Refine
Year of publication
Document Type
- Master's Thesis (42) (remove)
Language
- English (42) (remove)
Has Fulltext
- yes (42)
Is part of the Bibliography
- no (42)
Keywords
- WaterGAP (2)
- global water model (2)
- (n (1)
- AI Safety (1)
- ALICE (1)
- Activism (1)
- App ecosystem (1)
- Autonomous Driving (1)
- Bayesian Inference (1)
- Beryllium-7 (1)
Institute
In this thesis, Planck size black holes are discussed. Specifically, new families of black holes are presented. Such black holes exhibit an improved short scale behaviour and can be used to implement gravity self-complete paradigm. Such geometries are also studied within the ADD large extra dimensional scenario. This allows black hole remnant masses to reach the TeV scale. It is shown that the evaporation endpoint for this class of black holes is a cold stable remnant. One family of black holes considered in this thesis features a regular de Sitter core that counters gravitational collapse with a quantum outward pressure. The other family of black holes turns out to nicely fit into the holographic information bound on black holes, and lead to black hole area quantization and applications in the gravitational entropic force. As a result, gravity can be derived as emergent phenomenon from thermodynamics.
The thesis contains an overview about recent quantum gravity black hole approaches and concludes with the derivation of nonlocal operators that modify the Einstein equations to ultraviolet complete field equations.
As a part of this thesis, a Monte Carlo-based code has been developed capable of simulating the transition of proton beam properties to neutron beam properties as it occurs in the Li-7(p, n)Be-7 reaction. It is able to reproduce not only the angle-integrated energy distributions but it is also capable of predicting the angle-dependent neutron spectra as measured at Forschungszentrum Karlsruhe (Karlsruhe, Germany) and Physikalisch-Technische Bundesanstalt (Braunschweig, Germany). Since the code retains all three spatial dimensions as well as all three velocity dimensions, it provides very detailed information on the neutron beam. The resulting data can aid in many different aspects, for example it can be used in shielding construction, or for lithium target design. In this work, the code is used to predict the neutron beam properties expected at the Frankfurt Neutron Source at Stern-Gerlach-Zentrum (FRANZ) facility. For different proton beam energies, the neutron distribution in x/p_x, y/p_y, and z/p_z is shown as well as a Mollweide projection, which illustrates the kinematic collimation effect that limits the neutron cone opening angle to less than 180 degree.
The reanalysis products and derived products, ERA5 (Copernicus Climate Change Service, 2018) and W5E5 (WATCH Forcing Data (WFD) methodology applied to ERA5) (LANGE ET AL., 2021) have been recently published initiating a new phase of scientific research utilizing these datasets. ERA5 and W5E5 offer the possibility to reduce insecurities in model results through their improved quality compared to previous climate reanalyses (CUCCHI ET AL., 2020). The suitability of either climate forcing as input for the hydrological model WaterGAP and the influence of the models specific calibration routine has been evaluated with four model experiments. The model was validated by analysing the models ability to produce reasonable values for global water balance components and to reproduce observed discharge in 1427 basins as well as total water storage anomalies in 143 basins using well established efficiency metrics. Bias correction of W5E5 was found to lead to more global realistic mean precipitation and consequently discharge and AET values. In an uncalibrated model setup ERA5 results in better performances across all efficiency metrics. Model results produced with W5E5 as climate input were strongly improved through calibration ultimately leading to the best performances out of all four model experiments. However, model performances considerably improved through calibration with both climate forcings hence calibration was found to have the strongest effect on model performance. Furthermore, spatial differences in performance of either forcing were identified. Snow-dominated regions show an overall better performance with ERA5, while wetter and warmer regions are better represented with W5E5. Finally, it can be concluded that W5E5 should be preferred as climate input for impact modelling; however, depending on the spatial scale and region ERA5 should at least be considered, in particular for snow-dominated regions.
Representations of the reasons and actions of terrorists have appeared in German literature tracing back to the age of Sturm und Drang of the 18th century, most notably in Heinrich von Kleist's Michael Kohlhaas and Friedrich Schiller's Die Räuber, and more recently since the radical actions of the Red Army Faction during the late 1960s and early 1970s, such as in Uli Edel's film, The Baader Meinhof Complex. By referring to Walter Benjamin's system of natural law and positive law, which provides definitions of differing codes of ethics with relation to state laws and personal ethics, one should be able to understand that Michael Kohlhaas, Karl Moor, and the members of the RAF are indeed represented as terrorists. However, their actions and motives are not without an internal ethics, which conflicts with that of their respective state-sanctioned authorities. This thesis reveals the similarities and differences in motives, methods, and use of violence in Schiller, Kleist, and representations of the RAF and explores how the turn to terrorism can arise from a logical realization that ideologies of state law do not align with the personal sense of justice and law of the individual.
