The TutORials in Operations Research series is published annually by INFORMS as an introduction to emerging and classical subfields of operations research and management science. These chapters are designed to be accessible for all constituents of the INFORMS community, including current students, practitioners, faculty, and researchers. The publication allows readers to keep pace with new developments in the field and serves as augmenting material for a selection of the tutorial presentations offered at the INFORMS Annual Meeting.
Simulation Optimization in the New Era of AI
We review simulation optimization methods and discuss how these methods serve as the bedrock of modern AI techniques. Particularly, we focus on three areas: stochastic gradient estimation that plays a central role in training neural networks for deep learning and reinforcement learning, ranking and selection that can be used as the node selection policy in Monte Carlo tree search, and variance reduction that is critical for accelerating the training procedures in AI.
Speakers: Yiijie Peng and Chun-Hung Chen
Mixed-Integer Programming Approaches to Generalized Submodular Optimization and its Applications
Submodularity is an important concept in integer and combinatorial optimization. A classical submodular set function models the utility of selecting homogenous items from a single ground set, and such selections can be represented by binary variables. In practice, many problem contexts involve choosing heterogenous items from more than one ground set or selecting multiple copies of homogenous items, which call for extensions of submodularity. We refer to the optimization problems associated with such generalized notions of submodularity, Generalized Submodular Optimization (GSO.) GSO is found in wide-ranging applications, including infrastructure design, healthcare, online marketing, and machine learning. Due to the often highly nonlinear (even non-convex and non-concave) objective function and the mixed-integer decision space, GSO is a broad subclass of challenging mixed-integer nonlinear programming problems. In this tutorial, we first provide an overview of classical submodularity. Then we introduce two subclasses of GSO, for which we propose polyhedral theory for the mixed-integer set structures that arise from these problem classes. Our theoretical results lead to efficient and versatile exact solution methods that demonstrate their effectiveness in practical problems using real-world datasets.
Speaker: Simge Küçükyavuz
Pharmaceutical Supply Chains and Drug Shortages
While pharmaceutical (pharma) industry is vital to the economy and its supply chain efficiency directly affects the quality and cost of patient care, pharma supply chains have been largely under-researched in the healthcare field, compared to the thrived research in medical services/hospital operations by the INFORMS community. At the same time, the pharma industry faces complicated and unique economics and regulation environment. In this tutorial, we aim to provide a base understanding of the very complex pharmaceutical supply chain and present challenges it faces. Using drug shortages (a persistent and significant problem facing the pharma industry, government, and the society) as well as other examples, we demonstrate that the richness and uniqueness of the pharma supply chain provides great opportunities for research with impact.
Speaker: Hui Zhao
As management scientists, we increasingly rely on quantitative models to analyze problems that span across diverse realms such as supply chain management and finance. The models can be simulators or machine learning tools fitted to available data. Often, the complexity of the problems and the sophistication of the modeling exercises forces us to treat these models as black-boxes. Explainability and interpretability are then essential to reinforce stakeholders trust in quantitative models. The literature suggests analysts to answer questions such as: have we used the best available method for interpreting and explaining the results of my model? Is there a method that can help us in communicating the insights better to the decision-maker? As early as 1970, John D.C. Little writes that a process starts of finding out what it was about the inputs that made the outputs come out as they did, the task of sensitivity analysis. This TutORial reviews the role of sensitivity analysis in O.R., with a view of aiding explainability. We consider how O.R. researchers have been looking at sensitivity over time, and the techniques they have developed. We analyze available methods through the lens of four main managerial insights: factor prioritization (or feature importance), trend determination, interaction quantification, and stability (robustness). We discuss a variety of techniques and illustrate them by application to toy example and a realistic application, and illustrate the corresponding subroutines, with the goal of showcasing the wide palette of tools that the literature has made available to analysts.
Speaker: Emanuele Borgonovo
Stockpyl: A Python Package for Inventory Optimization and Simulation
Stockpyl is a Python package for inventory optimization and simulation. It implements classical single-node inventory models like the economic order quantity (EOQ), newsvendor, and Wagner–Whitin problems. It also contains algorithms for multiechelon inventory optimization (MEIO) under both stochastic-service model (SSM) and guaranteed-service model (GSM) assumptions. And, it has extensive features for simulating multi-echelon inventory systems. In this tutorial, we provide an overview of Stockpyl, including a short primer on the inventory models and algorithms that underlie the modules in Stockpyl. Python code snippets are used throughout, and a Jupyter notebook containing all of the code is available on the GitHub repository for the Stockpyl project.
Speaker: Lawrence Snyder
Supply-Chain-Centric View of Working Capital, Hedging and Risk Management: Integrated Supply Chain Finance (iSCF) Tutorial
Integrated Supply Chain Finance (iSCF) is a portfolio of effective operating, financial and risk mitigating practices and techniques within supply chains, which reflect strategic concerns of participating firm agents (decision makers) within the chain, and optimize not only the management of the working capital for liquidity, but also make effective use of assets for firm profitability and risk control. The main theory and research areas within iSCF are:
- Financing Working Capital in Supply Chains;
- Financial Hedging in support of Supply Chain Operations;
- Integrated Risk Management (IRM) in Supply Chains; and
- Supply Chain Contracts and Risk Management.
