As agents are built for ever more complex environments, methods that consider the uncertainty in the system have strong advantages. This uncertainty is common in domains such as robot navigation, medical diagnosis and treatment, inventory management, sensor networks and e-commerce. When a single decision maker is present, the partially observable Markov decision process (POMDP) model is a popular and powerful choice. When choices are made in a decentralized manner by a set of decision makers, the problem can be modeled as a decentralized partially observable Markov decision process (DEC-POMDP). While POMDPs and DEC-POMDPs offer rich frameworks for sequential decision making under uncertainty, the computational complexity of each model presents an important research challenge. As a way to address this high complexity, this thesis develops several solution methods based on utilizing domain structure, memory-bounded representations and sampling. These approaches address some of the major bottlenecks for decision-making in real-world uncertain systems. The methods include a more efficient optimal algorithm for DEC-POMDPs as well as scalable approximate algorithms for POMDPs and DEC-POMDPs. Key contributions include optimizing compact representations as well as automatic structure extraction and exploitation. These approaches increase the scalability of algorithms, while also increasing their solution quality.

Supply chains for time-sensitive products, and, in particular, for perishable products, pose specific and unique challenges. By definition, a perishable product has a limited lifetime during which it can be used, after which it should be discarded (Federgruen, Prastacos, and Zipkin (1986)). In this dissertation, I contribute to the analysis, design, and management of supply chain networks for perishable products with applications to healthcare. Specifically, I construct generalized network frameworks to capture perishable product supply chains in healthcare operating under either centralized or decentralized decision-making behavior. The dissertation is motivated by applications ranging from blood supply chains to pharmaceuticals, such as vaccines and medicines. The novelty of the modeling and computational framework includes the use of arc multipliers to capture the perishability of the healthcare product(s), along with waste management costs, and risk. The first part of the dissertation consists of a literature review of perishable product supply chains with a focus on healthcare along with an overview of the relevant methodologies. The second part of the dissertation formulates supply chains in healthcare operating under centralized decision-making behavior. In this part, I focus on both the operations management of and the sustainable design of blood supply chains and construct models for regionalized blood banking systems as belonging to the Red Cross. The third part of the dissertation considers competitive behavior, with a focus on the pharmaceutical industry. I construct an oligopoly supply chain network model, with differentiated brands to capture the competition among producers of substitutable drugs using game theory and variational inequality theory. Furthermore, using a case study based on real-world scenarios of a highly popular cholesterol-reducing branded drug, the impact of patent rights expiration of that brand is explored which coincides the time when its equivalent generic emerges into the markets. The calculated results are then compared to the observations from the real-word problem. Finally, the projected dynamical system formulation of the pharmaceutical network oligopoly model is derived. This dissertation is based on the following papers: Nagurney, Masoumi, and Yu (2012), Nagurney and Masoumi (2012), and Masoumi, Yu, and Nagurney (2012) as well as additional results and conclusions.

Protection of the environment is among the most pressing public policy challenges and will continue to be so long into the future. Pollution, specifically, has played a pivotal and substantial role in the degradation of the environment. Although numerous analytical techniques and computational solutions have been proposed by theoreticians from disciplines as varied as economics to environmental engineering and decision sciences, significant methodologies that provide a rigorous analysis for the modeling, qualitative analysis, and computation to solutions to environmental problems which are typically complex and large-scale remain yet to be harnessed. The goal of this dissertation is to present a series of models in marketable pollution permits that yield the profit-maximized quantities of the firms' products and the equilibrium quantities of the firms' emissions. In addition, the equilibrium allocation of pollution licenses and their prices are obtained. Furthermore, different modes of market structure, including oligopolistic behavior and non-compliant behavior, and market imperfections such as transactions costs are incorporated into the modeling framework. Investment in technology, specifically production technology and emission-abatement technology, are explicitly considered in the models. Lastly, the dissertation concludes with a shift from the static setting to a dynamic setting and the marketable pollution permit model is analyzed within a dynamic framework that allows firms to be noncompliant. The dissertation begins with a brief historical overview of the evolution of environmental economics followed by the theoretical foundations of the mathematical framework employed. The principal methodology that is utilized for the models in the dissertation is that of variational inequalities. We will also make use of the projected dynamical systems to analyze the models within a dynamic setting. We conclude this dissertation with possible extensions of the models developed and provide suggestions for future research. The dissertation is a major step in the advancement of mathematical methodologies coupled with environmental policy analysis to contribute fundamentally to the formulation and evaluation of environment policy. This research is highly interdisciplinary as it encompasses the fields of management science and operations research, environmental economics, and applied mathematics.

