The optimal implementation of matrix multiplication on modern computer architectures is of great importance for scientific and engineering applications. However, achieving the optimal performance for matrix multiplication has been continuously challenged both by the ever-widening performance gap between the processor and memory hierarchy and the introduction of new architectural features in modern architectures. The conventional way of dealing with these challenges benefits significantly from the blocking algorithm, which improves the data locality in the cache memory, and from the highly tuned inner kernel routines, which in turn exploit the architectural aspects on the specific processor to deliver near peak performance. A state-of-art improvement of the blocking algorithm is the self-tuning approach that utilizes "heroic" combinatorial optimization of parameters spaces. Other recent research approaches include the approach that explicitly blocks for the TLB (Translation Lookaside Buffer) and the hierarchical formulation that employs memoryriendly Morton Ordering (a spaceilling curve methodology). This thesis compares and contrasts the TLB-blocking-based and Morton-Order-based methods for dense matrix multiplication, and offers a qualitative model to explain the performance behavior. Comparisons to the performance of self-tuning library and the "vendor" library are also offered for the Alpha architecture. The practical benchmark experiments demonstrate that neither conventional blocking-based implementations nor the self-tuning libraries are optimal to achieve consistent high performance in dense matrix multiplication of relatively large square matrix size. Instead, architectural constraints and issues evidently restrict the critical path and options available for optimal performance, so that the relatively simple strategy and framework presented in this study offers higher and flatter overall performance. Interestingly, maximal inner kernel efficiency is not a guarantee of global minimal multiplication time. Also, efficient and flat performance is possible at all problem sizes that fit in main memory, rather than "jagged" performance curves often observed in blocking and self-tuned blocking libraries.

No description available.

No description available.

No description available.

No description available.

<p>Bifogad inspelning av konserten:</p><p></p><p>Repertoar:</p><p>Francesco Guerrero - Duo Seraphim</p><p>Francis Poulenc - Sept chansons nr 1, 2, 4, 5, 7</p><p></p><p>Medverkande: Radiokören</p><p></p>