This tutorial chapter will dedicate a section in each of the above topics, and it will elucidate the foundational models and the key results in each of these areas. It will conclude with thoughts on future research topics, and how emerging technologies may be shaping the future of iSCF decision making.
Speaker: Panos Kouvelis
Capturing Emerging Targets and Performing Reconnaissance Over Uncertain Terrain
This tutorial will introduce and discuss two topics that are related to the military and security domain. The first topic is focused on search path optimization for recording emerging targets. This considers the situation where targets emerge according to a nonhomogeneous space-time Poisson process during the mission. The only provided information is the time-dependent arrival rate function for each cell in the area. The single vehicle case is discussed first, along with examples and computational results. After this, the case of camouflaging targets and multiple vehicles is considered, with a focus on how these features complicate the problem and change routing strategies. The second topic is search and exploration problems on transportation networks with unknown characteristics. In such a scenario links are divided into two classes, one class being links that are operable and the other class being links whose status is unknown and only becomes known when the traveler arrives at one of the end nodes of the link. Prize collection problems on such networks will be detailed for the single traveler case. Examples and computational results will be presented. After discussion of these two topics, the tutorial will present emerging topics related to emergency response and reconnaissance applications.
Speaker: Rajan Batta
Integer Programming Games: a Gentle Computational Overview
Nash equilibria enlighten the structure of rational behavior in multi-agent decision-making. However, besides its existence, the concept is as helpful as one can efficiently compute it. Little is known about the computation of Nash equilibria in non-convex settings, a relevant context because non-convexities, often in the form of integer requirements. We provide a gentle overview of the recent bundle of work that deals with computing Nash equilibria for integer programming games. We do that by using the general and practically relevant context of attacking and protecting a critical infrastructure, and we highlight the characteristics and compare the differences of a sequential approach (Stackelberg game) versus a simultaneous one (Nash game). Finally, we guide the reader to the use of relevant software for computing Nash equilibria for integer programming games.
Speaker: Andrea Lodi
Nonlinear Dynamics Modeling and Control of Operational Data for Process Improvements
Whenever multifarious entities cooperate, compete, or interfere in manufacturing or service operations, there will be the rise of nonlinear and nonstationary dynamics. As complex systems evolve in time, operational dynamics deal with change. Whether the system settles down to the steady state, undergoes incipient changes, or deviates into more complicated variations, it is dynamics that help analyze system behaviors. Effective monitoring, modeling and control of nonlinear dynamics will increase system quality and integrity, thereby leading to significant economic and societal impacts. For example, manufacturing processes will make products with better quality and higher throughput. Gaining a deeper understanding of nonlinear dynamics of complex diseases will help improve the delivery of healthcare services, reduce the healthcare cost, and improve the health of our society.
However, nonlinear dynamics pose significant challenges for operations engineering. Particularly, nonlinear dynamical systems defy understanding based on the traditional reductionist’s approach, in which one attempts to understand a system’s behavior by combining all constituent parts that have been analyzed separately. In order to cope with system complexity and increase information visibility, modern industries are investing in advanced sensing modalities such as sensor networks and internet-of-things technology. Real-time sensing gives rise to rich datasets pertinent to operational dynamics. Realizing the full potential of such operational data for process improvements requires fundamentally new methodologies to harness and exploit complexity. Nonetheless, there is a critical gap in the knowledge base that pertains to integrating nonlinear dynamics research with operations engineering. The theory of nonlinear dynamics has been primarily studied in mathematics and physics. There is an urgent need to harness and exploit nonlinear dynamics for creating new products (or services) with exceptional features such as adaptation, customization, responsiveness, and quality in unprecedented scales.
This tutorial presents a review of nonlinear dynamics methods and tools for real-time system informatics, monitoring and control. Specifically, we will discuss the characterization and modeling of recurrence dynamics, network dynamics, and self-organizing dynamics hidden in operational data for process improvements. Further, we contextualize the theory of nonlinear dynamics with real-world case studies and discuss future opportunities to improve the design, monitoring, and control of manufacturing and service operations. We posit this work will help catalyze more in-depth investigations and multi-disciplinary research efforts in the intersection of nonlinear dynamics and data mining for operational excellence.
Speaker: Hui Yang
A Lyapunov Approach for Finite-sample Bounds of Stochastic Approximation of Contractive Operators
We consider the stochastic approximation (SA) method to find fixed points of contractive operator. SA, is a popular approach for solving fixed point equations when the information is corrupted by noise that was first proposed by Robins and Monro. We consider a type of SA algorithms for operators that are contractive under arbitrary norms (especially the l-infinity norm). We present finite sample bounds on the mean square error, which are established using a Lyapunov framework based on infimal convolution and generalized Moreau envelope. We also present concentration bounds on the tail error, even when the iterates are not bounded by a constant. These tail bounds are obtained using exponential supermartingales in conjunction with the Moreau envelop and a novel bootstrapping approach. We then present an overview of several generalizations that can be obtained using this approach, such as seminorm contractive operators, dissipative operators etc. We illustrate the utility of these bounds through applications in Reinforcement Learning.
Speaker: Siva Theja Maguluri