Proefschrift Universiteit van Amsterdam. / Met index, lit. opg. - Met samenvatting in het Nederlands.

Thesis (Ph. D.)--University of Massachusetts Amherst, 2009. / Includes bibliographical references (p. 227-236). Print copy also available.

Model Predictive Control (MPC) is an application of control that is highly popular due to its sensible approach and its ease of implementation. These qualities give MPC an advantage over Linear Quadratic (LQ) control, even though LQ will result in the optimal result where feasible. Recent advancements have resulted in greater computational power, which has given rise to the development of more complicated MPC algorithms, but there are instances when the complexity of the calculations involved will result in the amount of computations involved or ill-conditioning of the problem.

Configuring typical devices in the telemetry community requires the creation of complex, device-specific configuration files. While the grammar of the configuration files is vendor neutral, the device specific details are vendor specific. Thus, a naïve approach to building these files is to construct a file, test it against a device, and then iterate. The specification sheets (and other documents) for the device can serve as a guide, but the details of flight test configuration possibilities are immense and, in this community, typically not fully documented. This paper describes a process of creating a set of general rules describing characteristics of a configuration file and using those rules to discover the configuration constraints of telemetry devices automatically. The discovered constraints posed by a particular vendor’s device can then be quickly formed into a correct-by-construction constraint-based grammar for use in other systems.

Nonstandard cosmological models of two broad classes are examined: those in which there are more than four spacetime dimensions, and those in which there is a variable cosmological “constant” Λ. We test claims that a number of higher-dimensional models give rise to inflation. New constraints are placed on such models, and a number of them are ruled out. We then investigate the potential of variable-Λ theories to address the problem of the initial singularity. We consider a number of different phenomenological representations for this parameter, assessing their implications for the evolution of the cosmological scale factor as well as a range of observational data. In several cases we find nonsingular models which are compatible with observation. / Graduate

No description available.

I am interested in creation, the root of what architects do whether they are designing a master plan or a detail. The interesting thing about architecture is that an architect's creation does not exist in a vacuum. Numerous forces operate on the vision of an architect before it can become a reality; the most important being the building's context. Architecture without a site is conjecture and speculation and though it may have its place in some realm of study, it can never be real. For an idea to be manifested, it must be built someplace on this Earth and therefore I believe that architecture is a response to the context of the site. The site informs the architect on the material of the building's construction, how the building is oriented with respect to the sun, how and when the sun is allowed to enter the building, and how the building is approached and therefore entered. The context pushes and pulls on the building, informing the design process and demanding responses. The city offers a context dense in built structures and rich in history and architectural precedence. These parameters should be ever present in the mind of the architect during the design. A good design should respect and enhance the urban situation in which it resides as well as provide a useful scaffold for the architect to build from. By limiting the architect and forcing the integration of the new building with the existing context, the urban condition spurs ingenuity and can actually simplify the endless possibilities of architectural space. A rural site offers a different set of parameters under which the architect must operate. The shape of the land commands the largest influence on a rural design, should the architect look for contextual clues. The rural site can also set the architect free and allow a building to be a pure manifestation of the architect's mind, leaving the building program as the only constraint. The intent of the thesis is to investigate how architects design. How does a building come into existence? By designing two buildings at opposite ends of a contextual spectrum I hope to raise questions in my own mind about how context influences the decisions that are made in the design process. Recognizing these different sets of parameters can lead to a better understanding of context as a guiding force that shapes architecture. / Master of Architecture

