Dynamic memory allocators are a determining factor of an application's performanceand have the opportunity to improve a major performance bottleneck ontoday's computer hardware: data locality. To approach this problem, a memoryallocator must rst oer strategies that allow the locality problem to be addressed.However, while focusing on locality, an allocator must also not ignore the existing constraintsof allocation speed and fragmentation, which further complicate its design. Inorder for a locality improving technique to be successfully employed in today's largecode applications, its integration needs to be automatic, without user intervention.The alternative, manual integration, is not a tractable solution.In this dissertation we develop three novel memory allocators that explore dierentallocation strategies that enhance an application's locality. We conduct the rststudy that shows that allocation speed, fragmentation and locality improving goalsare antagonistic. We develop an automatic method that supplies allocation hintsfrom C++ STL containers to their allocators. This method allows applications tobenet from locality improving techniques at the cost of a simple re-compilation. Weconduct the rst study that quanties the eect of allocation hints on performance,and show that an allocator with high locality of reference can be as competitive asone using an application's spatial feedback.To further allow dynamic memory allocation to improve an application's performance,new and non-traditional strategies need be explored. We develop a generic software tool that allows users to examine unconventional strategies. The tool allowsusers not only to focus on allocation strategies rather than their implementation, butalso to compare and contrast various approaches.

Because of IC technology evolution and the increase of market demand, the communication industry grows vigorously in recent years. The voltage-controlled oscillator plays a key role in the RF transceiver and provides oscillation signals needed for upconversin and downconvertion. Usually, we separate the signals into I/Q channels for modulation and demodulation in upconversin and downconvertion. Because the quality of the local oscillator influences the performance of communication system, designing a voltage-controlled oscillator that can provide two identical signals in accurate quadrature is necessary. In this thesis, a new quadrature voltage-controlled oscillator is presented. We use two identical relaxation oscillators with adjustable Schmitt triggers to construct the cross-coupled architecture. This oscillator has accurate ( <1¢X) and stable quadrature outputs which are independent of operating frequency and process variations. This oscillator circuit is fabricated in TSMC 0.35£gm CMOS Mixed-Signal process provided by National Chip Implementation Center (CIC). Our design is verified by simulation and measurement results.

In this thesis, two types of geared linkages are studied in detail. One of the mechanisms is the geared five link mechanism for which the input and output shafts are collinear. The other is a two degree-of-freedom nine-link mechanism which is called the &quot / geared adjustable stroke mechanism&quot / . The geared adjustable stroke mechanism uses the geared five link mechanism in its structure. The geared adjustable stroke mechanism has been used in practice, but neither of these mechanisms have been studied in the literature. Analysis procedures are developed and expressions for the transmission angles of the mechanisms are derived. A synthesis procedure is proposed and charts are prepared for the design of such mechanisms.

none

In order to boost the exit value, it is not uncommon that issuers report earnings in excess of cash flow generated by its operations at the initial public offering (IPO). The discretionary activity of performing earnings management can mislead investors about the intrinsic value of the newly public firm. Within this study, I examine how earnings management will affect the stock market reaction upon the lockup expiration date, the IPO adjustable offering size, and how the backing of private equity or venture capital (PEVC) affects earnings management tendencies within IPO firms. Using a unique, hand-collected dataset of 56 Swedish newly public firms from 2007 - 2014, I show that IPO firms (i) manage their earnings at the full fiscal year prior to the IPO and that earnings management will result in a negative stock market reaction upon the lockup expiration date. More importantly, I show that (ii) high adjustable offerings do not affect this relationship indicating that earnings management has no impact on the adjustable part of the offering size within IPOs. I also find that (iii) IPO firms backed by PEVC firms are more eager to manipulate their earnings, and (iv) highly reputable PEVC firms do not mitigate the manipulation of earnings within IPO firms. The results taken together suggest that studying the stock market reaction on the lockup expiration date is important for manipulative IPO firm detection, and that a participation in IPOs backed by PEVC firms must be done with caution.

