Typescript (photocopy). / Digitized by Kansas Correctional Industries

A dissertation submitted in fulfillment of the requirements of the degree of Master of Science in the School of Computer Science and Applied Mathematics March 22, 2017 / Agent-based models, particularly those applied to financial markets, demonstrate the ability to produce realistic, simulated system dynamics, comparable to those observed in empirical investigations. Despite this, they remain fairly difficult to calibrate due to their tendency to be computationally expensive, even with recent advances in technology. For this reason, financial agent-based models are frequently validated by demonstrating an ability to reproduce well-known log return time series and central limit order book stylized facts, as opposed to being rigorously calibrated to transaction data. We thus apply an established financial agent-based model calibration framework to a number of intraday agent-based models employing realistic order matching procedures and demonstrate that while the parameters of these models rooted in market microstructure can indeed be meaningfully calibrated, those exclusively related to agent behaviors and incentives remain problematic, due to the presence of parameter degeneracies not identified by stylized fact-centric validation. We further argue that the observed parameter degeneracies are likely a consequence of the realistic matching processes employed in these models, which suggests that alternative approaches to linking data, phenomenology and market structure may be necessary and that the stylized fact-centric validation of intraday agent-based models is insufficient. / MT 2017

CMOS active pixel sensors (APS) have shown competitive performance with charge-coupled device (CCD) and offer many advantages in cost, system power reduction and on-chip integration of VLSI electronics. Among CMOS image sensors, sensors with logarithmic pixels are particularly applicable for outdoor environment where the light intensity varies over a wide range. They are also randomly accessible in both time and space. A major drawback comes from process variations during fabrication. This gives rise to a considerable fixed pattern noise (FPN) which deteriorates the image quality. In this thesis, a technique that greatly reduces FPN using on-chip calibration is introduced. An image sensor that consists of 64x64 active pixels has been designed, fabricated and tested. Pixel pitch is 18um x 19.2um? and is fabricated in a 0.5-um? CMOS process. The proposed pixel circuit considerably reduces the FPN as predicted in theoretical analysis. The measured FPN value is 2.29% of output voltage swing and column-wise FPN is 1.49% of mean output voltage over each column.

In this master’s thesis a model of a digitally compensated N-bit C-xC sar adc was developed.The architecture uses charge redistribution in a C-xC capacitor network to performthe conversion. Focus in the master’s thesis was set to understand how the charge is redistributedin the network during the conversion and calibration phase. Redundancy andparasitic capacitors is present in the system and rises the need for extra conversion steps aswell as a calibration algorithm. The calibration algorithm, Bit Weight Estimation, calculatesa weight corresponding to each bit which is used in the last conversion step to perform adigital weighting. The result of extensive calculations in different C-xC capacitor networkswas a model in Python of an N-bit C-xC sar adc. That model was used to create a model ofan eight-bit C-xC sar adc and finding suitable parameters for it through calculations andsimulations. The parameters giving the best inl was chosen. With the best parameters theC-xC sar adc static and dynamic performance was tested and showed an inl of less than1lsb, snr of 47:8 dB and enob of 7:6 bits.

none

Multibeam sonar systems are much more efficient than the convectional single-beam echo sounders for seafloor-mapping in hydrographic surveying. On the other hand, the operation of multibeam sonar systems needs to integrate more auxiliary sensor units. Because the world coordinates of each footprint is calculated based on the geometry of the sonar head relative to the GPS of the ship. Therefore, the resulting survey quality highly depends on the accuracy of the estimated mounting configuration of the sonar head, and other sensor units. Basically, the configuration parameters include the three Euler's angles, three linear translations and the asynchronous latency of signals between the transducer and other sensors. These parameters can not be measured directly. They can only be estimated from the post-process of the bathymetry data called patch test. Generally, the patch test requires the survey ship to follow several designated paths which are parallel, reciprocal or perpendicular to each other. Furthermore, the choice of seabed slope is also an important factor for the quality of the result. The contour plots of the seabed for the different paths are used to estimate the mounting configuration of the sonar head. In this work, we propose best-fitting a small flat patch to represent the seabed right beneath a segment of the path. A pair of patches from the two adjacent segments of reciprocal or perpendicular paths are selected for comparison. The difference between the two patches gives us an idea how the mounting parameters, i.e. the rolling, pitching and yawing angles, might be. If the parameters are accurately estimated, the two patches should be co-plane. We design several semi-positive definite functions and feed back control algorithms to steer the mounting angles to search for the solutions. One more advantage of this approach is that the variation of each mounting angles as the survey undergoes can be monitored. We test this idea with simulated seabed data, and field data as well. The result is within 1\% difference compared with respect to the angles calibrated by commercial software, such as Hypack.

