Article
Operations Research & Management Science
Wei Liu, Li Yang, Bo Yu
Summary: This paper proposes a distributionally robust mean-HMCR portfolio optimization model using kernel density estimation and phi-divergence to address the curse of dimensionality. Empirical tests demonstrate that the portfolio strategy obtained by the proposed model outperforms other strategies in most cases, showing higher quality in terms of performance.
ANNALS OF OPERATIONS RESEARCH
(2021)
Article
Computer Science, Information Systems
Jose Almeida, Joao Soares, Fernando Lezama, Zita Vale
Summary: This paper proposes a risk-based optimization approach for centralized day-ahead energy resource management (ERM) considering extreme events. The risk-averse strategy implements the conditional value-at-risk (CVaR) method to account for worst-case scenario costs. The case study shows that the risk-averse strategy increases operational costs but reduces worst-case scenario costs, providing safer and more robust solutions.
Article
Operations Research & Management Science
Kerem Ugurlu
Summary: A new operator for handling joint risk of different sources has been presented and investigated. It proposes a multivariate risk measure called multivariate average-value-at-risk, m AVaR(alpha), to quantify total risk. The operator satisfies the four axioms of a coherent risk measure and can be simplified to the conventional average-value-at-risk in certain cases. It also offers flexibility by allowing investors to choose different risk levels for each random loss. The research explores the connection between multivariate tail variance and m AVaR(alpha) using the Chebyshev inequality for tail events.
Article
Robotics
Astghik Hakobyan, Insoon Yang
Summary: This article introduces a novel model-predictive control method for mobile robots to avoid randomly moving obstacles, even without knowing the true probability distribution of uncertainty. By utilizing a statistical ball as the ambiguity set, it achieves a probabilistic guarantee of out-of-sample risk and resolves the infinite-dimensionality issue in the distributionally robust MPC problem. The proposed method demonstrates successful avoidance of randomly moving obstacles and guarantees out-of-sample risk even with a small sample size, outperforming its sample average approximation counterpart.
IEEE TRANSACTIONS ON ROBOTICS
(2022)
Article
Mathematics
Wenhua Lv, Linxiao Wei
Summary: This paper investigates a distributionally robust reinsurance problem that combines Glue Value-at-Risk and the expected value premium principle. The problem focuses on stop-loss reinsurance contracts with known mean and variance of the loss. The optimization problem is formulated as a minimax problem, where the inner problem involves maximizing over distributions with the same mean and variance. The paper provides analytical solutions and representations for the inner problem in different cases.
Article
Computer Science, Interdisciplinary Applications
J. G. Hoffer, S. Ranftl, B. C. Geiger
Summary: This article discusses how to find an input such that the output of a stochastic black box function is as close as possible to a target value. It fills the gap in current approaches by deriving acquisition functions for common criteria and demonstrating their compatibility with certain extensions of Gaussian processes. The experiments show that these derived acquisition functions can outperform classical Bayesian optimization.
COMPUTERS & INDUSTRIAL ENGINEERING
(2023)
Article
Mathematics, Applied
Sebastian Garreis, Thomas M. Surowiec, Michael Ulbrich
Summary: In the presence of uncertainty in engineering and natural science models, the incorporation of random parameters in partial differential equations is necessary. This leads to infinite-dimensional stochastic optimization problems, which often require the use of risk measures in the objective function. The proposed log-barrier risk measure method offers a novel approach to solving risk-averse PDE-constrained optimization problems.
SIAM JOURNAL ON OPTIMIZATION
(2021)
Article
Mathematics
Haoyu Chen, Kun Fan
Summary: Recent empirical evidence shows that financial risk exhibits a heavy-tailed distribution. Building on advances in generalized quantile risk measures, a tail value-at-risk (TVaR)-based expectile is proposed to capture tail risk compared to the classic expectile. This study not only presents the well-defined nature of the risk measure but also investigates its coherency properties. The asymptotic expansion of a TVaR-based expectile, with respect to quantiles, is studied for extreme risks commonly modeled by a regularly varying survival function. Additionally, a closed-form expression for the worst-case TVaR-based expectile is derived based on moment information, motivated by recent developments in distributionally robust optimization for portfolio selection. Numerical results demonstrate the performance of the new risk measure compared to classic risk measures, such as tail value-at-risk-based expectiles.