The main focus of this dissertation is the dynamic allocation of discrete-resources in the context of Discrete Event Systems (DES). For this purpose, we develop two algorithms that can be used to address such problems. The first one, is descent, in other words at every iteration proceeds to an allocation with a lower cost, and it is suitable for problems with separable convex structure. Furthermore, at every iteration it visits feasible allocations which makes it appropriate for use on-line. The second one, is incremental, that is, it starts with zero resources, and at every step it allocates an additional resource. Both algorithms are proven to converge in deterministic as well as stochastic environments. Furthermore, because they are driven by ordinal comparisons they are robust with respect to estimation noise and converge fast. To complement the implementation of the derived optimization algorithms we develop two techniques for predicting the system performance under several parameters while observing a single sample path under a single parameter. The first technique, Concurrent Estimation, can be directly applied to general DES. On the other hand, for the second one, referred to as Finite Perturbation Analysis (FPA), we demonstrate a general procedure for deriving this algorithm from the system dynamics. Moreover, both procedures can be used for systems with general event lifetime distributions. The dissertation ends with three applications of variations of the developed resource allocation methodologies on three practical problems. First, the incremental algorithm is used on a kanban-based manufacturing system to find the kanban allocation that optimizes a given objective function (e.g., throughput, mean delay). Next, a variation of the descent algorithm is used to resolve the channel allocation problem in cellular telephone networks as to minimize the number of lost calls. Finally, a combination of the FPA and kanban approaches are used to solve the ground holding problem in air traffic control to minimize the congestion over busy airports.

Network systems, including transportation and logistic systems, electric power generation and distribution networks as well as financial networks, provide the critical infrastructure for the functioning of our societies and economies. The understanding of the dynamic behavior of such systems is also crucial to national security and prosperity. The identification of new connections between distinct network systems is the inspiration for the research in this dissertation. In particular, I answer two questions raised by Beckmann, McGuire, and Winsten (1956) and Copeland (1952) over half a century ago, which are, respectively, how are electric power flows related to transportation flows and does money flow like water or electricity? In addition, in this dissertation, I achieve the following: (1) I establish the relationships between transportation networks and three other classes of complex network systems: supply chain networks, electric power generation and transmission networks, and financial networks with intermediation. The establishment of such connections provides novel theoretical insights as well as new pricing mechanisms, and efficient computational methods. (2) I develop new modeling frameworks based on evolutionary variational inequality theory that capture the dynamics of such network systems in terms of the time-varying flows and incurred costs, prices, and, where applicable, profits. This dissertation studies the dynamics of such network systems by addressing both internal competition and/or cooperation, and external changes, such as varying costs and demands. (3) I focus, in depth, on electric power supply chains. By exploiting the relationships between transportation networks and electric power supply chains, I develop a large-scale network model that integrates electric power supply chains and fuel supply markets. The model captures both the economic transactions as well as the physical transmission constraints. The model is then applied to the New England electric power supply chain consisting of 6 states, 5 fuel types, 82 power generators, with a total of 573 generating units, and 10 demand markets. The empirical case study demonstrates that the regional electricity prices simulated by the model match very well the actual electricity prices in New England. I also utilize the model to study interactions between electric power supply chains and energy fuel markets.

