221 research outputs found
Vindemia Iuris, Sive Dissertatio Iuridica De Vino
No full textQuam Magnifici ICtorum ordinis indultu, Sub Praesidio Dn. Joannis Strauchii ... Publicae disquisitioni exponit Christophorus Albertus Steininger/ Ratisp. A. A.S. MDCLXX. a.d. Kal. Decembr.Nicht identisch mit VD17 12:156149B (dort ohne Widmung
Sammelrezension: Zeit und Medien
No full textWerner Faulstich, Christian Steininger (Hg.): Zeit in den Medien - Medien der ZeitHans-Joachim Bieber, Hans Ottomeyer, Georg Christoph Tholen (Hg.): Die Zeit im Wandel der Zei
Vindemia iuris; sive, Dissertatio iuridica, de vino; oder, Von dem Weine. Quam magnifici ictorum ordinis indultu, sub. praesidio Joannis Strauchii ... Publicae disquisitioni exponit Christoph. Albertus Steininger ... 1670.
No full textMode of access: Internet
FFT specific compilation on IBM Blue Gene
No full textAlgorithmen zur digitalen Transformation von Signalen sind im Scientific Computing von größter Bedeutung und werden in zahlreichen Gebieten angewendet -- von der Echtzeit-Signalverarbeitung in eingebetteten Systemen bis zur numerischen Lösung partieller Differentialgleichungen im Rahmen komplexer Simulationen auf Supercomputern.Die Entwicklung und Publikation der schnellen Fourier-Transformation (FFT) durch Cooley und Tukey leitete die Entwicklung einer neuen Klasse schneller Signalverarbeitungsalgorithmen ein, die -- im Gegensatz zur direkten Auswertung des entsprechenden Matrix-Vektor-Produktes -- nicht eine Berechnungskomplexität von , sondern nur von haben.Für eine bestimmte Transformation gibt es aber nicht nur einen eindeutig bestimmten schnellen Algorithmus, sondern eine ganze Vielzahl äquivalenter Algorithmen. Diese Algorithmen unterscheiden sich nur unwesentlich im Hinblick auf ihren Rechenaufwand, umso mehr aber in ihrem Speicherzugriffsverhalten, was auf modernen Computersystemen mit ihren mehrstufigen Speicherhierarchien zu enormen Laufzeitunterschieden führen kann.Automatische Performance-Tuning-Systeme, wie zum Beispiel die den State-of-the-art verkörpernden Signalverarbeitungs-Programmbibliotheken FFTW und SPIRAL, führen auf einem gegebenen Computersystem eine Suche nach dem optimalen Algorithmus im Raum aller äquivalenten Algorithmen und Implementierungen durch. Da der Output von Performance-Tuning-Systemen aber in Form von C-Code erfolgt, ist deren Leistung durch die Qualität der verfügbaren Compiler beschränkt.Die vorliegende Arbeit stellt einen neu entwickelten Special-Purpose-Compiler vor, der für die Übersetzung laufzeitkritischer Codes existierende C Compiler ersetzt. Dieser Compiler, die Vienna MAP compiler tool-chain, besteht aus den folgenden Komponenten, die speziell den Bedürfnissen von Signalverarbeitungscodes angepasst sind: Der MAP Vectorizer extrahiert 2-weg SIMD-Parallelismus in numerischen Straight-Line-Codes. Der MAP Optimizer führt lokale Codeverbesserungen durch, wie sie von versierten Assembler-Programmierern manuell ausgeführt werden. Zuletzt erzeugt das MAP Backend Assemblercode für die Zielarchitektur.Die wichtigste Ziel-Architektur des MAP-Compilers ist der IBM PowerPC 440 FP2 Prozessor, der in IBM Blue-Gene-Systemen -- den derzeit schnellsten Supercomputern der Welt -- eingesetzt wird.FFTW-Grundroutinen, die mit Hilfe der Blue-Gene-Version des MAP-Compilers übersetzt werden, erreichen eine Effizienz von bis zu 80% und damit die dreifache Leistung jener Objekt-Codes, die von der aktuellsten Version des optimierenden IBM XL C Compilers erzeugt werden.Digital signal transforms are core algorithms in computational science and engineering, ranging from real-time signal processing in small-scale problems with stringent time constraints up to large-scale simulations based on partial differential equation solvers running on the world's largest supercomputers.Starting with Cooley and Tukey's work on the fast Fourier transform (FFT), a vast class of fast signal transform algorithms has been developed, pushing the computational complexity down from to . In