This dissertation investigates the nonlinear wave propagation phenomena in nonlinear metamaterials with nonlinear chains and nonlinear resonators using analytical and spectro-spatial analyses. In the first part of the thesis, the nonlinear metamaterials are modeled as a chain of masses with multiple local resonators attached to each cell. The nonlinearity stems from the chain's stiffness in one case and the local resonator's stiffness in another. Analytical approximates solutions are obtained for each case using perturbation techniques. These results are validated through numerical simulations and the results show good agreement. To further demonstrate the nonlinear wave propagation characteristics, spectro-spatial analyses are conducted on the numerical integration data sets. The wave profiles, short-term Fourier transform spectrograms, and contour plots of 2D Fourier transform show the presence of solitary waves for both sources of nonlinearity. In addition, spectro-spatial features demonstrate the presence of significant frequency shifts at different wavelength limits. indent The second part of the thesis studies a nonlinear electromechanical metamaterial and examines how the electromechanical coupling in the local resonator affects the wave propagation. Numerical examples indicate that the system can be used for simultaneous energy harvesting and vibration attenuation without any degradation in the size of bandgaps. Spectro-spatial analyses conducted on the electromechanical metamaterial also reveal the presence of solitons and frequency shifts. The presence of solitary wave in the electromechanical metamaterial suggests a significant improvement in energy harvesting and sensing techniques. The obtained significant frequency shift is employed to design an electromechanical diode, allowing voltage to be sensed and harvested only in one direction. Design guidelines and the role of different key parameters are presented to help designers to select the type of nonlinearity and the system parameters to improve the performance of acoustic diodes. indent The last part of this thesis studies the passive self-tuning of a metastructure via a beam-sliding mass concept. The governing equations of motions of the holding structure, resonator, and sliding mass are presented and discretized into a system of ODEs using Galerkin's projection. Given that the spatial parameters of the system continuously change over time (i.e., mode shapes and frequencies), instantaneous exact mode shapes and frequencies are determined for all possible slider positions. The numerical integration is conducted by continuously updating the spatial state of the system. The obtained exact mode shapes demonstrate that the resonance frequency of the resonator stretches over a wide frequency band. This observation indicates that the resonator can attenuates vibrations at a wide frequency range. Experiments are also conducted to demonstrate the passive self-tunability of the metastructure and the findings colloborate the analytical results. / Doctor of Philosophy / Metamaterials are artificially engineered structures that can offer incredible dynamical properties, which cannot be found in conventional homogeneous structures. Consequently, the global metamaterials market is expected to display a 23.6$%$ compound annual growth rate through 2027. Some of these exciting properties include, but not limited to, negative stiffness, negative mass, negative Poisson's ratio. The unique dynamic properties show the importance of metamaterials in many engineering applications, such as vibration reduction, noise control, and waveguiding and localization. However, beyond the linear characteristics of metamaterials, nonlinear metamaterials can exhibit more interesting nonlinear wave propagation phenomena, such as solitons, cloaking, tunable bandgaps, and wave non-reciprocity. indent This research work investigates wave propagation characteristics in nonlinear locally resonant metamaterials using analytical, numerical, and signal processing techniques. The nonlinearity stems from the chain in one case and from the local resonator in another. Numerical examples show the presence of solitary waves in both types of nonlinearity and significant frequency shift in certain frequency/wavenumber regions. The obtained significant frequency shift can be utilized to design mechanical diodes, where its operation range can be increased by introducing nonlinearity in the resonator. indent For simultaneous energy harvesting and vibration attenuation, integrating the local resonator with piezoelectric energy harvesters is also investigated in this research work with the presence of both types of nonlinearities. For weak electromechanical coupling, the results demonstrate that the band structure of the system is not affected by the electromechanical coupling. Therefore, the system can also be used for energy harvesting without any degradation in the vibration attenuation performance. This observation is also validated experimentally for the linear limit. Spectro-spatial analyses also reveal the presence of solitary output voltage waves, which can enhance the energy harvesting and sensing. The obtained significant frequency shift can be utilized to design an electromechanical diode where the wave can propagate and be harvested only in one direction. Numerical examples show that the performance of the electromechanical diode can be significantly improved by including nonlinearities in the local resonator. indent Another goal of this research work is the introduction of passive self-tuning mechanism to design self-tuning metastructure. The design of such a metastructure is motivated by the need for broadband devices that can adapt to changing environment. The passive self-tuning concept is achieved by a sliding mass coupled with a resonator. Analytical and experimental results show the ability of this system to tune itself to the excitation frequency, and hence, can control vibrations over a significantly wider frequency band as compared to conventional resonators.

Micro gas chromatography (μGC) system is a miniaturized and portable version of the conventional GC system, suitable for various applications such as healthcare and environmental analysis. The process of gas chromatography requires precise temperature control for the micro-fabricated preconcentrators and separation columns used since temperature changes directly affect retention time. Proportional Integral and Derivative (PID) controllers provide reliable temperature control and can be tuned to obtain the desired response. The conventional method of tuning the PID control parameters by trial and error is a tedious process and time-consuming process. This thesis aims to develop a PID auto-tuning and control system for auto-tuning microfabricated heaters in modular μGC systems. The developed system is based on the Ziegler Nichols rule-based PID tuning method for closed-loop systems, which uses the relay response of the micro-heater to calculate the PID tuning parameters. The system also includes an analysis system to verify the performance of the PID-tuned values and a tuning system where the PID values can be further tuned to obtain more precise control for the heaters. The aim of developing this system is to reduce the effective tuning time for heaters while satisfying the control requirements. In this thesis, we discuss the tuning methodology and the implementation of the PID tuning and control system, followed by a performance evaluation of the heaters tuned using the proposed system is discussed. / Master of Science / Gas chromatography (GC) is an established technique used for the qualitative and quantitative analysis of compounds present in a mixture. Micro-gas chromatography (μGC) systems are miniaturized versions of conventional GC systems. They are portable, energy-efficient, and facilitate on-site analysis in real-time, which is suitable for applications such as health care, forensics, and environmental analysis, requiring in-field analysis. GC is based on the principle that components of a gaseous mixture, when passed through a heated column coated with a stationary phase, separate out based on their extent of interaction with the stationary phase. The temperature control needs to be precise since it directly affects the process. PID control is the most common and reliable method for temperature control. It can be tuned to obtain the desired response, which can, however, be a tedious process. This thesis aims to develop a PID auto-tuning and control system for μ-fabricated heaters in μGC systems. As a part of this thesis, a system facilitating faster tuning of PID parameters for a given heater using the Ziegler Nichols closed-loop tuning method is developed. It uses the relay response of the micro-heater to determine the tuning value. The obtained PID values can be evaluated using the analysis system developed as a part of the system and can be further fine-tuned using the provided system to obtain the desired response. As a part of this thesis, we first discuss the development of the PID tuning and control system, after which the performance of the tuned values is evaluated for two micro-heaters.