For pediatric amputees in developing countries, where characteristically rapid growth of children is compounded by a lack of medical services, maintaining proper socket fit is a challenging but necessary endeavor. A socket design that adjusts for radial and longitudinal growth will allow patients to wear the same socket for a longer period of time saving them the expense of subsequent fittings and hardware. Manufacture of such a socket poses a challenge for contemporary manufacturing processes. Due to its ability to rapidly manufacture complex part geometries, Selective Laser Sintering (SLS) is particularly suited to this application. Several preliminary design concepts for a volume adjustable transtibial SLS prosthetic socket for pediatric amputees in developing countries have been generated. These current design concepts utilize fasteners such as ratchet hooks and threads. Results from design and validation of theoretical models of these fastener concepts are the focus of this thesis. / text

Modern electronics continues to push past boundaries of integration and functional density toward elusive, completely autonomous, self-powered microsystems. As systems continue to shrink, however, less energy is available on board, leading to short device lifetimes (run-time or battery life). Extended battery life is particularly advantageous in the systems with limited accessibility, such as biomedical implants and structure-embedded micro-sensors. The power management process usually requires compact and efficient power converters to be embedded in these microsystems. This dissertation introduces switched-capacitor (SC) power converter designs that make all these techniques realizable on silicon.Four different integrated SC power converters with multiple control schemes are designed here to provide low-power high-efficient power sources. First, a monolithic step-down power converter with subthreshold z-domain digital pulse-width modulation (DPWM) controller is proposed for ultra-low power microsystems. The subthreshold design significantly reduces the power dissipation in the controller. Second, an efficient monolithic master-slave complementary power converter with a feedback controller that purely operates in subthreshold operation region is discussed to tailor for the aforementioned ultra-low power applications. Third, we introduce an efficient monolithic step-down SC power stage with multiple-gain control and on-chip capacitor sizing for self-powered microsystems. The multiple-gain control helps the converter to constantly maintain high efficiency over a large input/output range. The size-adjustable pumping capacitors allow the output voltage to be regulated at different desired levels, with a constant 50% duty ratio. The monolithic implementations in these three integrated CMOS power converters effectively suppress noise and glitches caused by parasitic components due to bonding, packaging and PCB wiring. Fourth, an efficient step-up and step-down SC power converter with multiple-gain closed-loop controller is presented. The measurements and simulation results in these four power converters demonstrate the techniques proposed in this research. The approaches presented in this dissertation are evidently viable for realizing compact and high efficient SC power converters, contributing to next generation power-efficient microsystems designs.

Cast-in-place concrete in composite with steel sections is commonly used in bridge deck constructions. The shear transfer between the concrete and steel section is achieved by shear connectors and the strength calculation of conventional shear connectors, i.e. shear studs, is provided in various design codes in North America. Due to the fact that the strength equation is largely based on experimental results, the applicability of the equation is only warranted where the design matches the experimental configuration of the test specimens. Thus, the codes specify detailing requirement for the stud height and the elevation of the reinforcement mesh in relation to the stud height. However, these requirements, in particular, the elevation of the reinforcement mesh, may be difficult to meet accurately in construction practice. The implications of not meeting the mesh requirement to the strength of the shear stud and the remedy solutions are examined in this study. An experimental program involving the test of thirty-three push-out specimens was designed and conducted with a focus on the shear studs' performance. Testing parameters included reinforcement mesh position, shear stud height, presence of stud head, shear stud spacing, and steel flange surface treatment. In addition, the performance of a new type of shear studs, referred to as adjustable studs, was also studied experimentally. The ultimate load and load vs. slip curves were presented and discussed in the forms of tables and graphs. The failure modes were noted and the relationship between the failure modes and the ultimate capacity was discussed. Ultimate loads obtained from specimens were then used to assess the efficacy of code suggested values. Results showed that depending on the elevation of reinforcement mesh, three failure modes were observed including concrete related failure, combined concrete failure and bent studs and stud shear-off from the steel flange. The elevation of the reinforcement mesh had a significant effect on the ultimate load of the specimen. As the mesh elevation increased from intercepting the stud to being in flush with the top of the stud to above the stud, the ultimate load decreased. Specimens with unheaded shear studs had lower ultimate load than specimens with headed shear studs. Flange treatment had an impact on the ultimate load, where the coating on flanges resulted in a decrease in the ultimate load. Test results also showed that the close placement of the shear studs result in a reduction on the ultimate load when the other parameters were kept the same. In the comparison between conventional and adjustable shear studs, specimens with adjustable studs shared similar failure mode to those with conventional studs, but attained on average lower load capacity. The comparison with the code suggested values showed that the code suggested value is only ensured when double-layer reinforcement mesh is used and placed at code specified elevation. A single layer mesh intercepting the studs resulted in the ultimate load slightly lower than the code value. The code values for adjustable studs are markedly higher than the experimental value, which raises the question whether the code equation for conventional studs is directly transferrable to adjustable studs.