Article
Computer Science, Artificial Intelligence
Zhao-Rong Lai, Cheng Li, Xiaotian Wu, Quanlong Guan, Liangda Fang
Summary: In this study, a novel multitrend conditional value at risk (MT-CVaR) method is proposed for portfolio optimization, incorporating multiple trends and their influences to achieve state-of-the-art investing performance and risk management.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Economics
Noureddine Kouaissah
Summary: This paper presents robust portfolio optimization models that significantly improve upon conventional portfolio selection techniques by addressing estimation error and using an early-warning system. Empirical analyses based on the US stock market validate the proposed robust approaches and show their superiority in out-of-sample portfolios.
Article
Mathematics, Applied
Man Yiu Tsang, Tony Sit, Hoi Ying Wong
Summary: This paper introduces a distributionally robust multi-period portfolio model with ambiguity on asset correlations, utilizing the spectral risk measure for flexibility and protection. The SDDP algorithm is employed as a numerical method, and its convergence property is verified under finite scenarios.
APPLIED MATHEMATICS AND OPTIMIZATION
(2022)
Article
Economics
Haiyan Liu, Tiantian Mao
Summary: This paper studies a distributionally robust reinsurance problem by minimizing the maximum Value-at-Risk of the total retained loss of the insurer for all loss distributions with known mean and variance. A three-point distribution is proposed to achieve the worst-case VaR, and the closed-form solutions of the worst-case distribution and optimal deductible are obtained.
INSURANCE MATHEMATICS & ECONOMICS
(2022)
Article
Mathematics
Moch Panji Agung Saputra, Diah Chaerani, Andrea Sukono, Mazlynda Md Yusuf
Summary: The digitalization of bank data and financial operations poses a significant risk, which can be mitigated through ready reserve funds. To optimize reserve fund allocation and minimize excessive reserve funds, linear programming optimization approach with Extreme Value-at-Risk (EVaR) constraints is used. This approach considers potential losses from digital banking risks and produces an efficient allocation of bank reserve funds.
Article
Computer Science, Artificial Intelligence
Anushri Dixit, Mohamadreza Ahmadi, Joel W. Burdick
Summary: This paper investigates the problem of risk-averse receding horizon motion planning for agents with uncertain dynamics in the presence of stochastic, dynamic obstacles. The proposed model predictive control (MPC) scheme formulates the obstacle avoidance constraint using coherent risk measures. A waypoint following algorithm using the MPC scheme is also proposed and proved to be risk-sensitive and recursively feasible while guaranteeing finite-time task completion. The paper further explores commonly used coherent risk metrics and proposes a tractable incorporation within MPC. Simulation studies are conducted to illustrate the framework.
ARTIFICIAL INTELLIGENCE
(2023)
Article
Management
Stefano Nasini, Martine Labbe, Luce Brotcorne
Summary: This paper introduces a novel optimization framework for multi-market portfolio management, where the market-wise portfolio selection is delegated to specialized affiliates. The problem is characterized as a single-leader-multi-follower game, with theoretical insights and numerical solution approaches provided. The study shows that the problem is NP-Hard, and proposes a decomposition procedure and strong valid inequalities to improve computational efficiency.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2022)
Article
Computer Science, Software Engineering
Wenzhuo Yang, Melvyn Sim, Huan Xu
MATHEMATICAL PROGRAMMING
(2020)
Article
Computer Science, Hardware & Architecture
Selin Damla Ahipasaoglu, Karthik Natarajan, Dongjian Shi
Article
Computer Science, Software Engineering
Divya Padmanabhan, Karthik Natarajan, Karthyek Murthy
Summary: This paper identifies partial correlation information structures for evaluating the maximum expected value of mixed integer linear programs with random objective coefficients. A reduced semidefinite programming formulation is developed, leading to efficient polynomial-time solvable instances, particularly in appointment scheduling and project evaluation fields.
MATHEMATICAL PROGRAMMING
(2021)
Article
Management
Le Thi Khanh Hien, Melvyn Sim, Huan Xu
OPERATIONS RESEARCH
(2020)
Article
Management
Zhi Chen, Melvyn Sim, Peng Xiong
MANAGEMENT SCIENCE
(2020)
Article
Management
Taozeng Zhu, Jingui Xie, Melvyn Sim
Summary: The study introduces a joint estimation and robustness optimization framework to address the impact of estimation uncertainty in optimization problems. By incorporating both the parameter estimation process and the optimization problem seamlessly, the framework aims to obtain solutions that are immune to parameter perturbations. The size of the uncertainty set, based on the accuracy of parameter estimation from data using specific procedures, is maximized to achieve this goal.