Many researchers have studied the interfacial shear stress (ISS) in nanocomposites through theoretical calculation, computational simulation or sophisticated nanomanipulation experiment measurement. In this research, we attempt to directly calculate ISS values in actual nanocomposites based on a modified Cox's model using tensile test results of various macroscopic carbon nanotube (CNT) nanocomposites. Young's modulus, tensile strength and strain of CNT ropes rather than individual CNT properties were applied into the model. The effects of functionalization, CNT rope length, volume fraction and CNT type (SWNT, DWNT, MWNT) on interfacial shear stress were studied. It was found that the functionalization increased the mechanical properties of both interfacial bonding and DWNT and MWNT rope themselves; however, it decreased the mechanical properties of SWNT ropes. The major failure mode of the CNT nanocomposites was identified to be CNT rope rupture. The calculation results revealed that the ISS values in the nanocomposites are comparable with the ones reported in literature. / A Thesis submitted to the Department of Industrial and Manufacturing Engineering in partial fulfillment of the requirements for the degree of Master of Science. / Spring Semester, 2010. / December 14, 2009. / Interfacial Shear Stress, Functionalization, Carbon Nanotube, Nanocomposite, Epoxy / Includes bibliographical references. / Richard Liang, Professor Directing Thesis; Ben Wang, Committee Member; Okenwa Okoli, Committee Member.

As Vietnam emerges into world markets, Vietnamese organizations face a predicament: how can they move up the value chain thus avoiding the economic trap of being merely low-cost producers depending on cheap labor and therefore vulnerable to newly developing countries with even lower wages? Continuous improvement (CI) practices have proved fundamental to building and sustaining competitive advantage in companies in Europe, North and South America, as well as in Asian countries such as Japan, Singapore, India, and China. CI will also be critical to Vietnamese organizations if they are to build and sustain competitive advantage. However, little is known about the use of CI in Vietnam because the language barrier, lack of reliable business data, and the country's culture of government and corporate secrecy have made it extremely difficult for academics to do management research there. The research reported here investigated the use of CI practices in Vietnam during six months of fieldwork spread over four extended trips to the country. Data were collected from two questionnaires of 661 respondents, as well as extensive interviews of 130 executives, managers, and employees at twelve of Vietnam's leading companies. Information was also gathered from interviews and discussions with 440 business leaders, academics, and individuals who have extensive knowledge of Vietnam, including expatriates who have worked and lived in Vietnam for over 40 years. This dissertation discovered that Vietnamese managers face unique problems because the deeply ingrained top-down culture prevents lower-level employees from contributing to CI efforts. Executives and managers therefore see little potential in their employees, and so put relatively little effort into developing them. Furthermore, the leadership at most case companies pursued a low-cost strategy rather than initiating improvements that would enable their organizations to compete higher up the value-chain. This research recommends ways in which companies in Vietnam can address these challenges and thus enhance their CI efforts. Given the tremendous recent growth in business activity in Vietnam, a better understanding of the management practices of companies there—what works, what does not, and why—is important for both practitioners and academics.

Critically ill patients suffering from either closed head trauma or septic shock were studied retrospectively to see if the mathematical programming technique of Data Envelopment Analysis (DEA) could be used to develop models to assess an individual patient's progress in an Intensive Care Unit (ICU). Unlike current logistic regression models that focus on the mean values for groups of patients, the DEA models evaluate each patient individually by calculating an “efficiency” score based on a patient's ability to maximize output for a given set of physiologic inputs. Patients with high efficiency scores were found to have a better chance of making a full recovery than similarly injured patients that were inefficient, even when the latter had more “normal” values for their variables. New hybrid models that combine DEA with discriminant analysis and correspondence analysis were also developed and their potential role in the ICU is explored. DEA models in the ICU need further study before implementation but appear to offer physicians a deeper understanding of their patients and a better opportunity to improve patient outcome than presently used models based on regression.