In situ rainwater harvesting has a long history in arid and semi-arid regions of the world buffering water shortages for human consumption and agriculture. In the context of an Integrated Water Resource Management (IWRM) in the Cuvelai Basin in northern Namibia, roof top rainwater harvesting is being introduced to a rural community for the irrigation of household scale gardens for the cultivation of horticulture products. This study elaborates how harvested rainwater can be used for garden irrigation in a sustainable manner evaluating ecologic, economic and social implications. Considering local conditions eight cropping scenarios were designed, including different criteria as well as one and two annual planting seasons. These schemes were tested under present climate conditions and under three future climate change scenarios for 2050 with the help of a tank model designed to model monthly tank inflows and outflows. Special attention was laid on risk and uncertainty aspects of varying inter-annual and interseasonal precipitation and future climate change. A framework for the assessment of sustainability was adapted to the purposes of this study and indicators have been developed in order to assess the cropping and irrigation schemes for sustainability.
The study found that with the given tank size of 30 m³, depending on crop scenario, under optimized conditions a garden area of 60 to 90 m³ can be irrigated. The choice of crops highly impacts water use efficiency and economic profitability, compared to the considerably lower impact of amount of annual planting seasons and future climate change. In the case of worsening future climate conditions, adaptation measures need to be taken as especially the economic as well as the environmental situation are expected to exacerbate due to expected decreases in yields and revenues. Already under present conditions however, the economic dimension represents the most limiting factor to sustainability, particularly due to the excessive investment costs of the rainwater harvesting and gardening facility. Nonetheless, rainwater harvesting in combination with gardening can be regarded as successful in securing household nutrition, providing sufficient horticulture products for household consumption or market sale. At the same time with the optimal choice of crops the investment costs can be recovered within the end of the lifespan of the facility.
This study analyzes storyline structure in three Hausa home videos; Mai Kudi (The Rich Man), Sanafahna (with time truth shall dawn) and Albashi (Salary). The study measures storyline structure in these films against a Hollywood film industry model of story writing “the Hero's Journey”. It uses narrative analysis as its analytical tool, and narrative theory as its framework. After analyzing these videos, the study found that the major elements of storyline structure in Vogler's model formed the framework of the storyline structure in Hausa home videos analyzed. However, in spite of the preponderance of these elements within the storyline structure, there are significant variations to Vogler's model. Specifically, Vogler's model has some twelve stages spread on the universal structure of storytelling, i.e. beginning, middle and end. Few of these stages were found to exist in Hausa narrative structure, perhaps due to cultural differences between Western, Indian and Hausa cultures. The study therefore recommends screenwriters and producers to be aware of the existence of standard models of scriptwriting. It also recommends more training for script writers in the Hausa film industry.
Asymptotic giant branch (AGB) stars are initially low and intermediate mass stars undergoing recurrent hydrogen and helium shell burning. During the advanced stage of stellar evolution AGB stars follow after the helium core burning ceased and are located in the AGB of the Hertzsprung-Russell Diagram. One characteristic is their ability of element synthesis, especially carbon and nitrogen, which they eject in large amounts into the interstellar medium. But AGB stars also feature a slow-neutron capture process called s-process which forms approximately 50 % of all elements between Fe and Bi. The initial mass function emphasizes the importance of the synthesized ejecta of AGB stars since they are much more abundant than massive stars. Therefore, the abundance evolution of many elements in the universe is drastically affected by AGB stars. In order to understand chemical evolution in the universe their behavior must be known since their first appearance. In previous times less heavy elements were produced and available. Hence AGB stars with lower heavy element content, which means lower metallicity, must be investigated. They appear to behave substantially differently than stars of higher metallicity. Another issue is that AGB stars have mass-dependent characteristics from which follows a division into low-mass, massive and super AGB stars. Super AGB stars have the most open issues due to their large masses and initial mass boundaries that separate them from massive stars. Due to large spectroscopic surveys in the last years, many low metallicity stars have been analyzed. These findings make it necessary to complement those studies through stellar modeling. This work makes a step in this direction. The AGB star masses under investigation are 1M⊙, 1.65M⊙, 2M⊙, 3M⊙, 4M⊙, 5M⊙, 6M⊙ and 7M⊙ which include low-mass, massive and super AGB stars. Metallicities of Z = 6 x 10 exp-3 and Z = 1 x 10 exp-4 (for comparison, solar Z ~ 0.02) were chosen. These results are an extension of already available data, covering solar and half-solar metallicity, but without super AGB stars. Therefore physics input includes mainly well-established approaches rather than new theories. New physical approaches are included due to the low metallicity which makes the results a unique set of models. Additionally, extensive s-process network calculations lead to production factors of all included elements and isotopes. The s-process signatures of those stars were analyzed. The stellar evolution simulations presented in this work have been utilized for rate and especially sensitivity studies. One approach done was to analyze s-process branchings at 95Zr and 85Kr for stars at 3M⊙ with Z = 1 x 10 exp-2 and Z = 1 x 10 exp-3 respectively.