This thesis describes the implementation of zero-cost auto-tuning techniques for embedded Proportional Integral and Derivative (PID) controllers, specifically focusing on their application in the control of proportional valves within Micro Gas Chromatography (uGC) systems. uGC systems are miniaturized versions of conventional GC systems, and require precise temperature, flow and pressure control for the micro-fabricated preconcentrators and micro columns. PID controllers are widely used in process control applications due to their simplicity and effectiveness. The Commercial Off The Shelf (COTS) available controllers are expensive, bulky, need system compatibility and have high lead times. The proposed auto-tuner features simple Python-implemented empirical calculations based on Ziegler Nichols relay-based PID tuning method to determine the optimal PID gains. Leveraging Wi-Fi the system enables tuning for any embedded platform while visualizing transient response through the Graphical User Interface (GUI). The embedded-GUI interface provides a customizable auto-tuning experience extending usage across diverse temperature, pressure and flow regulation applications in environmental analysis. Specifically for uGC systems, the GUI integrates with existing hardware stack using minor software enhancements to enable rapid, automated PID tuning for thermal and flow control applications. The performance is analyzed by evaluating response metrics including overshoot, rise time, and steady-state error. / Master of Science / Commercially available flow and thermal regulators are expensive and bulky. In applications like micro gas chromatography (uGC) systems, these commercial tools to regulate actuator control reduce portability and may require different regulators for different control ranges. To overcome these challenges, we developed an open-source, transparent Proportional-Integral-Derivative (PID) auto-tuner for micro-electromechanical systems (MEMS) actuators in uGC systems. The proposed Python-based Graphical User Interface (GUI) approach leverages simple empirically-driven calculations to determine optimal gains. By interfacing with any embedded system through standard connection like Wi-Fi, the auto-tuner enables interactive, vendor-agnostic tuning while visualizing full transient response. This provides accessible, customizable auto-tuning capabilities to enhance closed-loop PID control across instrumentation and device applications at no or minimal additional hardware cost. In uGC systems, we utilize the same setup for thermal, flow, and pressure control, with additional sensor costs offset by the implementation of multiple closed loops on the same system.Precise temperature and flow control is critical in many applications, such as minimizing fluctuations in analyte retention times in uGC systems. PID control offers reliable closed-loop control for such applications, but tedious manual tuning is required for each system. The proposed auto-tuner presented in this work will greatly simplify PID tuning to improve temperature and flow rate precision in these systems. The performance is analyzed by evaluating response metrics including overshoot, rise time, and steady-state error. This thesis discusses the auto-tuning technique, PID implementation, and experimental performance analysis. Overall, this work presents a novel embedded PID automated methodology for rapid and precise thermal and flow control in uGC and other precision regulation applications. The proposed auto-tuning method provides effective tuning across a wide variety of applications such as motors, temperature and pressure control, and flow regulation systems.

This thesis project aims to improve text summarization in the financial application by fine-tuningGenerative Pre-trained Transformer 3.5 (GPT-3.5) . Through meticulous training andoptimization, the model was adeptly configured to accurately and efficiently condense complexfinancial reports into concise, informative summaries, specifically designed to support decision-making in professional business environments. Notable improvements were demonstrated inthe model's capacity to retain essential financial details while enhancing the readability andcontextual relevance of the text, as evidenced by superior ROUGE and BLEU scores whencompared to the baseline GPT-3.5 Turbo model. This fine-tuning approach not only underscoresGPT-3.5’s remarkable adaptability to domain-specific challenges but also marks a significantadvancement in the field of automated text summarization within the financial sector. Thefindings from this research highlight the transformative potential of bespoke NLP solutions,offering data-driven industries the tools to rapidly generate precise and actionable businessinsights, thus facilitating more informed decision-making processes