Thesis (S.M.)--Massachusetts Institute of Technology, Sloan School of Management, Operations Research Center, 1999. / Includes bibliographical references (leaves 103-106). / In this thesis, we develop practical solution methods for a supply chain optimization problem: a multi-echelon, un capacitated, time-expanded network of distribution cen­ters and stores, for which we seek the shipping schedule that minimizes total inventory, backlogging, and shipping costs, assuming deterministic, time-varying demand over a fixed time horizon for a single product. Because of fixed ordering and shipping costs, this concave cost network flow problem is in a class of NP-hard network design problems. We develop mathematical programming formulations, heuristic algorithms, and enhanced algorithms using approximate dynamic programming (ADP). We achieve a strong mixed integer programming (MIP) formulation, and fast, reliable algorithms, which can be extended to problems with multiple products. Beginning with a lot-size based formulation, we strengthen the formulation in steps to develop one which is a variation of a node-arc formulation for the network design problem. In addition, we present a path-flow formulation for the single product case and an enhanced network design formulation for the multiple product case. The basic algorithm we develop uses a dynamic lot-size model with backlogging together with a greedy procedure that emulates inventory pull systems. Four re­lated algorithms perform local searches of the basic algorithm's solution or explore alternative solutions using pricing schemes, including a Lagrangian-based heuristic. We show how approximate dynamic programming can be used to solve this sup­ply chain optimization problem as a dynamic control problem using any of the five algorithms. In addition to improving all the algorithms, the ADP enhancement turns the simplest algorithm into one comparable to the more complex ones. Our computational results illustrate that our enhanced network design formula­tion almost always produces integral solutions and can be used to solve problems of moderate size (3 distribution centers, 30 stores, 30 periods). Our heuristic methods, particularly those enhanced by ADP methods, produce near optimal solutions for truly large scale problems. / by Carl E. Wike. / S.M.

Thesis (Ph. D.)--Massachusetts Institute of Technology, Sloan School of Management, Operations Research Center, 1999. / Submitted to the Dept. of Electrical Engineering and Computer Science, MIT. / Includes bibliographical references (p. 110-114). / In the automobile manufacturing industry, vehicle production is an assembly-line process. Automobile companies typically sequence vehicle production based on workload balancing factors, with little consideration of vehicle colors. The resulting sequence usually has a small average paint block size. Because automobile manufacturers use expensive and sometimes pollutant chemicals to clean out old paint at each color change, they would like to increase the size of the paint blocks, while maintaining the original workload-balanced vehicle sequence. To achieve this goal, Ford and other automobile manufacturers are considering automated vehicle storage and retrieval systems that would allow them to perturb the original sequence around the vehicle painting station, creating larger paint blocks and then restoring the original sequence after painting. To evaluate these systems, it is necessary to develop a method for re sequencing cars and for calculating the resulting savings in paint cleanings. The problem of resequencing can be cast as a traveling salesman problem with time windows. For a real-life sequence size of 750 cars and windows of 75 slots per car in either direction, direct modeling using the strongest known ... formulation yields an integer program with up to 200,000 constraints and 14,000,000 variables. Reduced formulations. We exploit special problem structure to solve the LP relaxation of this problem quickly using Lagrangean relaxation. We prove and use an order-within-color property to construct an enumerative formulation, and use a greedy approach to bound the LP optimum. We decompose the problem and solve smaller enumerative formulations sequentially to generate a heuristic solution that empirically is within 2.5% of optimality. Because our heuristic and bounding procedure runs in a total of one minute, an automobile manufacturer could use the process to adjust the resequencing process in real time to compensate for vehicles that have been delayed in the original sequence due to production defects or other disruptions. We also establish worst-case bounds ranging from 2.5 to 6 for another related heuristic. / by Joel Scott Sokol. / Ph.D.