CMOS active pixel sensors (APS) have shown competitive performance with charge-coupled device (CCD) and offer many advantages in cost, system power reduction and on-chip integration of VLSI electronics. Among CMOS image sensors, sensors with logarithmic pixels are particularly applicable for outdoor environment where the light intensity varies over a wide range. They are also randomly accessible in both time and space. A major drawback comes from process variations during fabrication. This gives rise to a considerable fixed pattern noise (FPN) which deteriorates the image quality. In this thesis, a technique that greatly reduces FPN using on-chip calibration is introduced. An image sensor that consists of 64x64 active pixels has been designed, fabricated and tested. Pixel pitch is 18um x 19.2um? and is fabricated in a 0.5-um? CMOS process. The proposed pixel circuit considerably reduces the FPN as predicted in theoretical analysis. The measured FPN value is 2.29% of output voltage swing and column-wise FPN is 1.49% of mean output voltage over each column.

Thesis (M.S.)--Ohio State University, 2002. / Includes bibliographical references (leaves 60-62).

A prototype of the Muon Telescope Detector (MTD) was installed at STAR (Solenoidal Tracker at RHIC) during run year 2007. While the cosmic and beam tests showed a ~ 60 - 70 ps timing resolution for MTD, the actual performance in Au + Au collisions at STAR was found ~300 ps. In run year 10 STAR implemented a new electronics system for MTD and a cosmic ray trigger to study the performance of its several subsystems. With the cosmic ray data, this study shows that the timing resolution of MTD can reach 99 ps after a full calibration. / text

Fabric defects inspection is a vital step for fabric quality assessment. Many vision-based automatic fabric defect detection methods have been proposed to detect fabric flaws efficiently and accurately. Because the inspection methods are vision-based, image quality is of great importance to the accuracy of detection result. To our knowledge, most of camera lenses have radial distortion. So our goal in this project is to remove the radial distortion and achieve undistorted images. Much research work has been done for 2-D image correction, but the study for 1-D line scan camera image correction is rarely done, although line scan cameras are gaining more and wider applications due to the high resolution and efficiency on 1-D data processing. A novel line scan camera correction method is proposed in this project. We first propose a pattern object with mutually parallel lines and oblique lines to each pair of parallel ones. The purpose of the pattern design is based upon the fact that line scan camera acquires image one line at a time and it's difficult for one scan line to match the "0-D" marked points on pattern. We detect the intersection points between pattern lines and one scan line and calculate their position according to the pattern geometry. As calibrations for 2-D cameras have been greatly achieved, we propose a method to calibrate 1-D camera. A least-square method is applied to solve the pinhole projection equation and estimate the values of camera parameter matrix. Finally we refine the data with maximum-likelihood estimation and get the camera lens distortion coefficients. We re-project the data from the image coordinate to the world coordinate, using the obtained camera matrix and the re-projection error is 0.68 pixel. With the distortion coefficients ready, we correct captured images with an undistortion equation. We introduce a term of unit distance in the discussion part to better assess the proposed method. When testifying the undistortion results, we observe corrected image has almost identical unit distance with standard deviation of 0.29 pixels. Compared to the ideal distortion-free unit distance, the corrected image has only 0.09 pixel off the average, which proves the validity of the proposed method. / text