When performing transfer learning in Computer Vision, normally a pretrained model (source model) that is trained on a specific task and a large dataset like ImageNet is used. The learned representation of that source model is then used to perform a transfer to a target task. Performing transfer learning in this way had a great impact on Computer Vision, because it worked seamlessly, especially on tasks that are related to each other. Current research topics have investigated the relationship between different tasks and their impact on transfer learning by developing similarity methods. These similarity methods have in common, to do transfer learning without actually doing transfer learning in the first place but rather by predicting transfer learning rankings so that the best possible source model can be selected from a range of different source models. However, these methods have focused only on singlesource transfers and have not paid attention to multi-source transfers. Multi-source transfers promise even better results than single-source transfers as they combine information from multiple source tasks, all of which are useful to the target task. We fill this gap and propose a many-to-one task similarity method called MOTS that predicts both, single-source transfers and multi-source transfers to a specific target task. We do that by using linear regression and the source representations of the source models to predict the target representation. We show that we achieve at least results on par with related state-of-the-art methods when only focusing on singlesource transfers using the Pascal VOC and Taskonomy benchmark. We show that we even outperform all of them when using single and multi-source transfers together (0.9 vs. 0.8) on the Taskonomy benchmark. We additionally investigate the performance of MOTS in conjunction with a multi-task learning architecture. The task-decoder heads of a multi-task learning architecture are used in different variations to do multi-source transfers since it promises efficiency over multiple singletask architectures and incurs less computational cost. Results show that our proposed method accurately predicts transfer learning rankings on the NYUD dataset and even shows the best transfer learning results always being achieved when using more than one source task. Additionally, it is further examined that even just using one task-decoder head from the multi-task learning architecture promises better transfer learning results, than using a single-task architecture for the same task, which is due to the shared information from different tasks in the multi-task learning architecture in previous layers. Since the MOTS rankings for selecting the MTI-Net task-decoder head with the highest transfer learning performance were very accurate for the NYUD but not satisfying for the Pascal VOC dataset, further experiments need to varify the generalizability of MOTS rankings for the selection of the optimal task-decoder head from a multi-task architecture.
AI-based computer vision systems play a crucial role in the environment perception for autonomous driving. Although the development of self-driving systems has been pursued for multiple decades, it is only recently that breakthroughs in Deep Neural Networks (DNNs) have led to their widespread application in perception pipelines, which are getting more and more sophisticated. However, with this rising trend comes the need for a systematic safety analysis to evaluate the DNN's behavior in difficult scenarios as well as to identify the various factors that cause misbehavior in such systems. This work aims to deliver a crucial contribution to the lacking literature on the systematic analysis of Performance Limiting Factors (PLFs) for DNNs by investigating the task of pedestrian detection in urban traffic from a monocular camera mounted on an autonomous vehicle. To investigate the common factors that lead to DNN misbehavior, six commonly used state-of-the-art object detection architectures and three detection tasks are studied using a new large-scale synthetic dataset and a smaller real-world dataset for pedestrian detection. The systematic analysis includes 17 factors from the literature and four novel factors that are introduced as part of this work. Each of the 21 factors is assessed based on its influence on the detection performance and whether it can be considered a Performance Limiting Factor (PLF). In order to support the evaluation of the detection performance, a novel and task-oriented Pedestrian Detection Safety Metric (PDSM) is introduced, which is specifically designed to aid in the identification of individual factors that contribute to DNN failure. This work further introduces a training approach for F1-Score maximization whose purpose is to ensure that the DNNs are assessed at their highest performance. Moreover, a new occlusion estimation model is introduced to replace the missing pedestrian occlusion annotations in the real-world dataset. Based on a qualitative analysis of the correlation graphs that visualize the correlation between the PLFs and the detection performance, this study identified 16 of the initial 21 factors as being PLFs for DNNs out of which the entropy, the occlusion ratio, the boundary edge strength, and the bounding box aspect ratio turned out to be most severely affecting the detection performance. The findings of this study highlight some of the most serious shortcomings of current DNNs and pave the way for future research to address these issues.