Water sustainability, which is the effective management of water resources to meet growing demand while mitigating environmental impacts, is imperative for fostering a resilient and thriving future. To mitigate the increasing risk of water shortages and associated greenhouse gas (GHG) emissions, recycled water offers an alternative source. To address the dual challenges of resource scarcity and environmental sustainability, this study develops a circular internal structure by categorizing water utility into four operational divisions while introducing recycled water as a carryover product and considering GHG emissions as undesirable outputs within the evaluation framework. This proposed dynamic four-stage network based on the directional distance function (DDF) was employed to measure the technical efficiency of 47 integrated water and wastewater utilities in Australia from 2019 to 2023. The results reveal an aggregated overall efficiency score of 0.8080, suggesting a potential 19.2
Supply-chain disruptions vary in intensity, scale, and frequency, producing temporally heterogeneous impacts on network viability. The emphasis on exposure and topology has anchored resilience in network structure, obscuring how collaboration and trust govern its phase-specific temporal dynamics, including resistive capabilities, adaptation speed, containment, and recovery. This study thus develops an agent-based simulation of the Netherlands logistics network, in which nodes operate as autonomous agents that redistribute load under capacity, link, and trust constraints. Cascading failures are simulated with trust-gated, capacity-aware, and collaboration-dependent load reallocation, capturing both immediate contagion and multi-step redistribution dynamics. The framework compares rule-based heuristics with adaptive AI agents trained using supervised imitation, online multiplicative adaptation, and cross-entropy policy search. Temporal resilience is then measured across each phase specific metrics of absorptive, adaptive, and recovery phases across the collaboration structure. This points to the main contribution of this article, in which AI agents improve adaptation and cascade containment through targeted use of high-trust, high-capacity links, while recovery dynamics remain bounded by structural and exogenous constraints. This points to a layered view of resilience, where performance depends on the interaction between network configuration, trust-mediated coordination, and agent-level adaptation.
Demand information asymmetry is ubiquitous and severely undermines the performance of suppliers and online platforms in fresh e-tailing supply chains. Information sharing is vital for mitigating information asymmetry. It directly shapes members’ pricing and freshness-keeping decisions, and ultimately determines whether a platform adopts a reselling or an agency mode. We aim to study the interplay between demand information sharing strategy (sharing vs. non-sharing) and selling mode (reselling vs. agency) selection in a fresh e-tailing supply chain in this work. We find that the optimal strategy highly hinges on both consumer freshness sensitivity and prediction accuracy. Specifically, if consumer freshness sensitivity is moderate and the prediction accuracy is low, the platform prefers the agency mode and shares the demand information; otherwise, adopting the reselling mode and withholding information is the platform’s optimal choice. Under certain conditions, the platform’s optimal strategy can also improve the profitability of both the supplier and the whole supply chain. Furthermore, we examine the platform’s optimal strategy with a sophisticated supplier and find that, if consumer freshness sensitivity is moderate and the prediction accuracy is high, adopting the agency mode and sharing information will improve the platform’s profit. Interestingly, we confirm that the presence of a sophisticated supplier creates a win–win outcome for both parties under certain conditions.
This paper proposes an algorithm integrating random forests (RF) and data envelopment analysis (DEA), termed RF-DEA, to address several methodological challenges commonly encountered in DEA applications, e.g., sensitivity to outliers and the ‘curse of dimensionality’. We show that by leveraging the strengths of RF, the DEA estimator with a calibrated frontier and input/output vectors can effectively mitigate the influence of outliers, identify redundant variables, and account for the influence of environmental variables. Evidence from Monte Carlo simulations demonstrates that these properties of RF-DEA yield substantial improvements in estimation accuracy compared to the conventional DEA. Through an empirical application to global production efficiency, we illustrate that this easy-to-implement method constitutes a viable alternative DEA estimator, capable of disentangling some practical challenges in efficiency analysis.