Area exploration and information gathering are one of the fundamental problems in mobile robotics. Much of the current research in swarm robotics is aimed at developing practical solutions to this problem. Exploring large environments poses three main challenges. Firstly, there is the problem of limited connectivity among the robots. Secondly, each of the robots has a limited battery life which requires the robots to be recharged each time they are running out of charge. Lastly, the robots have limited memory to store data. In this work, we mainly focus on the memory and energy constraints of the robot swarm. The memory constraint forces the robots to travel to a centralized data collection center called sink, to deposit data each time their memory is full. The energy constraint forces the robots to travel to the charging station called dock to recharge when their battery level is low. However, this navigation plan is inefficient in terms of energy and time. There is additional energy dissipation in depositing data at the centralized sink. Moreover, ample amount of time is spent in traveling from one end of the arena to the sink owing to the memory constraint. The goal is that the robots perform data gathering in the least time possible with the optimal use of energy. Both the energy and time spent while depositing data at the sink act as an additional overhead cost to this goal. In this work, we propose to study an algorithm to tackle this scenario in a decentralized manner. We implement a dynamic task allocation algorithm which accomplishes the goal of exploration with data gathering by assigning roles to robots based on their memory buffer and energy levels. The algorithm assigns two sets of roles, to the entire group of robots, namely: Role A is the data gatherer, a robot which does the task of workspace exploration and data gathering, and Role B is data relayer, a robot which does the task of data transportation from data gatherers to the sink. By this division of labor, the robots dynamically decide which role to choose given the contradicting goals of maximizing data gathering and minimizing energy loss. The choice of a robot to perform the task of data gathering or data relaying is the key problem tackled in this work. We study the performance of the algorithm in terms of task distribution, time spent by the robots on each task and data throughput. We analyze the behavior of the robot swarm by varying the energy constraints, timeout parameter as well as strategies for relayer choice. We also test whether the algorithm is scalable.

We consider joint chance-constrained programs with random lefthand sides. The motivation of this project is that this class of problem has many important applications, but there are few existing solution methods. For the most part, we deal with the subclass of problems for which the underlying parameter distributions are discrete. This assumption allows the original problem to be formulated as a deterministic equivalent mixed-integer program. We rst approach the problem as a mixed-integer program and derive a class of optimality cuts based on irreducibly infeasible subsets of the constraints of the scenarios of the problem. The IIS cuts can be computed effciently by means of a linear program. We give a method for improving the upper bound of the problem when no IIS cut can be identifi ed. We also give an implementation of an algorithm incorporating these ideas and finish with some computational results. We present a tabu search metaheuristic for fi nding good feasible solutions to the mixed-integer formulation of the problem. Our heuristic works by de ning a sufficient set of scenarios with the characteristic that all other scenarios do not have to be considered when generating upper bounds. We then use tabu search on the one-opt neighborhood of the problem. We give computational results that show our metaheuristic outperforming the state-of-the-art industrial solvers. We then show how to reformulate the problem so that the chance-constraints are monotonic functions. We then derive a convergent global branch-and-bound algorithm using the principles of monotonic optimization. We give a finitely convergent modi cation of the algorithm. Finally, we give a discussion on why this algorithm is computationally ine ffective. The last section of this dissertation details an application of joint chance-constrained stochastic programs to a vaccination allocation problem. We show why it is necessary to formulate the problem with random parameters and also why chance-constraints are a good framework for de fining an optimal policy. We give an example of the problem formulated as a chance constraint and a short numerical example to illustrate the concepts.

The simulation as a stand alone optimization tool of a complex system such as a vertical integrated mining operation, significantly over simplifies the actual picture of the system processes involved resulting in an unaccountable effort and resources being spent on optimizing Non Value Added (NA) processes. This study purposed to develop a discrete stochastic simulation-optimization model to accurately capture the dynamics of the system and to provide a structured way to optimize the Value Added (VA) processes. The mine operation model to be simulated for this study is designed as a hybrid level throughput model to identify the VA processes in a mining operation. This study also allows a better understanding of the impact of variation on the likelihood of achieving any given overall result. The proposed discrete stochastic simulation- optimization model provides the ability for a process manager to gain realistic understanding of what a process can do if some factors constraining the process were to be optimized i.e. to conduct what-if analysis. Another benefit of this approached technique is to be able to estimate dependable and reasonable returns on a large optimization related expenditure. The inputs into the model are the capability of the processes which are entered using various variables depending on how much information is available; simple inputs for least amount of information to detailed inputs for well known process to combinational inputs for somewhere in between. The process bottlenecks are identified and measured using the outputs of the model which include production output, severity of constraints, capacity constraints and cumulative bottleneck plots. Once a base case has been identified and documented then the inputs can be modified to represent the business initiatives and the outputs can be compared to the base case to evaluate the true value of the initiative.