Games have been used to evaluate and advance techniques in the eld of Articial Intelligence since before computers were invented. Many of these games have been deterministic perfect information games (e.g. Chess and Checkers). A deterministic game has no chance element and in a perfect information game, all information is visible to all players. However, many real-world scenarios involving competing agents can be more accurately modeled as stochastic (non-deterministic), im- perfect information games, and this dissertation investigates such games. Poker is one such game played by millions of people around the world; it will be used as the testbed of the research presented in this dissertation. For a specic set of games, two-player zero-sum perfect recall games, a recent technique called Counterfactual Regret Minimization (CFR) computes strategies that are provably convergent to an -Nash equilibrium. A Nash equilibrium strategy is very useful in two-player games as it maximizes its utility against a worst-case opponent. However, once we move to multiplayer games, we lose all theoretical guarantees for CFR. Furthermore, we have no theoretical guarantees about the performance of a strategy from a multiplayer Nash equilibrium against two arbitrary op- ponents. Despite the lack of theoretical guarantees, my thesis is that CFR-generated agents may perform well in multiplayer games. I created several 3-player limit Texas Holdem Poker agents and the results of the 2009 Computer Poker Competition demonstrate that these are the strongest 3-player computer Poker agents in the world. I also contend that a good strategy can be obtained by grafting a set of two-player subgame strategies to a 3-player base strategy when one of the players is eliminated.

This dissertation develops and analyzes differential equation-based mathematical models and efficient numerical methods and algorithms for genetic regulatory network identification. The primary objectives of the dissertation are to design, analyze, and test a general variational framework and numerical methods for seeking its approximate solutions for reverse engineering genetic regulatory networks from microarray datasets using the approach based on differential equation modeling. In the proposed variational framework, no structure assumption on the genetic network is presumed, instead, the network is solely determined by the microarray profile of the network components and is identified through a well chosen variational principle which minimizes a biological energy functional. The variational principle serves not only as a selection criterion to pick up the right biological solution of the underlying differential equation model but also provide an effective mathematical characterization of the small-world property of genetic regulatory networks which has been observed in lab experiments. Five specific models within the variational framework and efficient numerical methods and algorithms for computing their solutions are proposed and analyzed in the dissertation. Model validations using both synthetic network datasets and real world subnetwork datasets of Saccharomyces cerevisiae (yeast) and E. Coli are done on all five proposed variational models and a performance comparison vs some existing genetic regulatory network identification methods is also provided. As microarray data is typically noisy, in order to take into account the noise effect in the mathematical models, we propose a new approach based on stochastic differential equation modeling and generalize the deterministic variational framework to a stochastic variational framework which relies on stochastic optimization. Numerical algorithms are also proposed for computing solutions of the stochastic variational models. To address the important issue of post-processing computed networks to reflect the small-world property of underlying genetic regulatory networks, a novel threshholding technique based on the Random Matrix Theory is proposed and tested on various synthetic network datasets.

This thesis investigates the computation of piecewise polynomial solutions to ill- conditioned linear systems of equations when noise on the linear measurements is observed. Specifically, we enhance our understanding of and provide qualifications on when such ill-conditioned systems of equations can be solved to a satisfactory accuracy. We show that the conventional condition number of the coefficient matrix is not sufficiently informative in this regard and propose a more relevant conditioning measure that takes into account the decay rate of singular values. We also discuss interactions of several factors affecting the solvability of such systems, including the number of discontinuities in solutions, as well as the distribution of nonzero entries in sparse matrices. In addition, we construct and test an approach for computing piecewise polynomial solutions of highly ill-conditioned linear systems using a randomized, SVD-based truncation, and L1-norm regularization. The randomized truncation is a stabilization technique that helps reduce the cost of the traditional SVD truncation for large and severely ill-conditioned matrices. For L1-minimization, we apply a solver based on the Alternating Direction Method. Numerical results are presented to compare our approach that is faster and can solve larger problems, called RTL1 (randomized truncation L1-minimization), with a well-known solver PP-TSVD.