Sequential decision-making under uncertainty lies at the core of operations research, statistics, and modern machine learning. Among its most influential formulations is the multi-armed bandit (MAB) problem, which formalizes the exploration-exploitation trade-off through adaptive allocation of uncertain alternatives. This article presents a comprehensive festschrift in honor of Michael N. Katehakis, whose work has played a central role in shaping the structural and information-theoretic foundations of bandit theory and its applications. We trace the development of project index theory and discounted bandits, highlighting the structural decomposition principles that reduce complex stochastic control problems to scalar index comparisons. We then examine the Lai-Robbins lower bound and the Burnetas-Katehakis asymptotic optimality framework, which constructs allocation rules attaining the exact logarithmic regret constant dictated by Kullback–Leibler divergence. These contributions establish a unified perspective in which dynamic programming, large deviations theory, and statistical efficiency converge. Building on this foundation, we survey extensions to constrained, restless, contextual, combinatorial, non-stationary, and adversarial bandits, as well as connections to reinforcement learning. Across these generalizations, a common structural paradigm persists: optimal exploration is governed by quantified uncertainty, and complex allocation problems admit tractable solutions through indexability and Lagrangian relaxation. We further examine applications in adaptive clinical trials, dynamic pricing and revenue management, inventory control with learning, online recommendation systems, finance, and service operations. In each domain, index-based and regret-optimal policies provide computationally implementable decision rules with provable long-run efficiency guarantees. By situating these developments within the broader evolution of sequential decision theory, this article highlights the enduring influence of Katehakis’s contributions and underscores the continuing relevance of structural and information-theoretic principles in modern data-driven optimization.
In light of the escalating popularity of online shopping and the urgent need to reduce unnecessary driving to mitigate environmental impacts, it has become increasingly important to provide cost-effective solutions for attended home deliveries. Extensive research efforts have been dedicated to addressing challenges related to integrating demand management and vehicle routing with time windows. In this paper, we present two key contributions. Firstly, we propose an enhanced method for estimating opportunity cost, by leveraging a dynamic-routing and distribution approach that incorporates forecast orders. This approach allows for more accurate revenue-loss assessments, ultimately leading to improved decision-making on the delivery charges. Secondly, we introduce a dynamic slot-combination strategy, aiming to fully exploit the flexibility that customers possess in receiving their delivery, which enhances overall route efficiency and customer satisfaction. Importantly, our proposed augmented time-windows approach can be easily implemented within existing systems, employing standard time windows, without necessitating any strategic changes or complex computational modifications to the routing system. To assess the performance of our proposed approach, we conducted exhaustive experiments on real data. The results demonstrate that our generated solution outperforms both recent and current state-of-the-art approaches in terms of profitability and delivery efficiency. This signifies the effectiveness and practicality of our proposed methodology in addressing the challenges associated with attended home deliveries.
Sustainability and emissions reduction are crucial in modern waste management. This article explores the integration of circular economy principles with advanced optimization techniques to enhance resource efficiency and minimize environmental impact. The focus is on a multi-period bi-objective vehicle routing problem for municipal waste collection, integrating inventory dynamics at bins, capacity constraints, and service time windows to prevent overflow and ensure operational feasibility over a planning horizon. This research aims to optimize travel costs and load dependent CO_2 emissions combining an exact multi-objective approach using Gurobi with an approximate approach based on NSGA-II, enhanced through local search, and selects final compromise solutions using TOPSIS under different decision priorities. The proposed methodology is validated on a real-world case study from a zone in municipality of Tlemcen, Algeria, across multiple instances. The results highlight clear cost–emission tradeoffs, demonstrate the scalability limits of exact solvers in multi-period settings, and show that the hybrid NSGA-II approach consistently provides high quality, near-optimal solutions for large scale sustainable waste collection planning.
This paper studies a stochastic Home Health Care Routing Problem (HHCSRP) that integrates routing, assignment, and scheduling decisions under uncertainty. The problem considers heterogeneous caregivers, skill requirements, time windows, and patient-career preferences. Unlike most existing works, we jointly model stochastic patient demand and stochastic service times within a multi-objective framework. We formulate the problem as a bi-objective stochastic mixed-integer program that minimizes total operational costs while maximizing service quality. To address uncertainty, we propose a hybrid approach combining chance-constrained programming and recourse modeling, and derive an equivalent deterministic formulation using goal programming techniques. Computational experiments in benchmark and real-world-inspired cases demonstrate the effectiveness of the proposed model in producing robust and cost-effective schedules. The main contribution lies in the integrated treatment of multiple uncertainty sources within a unified multi-objective HHCSRP and the development of a tractable chance-constrained recourse framework.
