algorithms to rapidly approach the optimal solution under different problem settings.
{"title":"Solving Markov decision processes via state space decomposition and time aggregation","authors":"Rodrigo e Alvim Alexandre, Marcelo D. Fragoso, Virgílio J.M. Ferreira Filho, Edilson F. Arruda","doi":"10.1016/j.ejor.2025.01.037","DOIUrl":"https://doi.org/10.1016/j.ejor.2025.01.037","url":null,"abstract":"Although there are techniques to address large scale Markov decision processes (MDP), a computationally adequate solution of the so-called curse of dimensionality still eludes, in many aspects, a satisfactory treatment. In this paper, we advance in this issue by introducing a novel multi-subset partitioning scheme to allow for a distributed evaluation of the MDP, aiming to accelerate convergence and enable distributed policy improvement across the state space, whereby the value function and the policy improvement step can be performed independently, one subset at a time. The scheme’s innovation hinges on a design that induces communication properties that allow us to evaluate time aggregated trajectories via absorption analysis, thereby limiting the computational effort. The paper introduces and proves the convergence of a class of distributed time aggregation algorithms that combine the partitioning scheme with two-phase time aggregation to distribute the computations and accelerate convergence. In addition, we make use of Foster’s sufficient conditions for stochastic stability to develop a new theoretical result which underpins a partition design that guarantees that large regions of the state space are rarely visited and have a marginal effect on the system’s performance. This enables the design of approximate algorithms to find near-optimal solutions to large scale systems by focusing on the most visited regions of the state space. We validate the approach in a series of experiments featuring production and inventory and queuing applications. The results highlight the potential of the proposed algorithms to rapidly approach the optimal solution under different problem settings.","PeriodicalId":55161,"journal":{"name":"European Journal of Operational Research","volume":"1 1","pages":""},"PeriodicalIF":6.4,"publicationDate":"2025-02-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443574","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"管理学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：