practice, however, there is not just one unique algorithm, but a large number of fast algorithms for computing one specific transform. These diverse algorithms are equivalent, but may differ significantly with regard to their memory access behavior, which causes tremendous runtime differences on all common-place machines with deep memory hierarchies.Automatic performance tuning systems---like the state-of-the-art signal transform libraries FFTW and SPIRAL---search the space of suitable algorithms and implementations, automatically generating a large number of promising codes. To obtain the best performing code on a given target hardware, the search process is guided by empirical runtime measurements. However, as the program generators used in automatic performance tuning systems produce high-level C code, the performance of these systems is clearly limited by the quality of available compilers.The thesis at hand presents a newly developed special-purpose compiler---the Vienna MAP compiler tool chain---as a replacement for general purpose high-level compilers in the context of automatic performance tuning systems. The MAP tool chain is composed of several generic components, consecutively focusing on specific properties of signal transform codes. The MAP vectorizer extracts 2-way SIMD-style parallelism out of numerical straight line code. The MAP optimizer performs local code improvements similar to the ones that experienced programmers would achieve by hand. Finally, the MAP backend produces assembly code for the specific target architecture.The primary target architecture of the MAP compiler is IBM's PowerPC 440 FP2 processor used in all Blue Gene systems, currently being the fastest supercomputers worldwide. FFTW's core routines were compiled by the Blue Gene version of the MAP compiler. The resulting assembly code boasts an unprecedented performance---reaching a level of 80% efficiency. That way, MAP compiled codes are up to three times as fast as object codes obtained with the latest version of IBM's optimizing XL C compiler
Information avoidance in the context of privacy decision making
No full textWhen making privacy decisions (mostly this concerns data disclosure in an online context) an individual is confronted with information like Terms of Service or data policies that are meant to support decision making. However, such information might be avoided by individuals. Such behaviour contradicts the efforts of legislators like the European Union (EU) to ensure an informed decision making of citizens.
This thesis examines this phenomenon of ‘information avoidance’ explicitly in a privacy context. Furthermore, it is investigated how information avoidance in privacy decisions is influenced (or influences) transparency measures. Thereby, the usefulness of transparency measures in the presence of information avoidance is investigated. This thesis finds that privacy decision making is neither completely rational nor irrational. There are influences besides a rational calculus. The same yields for information avoidance: It is neither purely rational nor irrational. Based on these findings, two sets of reasons for information avoidance in privacy decision making are identified: There are information-driven reasons and situational factors that determine the decision whether to avoid information. When comparing the reasons of information avoidance and the possible effects of transparency, it becomes evident that transparency measures can mitigate (especially) situational driven reasons of information avoidance. However, considering the EU’s General Data Protection Regulation (GDPR) as an concrete example for transparency regulation and examining its impact to date, it becomes obvious that current transparency measures are still not being implemented in a way that ensures the full potential of transparency is realised. Information that is meant to support decision making (e.g. data policies) is still too complex, provided in a false form or not at the right time.