Multi-sensor bearing fault diagnosis can take full advantage of the complementary information across heterogeneous signals; however, real-world deployments often involve sensor dropout, temporal gaps, and sporadic point-wise data loss. To improve diagnostic robustness under diverse mixed missingness conditions, this paper proposes a Missing-aware Adaptive Expert Fusion Framework (MAEFF), which enables adaptive decision-making under incomplete observations through reliability-guided multi-sensor fusion and relation-aware residual correction. Unlike methods that rely on signal reconstruction or data imputation, MAEFF directly incorporates missing-awareness into the fusion process itself, and further improves robustness and training stability through missingness-injected data and a staged optimization strategy. Extensive experiments on the Ottawa multi-sensor bearing dataset and the KAIST rotating-machine dataset show that, under three practically motivated missingness patterns, MAEFF achieves strong diagnostic performance under fully observed inputs and degrades gracefully as missingness increases. Across missingness ratios ranging from fully observed inputs to settings where every segmented signal window contains missing observations, MAEFF consistently maintains high accuracy and competitive macro-F1 scores, while outperforming advanced multi-sensor fusion baselines under severe missingness and achieving a favorable accuracy-efficiency trade-off. These results indicate that robust missing-aware fusion is an important direction for practical intelligent diagnosis, and that the proposed framework provides a flexible foundation for building resilient multi-sensor PHM systems in non-ideal sensing environments.
In many hospitals, treatment areas are typically dedicated to a single medical or surgical specialty. While shared use by multiple specialties is uncommon, it can be necessary, or even advantageous in certain situations. This article introduces a proof-of-concept mathematical framework for strategically reorganizing treatment-area functions to enhance resilience and operational flexibility. We focus on configuring a hospital so it can robustly treat a selected repertoire of patient caseloads (a.k.a., case mix) should they arise. Using mixed-integer programming and metaheuristic optimization, our solution approach determines which patient types should share each treatment area, maximizing the number of targeted caseloads that can be accommodated. When caseloads vary greatly or are infeasible, the model seeks a best-fit configuration that maximizes fulfilment across all caseload profiles. Because solving the exact model is computationally intractable, we propose and evaluate two matheuristics, one that utilises Simulated Annealing and the other Variable Neighbourhood Search. Numerical testing on a large, real-world instance demonstrates their effectiveness in finding near-optimal, practical solutions, providing strong motivation for further research and validation across diverse healthcare settings.
Practical decision-making in business and government regularly involves interaction with competitors (and other agents such as suppliers and customers). These interactions can have a significant effect on optimal decision making. To determine optimal prices, for example, a monopoly firm may need to vary prices to learn consumer demand, but, with competition, varying prices may provide information to competitors that is counterproductive. This paper shows that, while firms in competition may be able to learn optimal best-response strategies over time, in some settings, this result is not an ex ante optimal strategy for the competitors when actions reveal information to others. In this situation, the firms can achieve a collaborative outcome in equilibrium by not fully exploring and learning the environment with both firms earning more than they would in a competitive equilibrium with full information. The result has implications for policies (such as price or production controls) that restrict firm actions or that require disclosures.
Environmental, social, and governance (ESG) metrics are widely adopted tools for assessing corporate sustainability. Despite their importance for investors, regulators, and other stakeholders, ESG scores suffer from major drawbacks, including score divergence across providers, opaque and biased methodologies, and data quality issues. This paper focuses on Eikon Refinitiv, a prominent ESG data provider, to explore two interrelated challenges: the distortion of association measures due to missing data imputation and the potential redundancy among ESG key performance indicators (KPIs). First, we show that Refinitiv’s percentile ranking scheme preserves rank correlations (Kendall’s tau, Spearman’s rho) but not Pearson’s correlation when data are complete, but Refinitiv’s imputation for missing values with zeros inflates all association measures. Second, we develop an optimization-based tool to identify informational redundancy among KPIs. We formulate it as a cardinality-constrained rank correlation maximization problem and solve it via stochastic hill climbing with random restarts. Across three sectors and 10 years, we find that only 20–25
This paper mainly deals with a multi-criteria traffic network equilibrium problem with flexible demands and capacity constraints of arcs. A multi-objective minimum cost flow problem with flexible demands is firstly introduced. A set of (weak) vector minimum cost flows is obtained by using an inexact multi-objective augmented Lagrangian method. Since (weak) vector minimum cost flows may not be a (weak) vector equilibrium flows, we pay attention to constructing equivalent optimization problems for weak vector equilibrium flows and vector equilibrium flows in terms of an activation function ReLU and a vector version of the Heaviside step function, respectively. In addition, numerical methods based on smoothing the objective functions are proposed to generalize the set of the (weak) vector equilibrium flows. Finally, extensive numerical examples are given to illustrate excellence of our algorithms compared with existing algorithms. It is shown that the proposed smoothing method is more efficient and can find more vector equilibrium flows.