An adjustable systemic-pulmonary artery (SPA) shunt is being developed that consists of apolytetrafluoroethylene (PTFE) graft with a screw plunger mechanism. This device would allowfull control of flow through SPA shunts used to augment pulmonary blood flow in neonates bornwith single ventricle physiology. The objective of this study is to evaluate the changes this mechanismhas on flow fields for a 4 mm and 5 mm adjustable SPA shunt. Two in vitro models wereexamined; an idealized model with an axisymmetric constriction and a model developed from 3-Dreconstruction of the actual shunt under asymmetric constriction. These models were used to measurethe instantaneous velocity and vorticity fields using Particle Image Velocimetry (PIV) underboth steady and pulsatile flow conditions. Recirculation regions and maximum values of velocity,vorticity, and shear stress are compared between the 4 mm and 5 mm models. The results indicatethat for the idealized model of both shunts, separation regions are much smaller, persistingfor approximately 0-1.75 diameters downstream of the constriction, while for the realistic modelsseparation regions of 2.5 diameters downstream were observed. Additional models of a 4 mm and5 mm shunt were tested under pulsatile conditions matching Re of 1061 and 849 and a Womersleynumber of 4.09 and 5.12, respectively, as seen in vivo. The maximum shear rates observed in bothshunts are within an allowable range without inducing platelet aggregation or hemolysis. However,regions of reverse flow exist distal to the throat, leading to possible concerns of plaque formation.Further in vivo testing will be needed to address this concern. This work is part of an extensiveeffort in developing a completely implantable adjustable systemic-pulmonary artery shunt.

The well known synthesis problem of obtaining dimensions of a four-link mechanism such that a point on the coupler link traces a desired path has been extensively studied. There are two types of path generation–path specified by a finite number of precision points where the prescribed points must be exactly traced, and continuous path generation where the path is approximately traced by the coupler point. In various application, more than one or multiple paths are required to be traced by the coupler point and in such cases, adjustable four-link mechanisms where one of the dimension or parameters of the mechanism can be changed is a possible solution. This thesis deals with the synthesis of planar and spherical adjustable four-link crank-rocker type mechanisms for multiple continuous path generation. Approximate multiple path generation is typically solved as an optimization problem where the dimensions and parameters of the four-link mechanism are obtained such that the objective functions, typically in terms of an error between the desired and obtained path, is minimized. In this thesis, we present a two-stage optimization to obtain four-link mechanism dimensions such that the adjustable four-link mechanism can approximately trace multiple desired paths. In the first stage, the parameters in the driving side of the four-link mechanism is obtained and in the second stage, the parameters of the driven side are obtained. In case of adjustable planar four-link mechanism, a novel optimization objective function based on circle-fitting is used and for spherical adjustable mechanisms a novel plane-fitting based objective function is used. The use of these objective functions results in a lesser number of variables to be searched and thus the method presented in this work is more efficient than existing optimization based algorithms available in literature. Several examples are presented for synthesis of adjustable planar and spherical four-link mechanism for tracing multiple paths. In particular, a spherical mechanism which can generate an oval and an‘ 8’shaped path by one single adjustment is synthesized. This mechanism has been made using 3D printing and it is shown that the mechanism indeed traces the desired oval and ‘8’ shaped paths. This mechanism is being planned for use in a flapping wing micro air vehicle where the oval shaped path is known to make the vehicle to move forward while the ‘8’ shaped path results in a hovering motion.