We study a two-way amplify and forward relay network with two transceivers which communicate through a network of nr relays while there is no direct link between the two transceivers. Each relay is equipped with a single antenna for transmitting and receiving. We study the minimization of the total transmit power that is used in all of the network nodes given the condition that the transceiver which calculates the optimal transmitting power has a full knowledge about the channels between itself and the relays and the variance with zero mean of the channels between the relays and the other transceiver. The total average power is minimized subject to a soft constraint which guarantees that the outage probability is below a certain level. The optimal solution is derived in closed form and leads to a single relay selection criterion. / UOIT

The current thesis presents an exploratory study on municipal solid waste generation and minimization in Sweden, with a focus on their connection to basic socio-economic parameters. The fundamental goal of the study is to investigate into correlations and interdependencies between waste generation, waste minimization and basic socio-economic characteristics on municipal level, and to search for models for explanation of waste management parameters through socio-economic factors. Theoretical background involves reasoning on the role of municipal waste management in sustainable development, and extensive analysis of framework, legislation and organization of municipal solid waste management in Sweden. Practical part presents correlation analysis of data, which proved that socio-economic parameters do not explain differences in waste management performance of Swedish municipalities; however they are closely connected to differences between municipalities in aspect of presence of waste-related data.

This dissertation presents dynamic reactive power control of isolated power systems. Isolated systems include MicroGrids in islanded mode, shipboard power systems operating offshore, or any other power system operating in islanded mode intentionally or due to a fault. Isolated power systems experience fast transients due to lack of an infinite bus capable of dictating the voltage and frequency reference. This dissertation only focuses on reactive control of islanded MicroGrids and AC/DC shipboard power systems. The problem is tackled using a Model Predictive Control (MPC) method, which uses a simplified model of the system to predict the voltage behavior of the system in future. The MPC method minimizes the voltage deviation of the predicted bus voltage; therefore, it is inherently robust and stable. In other words, this method can easily predict the behavior of the system and take necessary control actions to avoid instability. Further, this method is capable of reaching a smooth voltage profile and rejecting possible disturbances in the system. The studied MicroGrids in this dissertation integrate intermittent distributed energy resources such as wind and solar generators. These non-dispatchable sources add to the uncertainty of the system and make voltage and reactive control more challenging. The model predictive controller uses the capability of these sources and coordinates them dynamically to achieve the voltage goals of the controller. The MPC controller is implemented online in a closed control loop, which means it is self-correcting with the feedback it receives from the system.

In this master thesis a visual odometry system is implemented and explained. Visual odometry is a technique, which could be used on autonomous vehicles to determine its current position and is preferably used indoors when GPS is notworking. The only input to the system are the images from a stereo camera and the output is the current location given in relative position. In the C++ implementation, image features are found and matched between the stereo images and the previous stereo pair, which gives a range of 150-250 verified feature matchings. The image coordinates are triangulated into a 3D-point cloud. The distance between two subsequent point clouds is minimized with respect to rigid transformations, which gives the motion described with six parameters, three for the translation and three for the rotation. Noise in the image coordinates gives reconstruction errors which makes the motion estimation very sensitive. The results from six experiments show that the weakness of the system is the ability to distinguish rotations from translations. However, if the system has additional knowledge of how it is moving, the minimization can be done with only three parameters and the system can estimate its position with less than 5 % error.

The effectiveness of partial ozonation of return activated sludge for enhancing denitrification and waste sludge minimization were examined. A pair of nitrifying sequencing batch reactors was operated in either aerobic or alternating anoxic/aerobic conditions, with one control and one ozonated reactor in each set. The amount of solids decreased with the ozone dose. Biomass in the anoxic/aerobic reactor was easier to destroy than in the aerobic one, generating approximately twice as much soluble chemical oxygen demand (COD) by cell lysis. Increased COD favoured production of extracellular polymers in ozonated reactors, enhancing flocculation and improving settling. Floc stability was also strengthened in prolonged operation in alternating treatment, resulting in declined solids destruction. Dewaterability was better in alternating reactors than in aerobic ones indicating that incorporation of an anoxic zone for biological nutrient removal leads to improvement in sludge dewatering. The negative impact of ozonation on dewaterability was minimal in terms of the long-term operation. Ozone successively destroyed indicator estrogenic compounds, contributing to total estrogen removal from wastewater. Denitrification rate improved up to 60% due to additional carbon released by ozonation. Nitrification rates deteriorated much more in the aerobic than in the alternating reactor, possibly as a result of competition created by growth of heterotrophs receiving the additional COD. Overall, ozonation provided the expected benefits and had less negative impacts on processes in the alternating treatment, although after prolonged operation, benefits could become less significant. The alternating anoxic/aerobic reactor achieved twice the nitrification rates of its aerobic counterpart. Higher removal rates of estrogens were associated with higher nitrification rates, supporting the contention that the nitrifying biomass was responsible for their removal. The alternating treatment offered the better estrogen biodegradation. Microbial populations in both reactors were examined with fluorescent in situ hybridization. Dominance of rapid nitrifiers like Nitrosomonas and Nitrobacter (79.5%) in the alternating reactor, compared to a dominance of slower nitrifiers like Nitrosospira and Nitrospira (78.2%) in the aerobic reactor were found. The findings are important to design engineers, as reactors are typically designed based on nitrifiers’ growth rate determined in strictly aerobic conditions. / October 2008