This study bridges operational research and financial analysis to investigate the interdependencies between price discovery processes and media sentiments related to commodities in the Chinese futures market. Across a data-rich span from January 2018 to May 2023, we systematically aggregate and assess daily sentiment data extracted from a diverse array of media sources. By adapting the Robustly-optimized Bidirectional Encoder Representations from Transformers pretraining approach, our Operational Research processing improves traditional sentiment scoring methods, enabling a holistic assimilation of semantic data that encapsulates both influential and consequential market factors. Our empirical investigation focusing on the predictive modelling of Chinese commodity futures’ returns, identifies sentiment as a key extra element, also confirming the most relevant factors already identified in the literature, showing an enhanced adjustment effect of sentiment in the Chinese market compared to findings in other international futures markets. The incorporation of extensive semantic analysis into our Operational Research models affords a fresh vantage point upon sentiment’s role in shaping market behaviours, thus informing some past conclusions. This confluence of Operational Research textual analysis methodology and financial analysis enriches the available toolkit for market strategists, and offers a new conduit for academic and practitioner inquiries into financial markets’ dynamics.
We propose a novel statistical procedure which combines forecasting reconciliation techniques and unsupervised clustering algorithms to forecast annual greenhouse gases emissions at the territorial level. Specifically, we aim at predicting future emissions from the agricultural sector for Europe by aggregating bottom-level regional forecasts to the national and continental levels. Fuzzy clustering is used to define aggregations of bottom and middle-level series into clusters based on the similarity of emissions patterns; then, the fuzzy hierarchies are combined with the original administrative aggregation structure to improve the prediction accuracy of forecasting models. Empirical results show that fuzzy clustering aggregation, by leveraging uncertainty compared to crisp approaches, enhances forecast reconciliation accuracy at all levels of the hierarchy. Our approach offers a scalable solution to enhance forecast accuracy and reliability, particularly for short time series characterized by complex hierarchical structures.
The park-and-loop routing problem is a variant of the vehicle routing problem in which each driver can operate walking subtours while their truck is parked at a customer. A maximum walking distance per route and a maximum route duration are considered. This problem is particularly relevant in city logistics, where customers are clustered, and truck use can be reduced, thereby decreasing traffic congestion and emissions. We present a new exact branch-cut-and-price approach to solve the problem. In this approach, partial routes corresponding to walking subtours are enumerated using an efficient dynamic program. The problem is solved using a state-of-the-art exact branch-cut-and-price algorithm, which employs a bidirectional labeling algorithm for the pricing problem. Computational experiments demonstrate that the algorithm increases the size of optimally solved instances by 50
In this paper, we consider the distributed permutation flowshop scheduling problem with a total weighted tardiness objective. Our focus is on dispatching rules, which are widely used in practical settings and can quickly provide solutions for even quite large instances. We consider a total of 23 dispatching rules. Twelve are adaptations to the distributed flowshop setting of existing heuristics, two are simple distributed flowshop rules, four are modified versions of two procedures developed for the distributed flowshop with a blocking constraint and five are newly developed procedures. These dispatching rules are combined with three factory assignment rules, as well as with two different approaches (one newly proposed) for dealing with the last few jobs. Therefore, a total of 138 procedures are considered and evaluated. Extensive preliminary tests were performed to determine appropriate values for the parameters required by some of the procedures. The heuristics were then evaluated on a dataset that is both quite large, and diverse in terms of due dates. Multiple performance measures, as well as statistical tests, were used to evaluate the procedures, the factory assignment rules and the two ways of dealing with the last few jobs. The computational results show that a new approach for dealing with the last few jobs outperforms its traditional counterpart. The top positions in terms of solution quality were consistently occupied by the newly developed rules. In particular, the WIT_WCT_3 procedure achieved, with a single exception, the best performance across all solution quality performance measures and factory assignment rules.