Thus, while transparency measures can be an effective mean for informed decision making, their implementation is crucial: There are various mechanisms or tools that can provide transparency but they have to be empowered in an effective way by regulation. In this context, also the empowerment of such regulation has to be ensured. More research is indispensable to evaluate the effects of transparency requirements since regulations which require so (e.g. GDPR) are usually not long in place and the effects are still unfolding.. The topic of information avoidance is especially relevant in this context. Therefore, there should be further research on information avoidance with a focus on privacy decision-making, as most research on information avoidance has so far been conducted in a more general context
Characterization of a reduced graphene oxide based biosensor via infrared spectroscopy
No full textThe fabrication of a small and user-friendly sensor, which has a high dynamic range in which it is sensitive and specific for certain analytes is a major topic in research in the last decades. The first fabrication of a graphene monolayer in 2004 caused a boom in material research because its unique electrical properties make it a potential candidate to replace materials like copper (a conductor) and silicon (a semiconductor). However, many preparation steps have to be optimized. Furthermore, diverse factors which could potentially effect the functionality of the device have to be determined. Graphene is often used in biosensors which utilise biomolecules like DNA, aptamers or proteins. Those can specifically interact with the analyte which is further detected by the device. A grapheneba.sed field-effect transistor (GFET) was developed in our group which uses a protein as a sensing molecule. The protein is the odorant binding protein 14 from the honey bee (Apis mellifera). It binds specific odorants and can consequently be used for an artificial nose. The first part of the thesis analyses the fabrication of the FET step by step via surface sensitive methods. Beside static measurements, dynamic measurements were done for critical steps. This data was used for optimization of the setup. The graphene product of Freicht et al. showed the best result for optimal surface coverage. Hydroiodic acid showed the best reduction properties. Measurements, done via infrared spectroscopy, showed that OBP14 adsorbs stable via the PBSE linker on the surface. Further experiments were done via polarized attenuated total reflection (ATR) infrared spectroscopy to determine the orientation of certain vibrations and the surface concentration of the protein and linker. The protein concentration was determined to be 9.68 ± 3.03pM I cm2 or 157 ± 50ng I cm2 respectively. This is equivalent to a surface concentration of approximately 60%. The last part deals with the effect of pH and ligand binding on the sensor protein. Therefore, the protein was immobilized via His-tag on the surface of an ATR-crystal and the environment changed periodically. The decrease in pH showed a significant spectral change which is attributed to the weakening of H-bridges in the protein and rearrangement of helix 7, which causes opening of the binding pocket. Additionally, a theoretical mechanism for ligand uptake and release was proposed. The ligand modulation showed only a very weak signal. Due to the low intensities and poor signal-to-noise ratio it was not possible to obtain clear information on structural changes or the kinetics of ligand binding.Doctor of Philosophy (MSE
Geschichtliche Rekonstruktion der linken Terrorbewegung in der BRD von 1968 - 1998 dargestellt am Beispiel der RAF
No full text1970 erklärte eine kleine linksradikale Gruppierung dem Staat den Krieg. Die BürgerInnenkinder aus der Mitte der Gesellschaft nannten sich Rote Armee Fraktion. Diese Diplomarbeit soll rekonstruieren, wie aus der an sich friedlichen StudentInnenbewegung, Anti- Vietnam- Bewegung und den Protesten gegen die fehlende Aufarbeitung der NS- Zeit blutiger Terror entstand, der die Republik an den Rand des Staatsnotstands trieb. Gerade nach der Haftentlassung des ehemaligen RAF- Terroristen Christian Klar am 3. Januar 2008 - er war 26 Jahre lang im Gefängnis- rückt dieses Thema wieder ins Blickfeld der Öffentlichkeit. Nach der Freilassung Klars ist nur noch Birgit Hogefeld, Mitglied der letzten RAF- Gruppe, in Haft. Ob damit ein Schlussstrich unter dieses dunkle Kapitel deutscher Geschichte gezogen werden kann, bleibt fraglich. In der vorliegenden Arbeit werden Widersprüche aufgedeckt, in die sich die RAF innerhalb ihres langjährigen Bestehens immer mehr verstrickt hat. Natürlich sollen auch Parallelitäten und Gemeinsamkeiten innerhalb der linken Terrororganisationen und in der RAF, im speziellen in Hinblick auf deren Weiterbestand, untersucht werden. Fragen nach Kooperationen mit anderen terroristischen Organisationen im In- und Ausland, konkrete Handlungsweisen, Funktionslogiken, Gewaltbereitschaft und ihr Umgang mit den Medien und Staat bieten sich in der Untersuchung durchaus an
- …