MANAGEMENT SCIENCE
(2022)
Article
Management
Yu Zhang, Zhenzhen Zhang, Andrew Lim, Melvyn Sim
Summary: The study proposes a distributionally robust optimization model for a vehicle routing problem with time windows, utilizing a Wasserstein distance-based ambiguity set to minimize delays while limiting travel costs. The proposed solution significantly improves on-time arrival performance with only modest increases in expenditures and outperforms traditional decision criteria in out-of-sample simulations.
OPERATIONS RESEARCH
(2021)
Article
Management
Bikramjit Das, Anulekha Dhara, Karthik Natarajan
Summary: The study shows that for the distributionally robust newsvendor problem, the order quantity selection is optimal for heavy-tailed distributions as observed in Scarf's work, and remains valid even under information on the first and αth moments within the confidence set.
OPERATIONS RESEARCH
(2021)
Article
Management
Daniel Zhuoyu Long, Melvyn Sim, Minglong Zhou
Summary: We present a general framework for robust satisficing that favors solutions which can achieve an acceptable target even when the actual probability distribution deviates from the empirical distribution. By balancing the model's ability to withstand uncertainty, the decision maker specifies an acceptable target or loss of optimality compared to the empirical optimization model. Through numerical studies, it is shown that solutions to the robust satisficing models are more effective in improving out-of-sample performance.
OPERATIONS RESEARCH
(2023)
Article
Management
Patrick Jaillet, Gar Goei Loke, Melvyn Sim
Summary: This study introduces a new approach to address the workforce planning issues of hiring, dismissing, and promoting while ensuring compliance with organizational constraints. By considering factors such as employees' time-in-grade, a new model is proposed to meet constraints under uncertainty, providing insights into HR management.
OPERATIONS RESEARCH
(2022)
Article
Management
Jingui Xie, Gar Goei Loke, Melvyn Sim, Shao Wei Lam
Summary: Bed shortages in hospitals have negative consequences on patient satisfaction and medical outcomes. Traditional metrics such as bed occupancy rates (BORs) are insufficient in capturing the risk of bed shortages. We propose the bed shortage index (BSI) to capture more facets of bed shortage risk. Our metric is based on the riskiness index by Aumann and Serrano and can be easily computed without additional assumptions. We also propose optimization models to plan for bed capacity using this metric.
OPERATIONS RESEARCH
(2023)
Article
Management
Georgia Perakis, Melvyn Sim, Qinshen Tang, Peng Xiong
Summary: We propose a new distributionally robust optimization model for a two-period, multiitem joint pricing and production problem, which utilizes historical demand and side information for demand prediction. By introducing a partitioned-moment-based ambiguity set, we characterize the residuals of an additive demand model and determine the evolution of the second-period demand from the first-period information in a data-driven setting. We investigate the problem by proposing a cluster-adapted markdown policy and affine recourse adaptation, reformulating it as a mixed-integer linear optimization problem and solving it to optimality using commercial solvers. We also extend our framework to ensemble methods using ambiguity sets constructed from different clustering approaches. Numerical experiments and a case study demonstrate the benefits of the cluster-adapted markdown policy and the partitioned moment-based ambiguity set in improving the mean profit over empirical models in most out-of-sample tests.
MANAGEMENT SCIENCE
(2023)
Article
Engineering, Manufacturing
Minglong Zhou, Melvyn Sim, Shao-Wei Lam
Summary: The study addresses the issue of advance scheduling of ward admission requests in a public hospital and proposes a resource satisficing framework to reduce the risks of resource overutilization.
PRODUCTION AND OPERATIONS MANAGEMENT
(2022)
Article
Management
Georgia Perakis, Melvyn Sim, Qinshen Tang, Peng Xiong
Summary: We propose a new distributionally robust optimization model for a two-period, multiitem joint pricing and production problem, which can be implemented in a data-driven setting using historical demand and side information. We introduce a new partitioned-moment-based ambiguity set to characterize the residuals of an additive demand model and investigate the problem by proposing a cluster-adapted markdown policy and an affine recourse adaptation. Experimental results demonstrate the effectiveness of the cluster-adapted markdown policy and the partitioned moment-based ambiguity set in improving the mean profit.
MANAGEMENT SCIENCE
(2023)
Article
Computer Science, Interdisciplinary Applications
Anulekha Dhara, Bikramjit Das, Karthik Natarajan
Summary: The paper introduces a novel approach in risk management by utilizing ambiguity sets with a tree structure to compute and minimize the worst-case bound on a portfolio. This method provides flexibility in modeling and computational properties.
INFORMS JOURNAL ON COMPUTING
(2021)