The objective of the proposed research is to develop a new algorithm for range and Doppler ambiguity resolution in radar detection data using L1 minimization methods for sparse signals and to investigate the properties of such techniques. This novel approach to ambiguity resolution makes use of the sparse measurement structure of the post-detection data in multiple pulse repetition frequency radars and the resulting equivalence of the computationally intractable L0 minimization and the surrogate L1 minimization methods. The ambiguity resolution problem is cast as a linear system of equations which is then solved for the unique sparse solution in the absence of errors. It is shown that the new technique successfully resolves range and Doppler ambiguities and the recovery is exact in the ideal case of no errors in the system. The behavior of the technique is then investigated in the presence of real world data errors encountered in radar measurement and detection process. Examples of such errors include blind zone effects, collisions, false alarms and missed detections. It is shown that the mathematical model consisting of a linear system of equations developed for the ideal case can be adjusted to account for data errors. Empirical results show that the L1 minimization approach also works well in the presence of errors with minor extensions to the algorithm. Several examples are presented to demonstrate the successful implementation of the new technique for range and Doppler ambiguity resolution in pulse Doppler radars.

Forced air cooling of electronic packages is enhanced through the use of extended surfaces or heat sinks that reduce boundary resistance allowing heat generating devices to operate at lower temperatures, thereby improving reliability. Unfortunately, the clearance zones or bypass regions surrounding the heat sink, channel some of the cooling air mass away from the heat sink, making it difficult to accurately estimate thermal performance. The design of an "optimized" heat sink requires a complete knowledge of all thermal resistances between the heat source and the ambient air, therefore, it is imperative that the boundary resistance is properly characterized, since it is typically the controlling resistance in the path. Existing models are difficult to incorporate into optimization routines because they do not provide a means of predicting flow bypass based on information at hand, such as heat sink geometry or approach velocity. <br /><br /> A procedure is presented that allows the simultaneous optimization of heat sink design parameters based on a minimization of the entropy generation associated with thermal resistance and fluid pressure drop. All relevant design parameters such as geometric parameters of a heat sink, source and bypass configurations, heat dissipation, material properties and flow conditions can be simultaneously optimized to characterize a heat sink that minimizes entropy generation and in turn results in a minimum operating temperature of an electronic component. <br /><br /> An analytical model for predicting air flow and pressure drop across the heat sink is developed by applying conservation of mass and momentum over the bypass regions and in the flow channels established between the fins of the heat sink. The model is applicable for the entire laminar flow range and any type of bypass (side, top or side and top both) or fully shrouded configurations. During the development of the model, the flow was assumed to be steady, laminar, developing flow. The model is also correlated to a simple equation within 8% confidence level for an easy implementation into the entropy generation minimization procedure. The influence of all the resistances to heat transfer associated with a heat sink are studied, and an order of magnitude analysis is carried out to include only the influential resistances in the thermal resistance model. Spreading and material resistances due to the geometry of the base plate, conduction and convection resistances associated with the fins of the heat sink and convection resistance of the wetted surfaces of the base plate are considered for the development of a thermal resistance model. The thermal resistance and pressure drop model are shown to be in good agreement with the experimental data over a wide range of flow conditions, heat sink geometries, bypass configurations and power levels, typical of many applications found in microelectronics and related fields. Data published in the open literature are also used to show the flexibility of the models to simulate a variety of applications. <br /><br /> The proposed thermal resistance and pressure drop model are successfully used in the entropy generation minimization procedure to design a heat sink with bypass for optimum dimensions and performance. A sensitivity analysis is also carried out to check the influence of bypass configurations, power levels, heat sink materials and the coverage ratio on the optimum dimensions and performance of a heat sink and it is found that any change in these parameters results in a change in the optimized heat sink dimensions and flow conditions associated with the application for optimal heat sink performance.