Article
Management
Elvan Gokalp, M. Selim Cakir, Hasan Satis
Summary: The COVID-19 pandemic has caused hospitals to be flooded with patients, leading to overcrowding of healthcare resources. Instead of reactively expanding capacities, a proactive approach is proposed to address future uncertainties in demand due to COVID-19. A stochastic and dynamic model is developed to determine the appropriate amount of capacity increase in critical hospital resources. Experiments based on data from a large tertiary hospital in Turkey demonstrate that Approximate Dynamic Programming outperforms a benchmark myopic heuristic. Sensitivity analysis is also conducted to examine the impact of different epidemic dynamics and cost parameters on the results.
JOURNAL OF THE OPERATIONAL RESEARCH SOCIETY
(2023)
Article
Engineering, Industrial
Franco Quezada, Celine Gicquel, Safia Kedad-Sidhoum
Summary: This paper aims to optimize the production planning of a three-echelon remanufacturing system under uncertain input data. A multi-stage stochastic integer programming approach is considered, and scenario trees are used to represent the uncertain information structure. A new dynamic programming formulation is introduced based on a partial nested decomposition of the scenario tree. A new approximate stochastic dual dynamic integer programming algorithm is proposed based on this partial decomposition. The numerical results show that the proposed solution approach can provide near-optimal solutions for large-size instances with a reasonable computational effort.
INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH
(2023)
Article
Management
Kerem Akartunali, Stephane Dauzere-Peres
Summary: This paper proposes and studies a novel approach to modeling uncertainty in demand for the single-item dynamic lot sizing problem. The uncertainty is not related to the demand quantity, but rather to the demand timing. The paper introduces a modeling technique that naturally correlates uncertain demands in different periods, and presents dynamic programming algorithms for the general case and special cases with stochastic demand timing. Additionally, the paper proves that the most general case, where the backlog cost depends both on the time period and the stochastic demand, is NP-hard.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2022)
Article
Management
Dimitris Bertsimas, Christopher McCord, Bradley Sturt
Summary: This article presents a data-driven approach that incorporates side information into multi-stage stochastic programming using predictive machine learning methods. The proposed method achieves asymptotic optimality for multi-period stochastic programming with side information, and introduces a general-purpose approximation for dynamic problems that produces high-quality solutions.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2023)
Article
Management
Luca Bertazzi, Riccardo Mogre, Nikolaos Trichakis
Summary: We solve the problem of managing a project or complex operational process by controlling the pace of its activities. We propose a Markov decision process in which the manager balances expediting costs and delay costs by expediting certain activities. We find exact solution methods that significantly reduce computational burden and demonstrate the applicability of our approach to various problems.
MANAGEMENT SCIENCE
(2023)
Article
Operations Research & Management Science
Thuener Silva, Davi Valladao, Tito Homem-de-Mello
Summary: The study introduces a data-driven prescriptive analytics framework that integrates machine learning and dynamic optimization mechanisms to build a bridge from data to decisions. This framework is applicable to dynamic decision problems and demonstrates consistency and efficiency.
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS
(2021)
Article
Transportation Science & Technology
Panagiotis Typaldos, Markos Papageorgiou
Summary: This paper focuses on developing efficient numerical algorithms to solve the stochastic Green Light Optimal Speed Advisory (GLOSA) problem. Two modified versions of Dynamic Programming, namely Discrete Differential Dynamic Programming (DDDP) and Differential Dynamic Programming (DDP), are proposed and tested. The results show that these algorithms can achieve instantaneous or extremely fast solutions with reduced computation times compared to the standard SDP algorithm.
TRANSPORTATION RESEARCH PART C-EMERGING TECHNOLOGIES
(2023)
Article
Management
David B. Brown, Jingwei Zhang
Summary: This paper investigates fluid relaxations for dynamic resource allocation problems. It provides easy-to-implement feasible fluid policies that achieve near-optimal performance and develops a class of fluid-budget balancing policies that achieve near-optimal performance under certain nondegeneracy assumptions.
OPERATIONS RESEARCH
(2023)
Article
Management
Qichen Deng, Bruno F. Santos
Summary: This paper proposes a lookahead approximate dynamic programming methodology to address aircraft maintenance check scheduling, aiming to minimize wasted utilization intervals between maintenance checks while reducing the need for additional maintenance slots. Through validation, the methodology is shown to potentially decrease the number of checks and increase aircraft availability while also respecting airworthiness regulations.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2022)
Article
Management
Saif Benjaafar, Daniel Jiang, Xiang Li, Xiaobo Li
Summary: The paper explores an optimal policy for systems with a general network structure in the context of on-demand rental services. It demonstrates that the optimal policy can be described in terms of a specific region in the state space and proposes a provably convergent approximate dynamic programming algorithm to handle high-dimensional problems.
MANAGEMENT SCIENCE
(2022)
Article
Computer Science, Interdisciplinary Applications
Elvan Gokalp
Summary: This study proposes a mathematical model based on stochastic dynamic programming to determine the right amount of fertilizer for plants in a citrus orchard. Through comparisons, it is found that the ADP algorithm outperforms other heuristic algorithms in various parameter ranges, which may lead to excessive fertilization in large orchards.
COMPUTERS & INDUSTRIAL ENGINEERING
(2021)
Article
Management
Santiago R. Balseiro, David B. Brown, Chen Chen
Summary: This research focuses on the dynamic pricing of resources relocating over a network, specifically in hub-and-spoke structured networks. By developing a dynamic pricing policy and performance bound based on Lagrangian relaxation, the study shows that the Lagrangian policy loses no more than O(root ln n/n) in performance compared to the optimal policy in a supply-constrained large network regime, implying asymptotic optimality as n grows large. Additionally, it is demonstrated that no static policy is asymptotically optimal in the large network regime. The research extends the Lagrangian relaxation to provide upper bounds and policies for general networks with multiple interconnected hubs and spoke-to-spoke connections, incorporating relocation times and examining performance on numerical examples.
MANAGEMENT SCIENCE
(2021)
Article
Management
Christos Zacharias, Nan Liu, Mehmet A. Begen
Summary: In this paper, a novel dynamic programming framework is introduced to address the theoretical and practical problem of simultaneous consideration of appointment day and time of day in dynamic scheduling decisions. The framework incorporates these scheduling decisions in two timescales and aims to leverage the latest theoretical developments in tackling the joint problem. The authors establish theoretical connections and develop a practically implementable and computationally tractable scheduling paradigm with performance guarantees. Numerical experiments demonstrate the effectiveness of the proposed approach.
OPERATIONS RESEARCH
(2022)
Article
Engineering, Electrical & Electronic
Bruno Colonetti, Erlon Finardi, Samuel Brito, Victor Zavala
Summary: Unit commitment is a complex problem in power system operations that has yet to be fully solved. Operators currently use optimization solvers and simplifications to address the problem, but solving it in a timely manner remains a challenge. This study proposes a parallel dynamic integer programming approach for solving the unit commitment problem, which has been successfully applied to different power systems with impressive speed-ups compared to sequential strategies.
IEEE TRANSACTIONS ON POWER SYSTEMS
(2023)
Article
Operations Research & Management Science
N. Azevedo, D. Pinheiro, S. Pinheiro
Summary: In this paper, we study a stochastic optimal control problem with state variable dynamics described by a stochastic differential equation modulated by a semi-Markov process. We provide a detailed proof of the dynamic programming principle and show that the value function can be characterized as a viscosity solution of the corresponding Hamilton-Jacobi-Bellman equation. We illustrate our results with an application to the generalization of Merton's optimal consumption-investment problem to financial markets with semi-Markov switching.
