Refine
Year of publication
- 2018 (153) (remove)
Document Type
- Doctoral Thesis (153) (remove)
Language
- English (153) (remove)
Is part of the Bibliography
- yes (153) (remove)
Keywords
- Fernerkundung (3)
- Magnetismus (3)
- climate change (3)
- magnetism (3)
- remote sensing (3)
- uncertainty (3)
- Angriffserkennung (2)
- Bakterien (2)
- Big Data (2)
- Biodiversität (2)
Institute
- Institut für Physik und Astronomie (28)
- Institut für Biochemie und Biologie (27)
- Institut für Chemie (24)
- Institut für Geowissenschaften (22)
- Hasso-Plattner-Institut für Digital Engineering GmbH (14)
- Extern (13)
- Wirtschaftswissenschaften (8)
- Department Psychologie (6)
- Department Linguistik (5)
- Institut für Ernährungswissenschaft (5)
Answer Set Programming (ASP) is a declarative problem solving approach, combining a rich yet simple modeling language with high-performance solving capabilities. Although this has already resulted in various applications, certain aspects of such applications are more naturally modeled using variables over finite domains, for accounting for resources, fine timings, coordinates, or functions. Our goal is thus to extend ASP with constraints over integers while preserving its declarative nature. This allows for fast prototyping and elaboration tolerant problem descriptions of resource related applications. The resulting paradigm is called Constraint Answer Set Programming (CASP).
We present three different approaches for solving CASP problems. The first one, a lazy, modular approach combines an ASP solver with an external system for handling constraints. This approach has the advantage that two state of the art technologies work hand in hand to solve the problem, each concentrating on its part of the problem. The drawback is that inter-constraint dependencies cannot be communicated back to the ASP solver, impeding its learning algorithm. The second approach translates all constraints to ASP. Using the appropriate encoding techniques, this results in a very fast, monolithic system. Unfortunately, due to the large, explicit representation of constraints and variables, translation techniques are restricted to small and mid-sized domains. The third approach merges the lazy and the translational approach, combining the strength of both while removing their weaknesses. To this end, we enhance the dedicated learning techniques of an ASP solver with the inferences of the translating approach in a lazy way. That is, the important knowledge is only made explicit when needed.
By using state of the art techniques from neighboring fields, we provide ways to tackle real world, industrial size problems. By extending CASP to reactive solving, we open up new application areas such as online planning with continuous domains and durations.
The development of self-adaptive software requires the engineering of an adaptation engine that controls the underlying adaptable software by a feedback loop. State-of-the-art approaches prescribe the feedback loop in terms of numbers, how the activities (e.g., monitor, analyze, plan, and execute (MAPE)) and the knowledge are structured to a feedback loop, and the type of knowledge. Moreover, the feedback loop is usually hidden in the implementation or framework and therefore not visible in the architectural design. Additionally, an adaptation engine often employs runtime models that either represent the adaptable software or capture strategic knowledge such as reconfiguration strategies. State-of-the-art approaches do not systematically address the interplay of such runtime models, which would otherwise allow developers to freely design the entire feedback loop.
This thesis presents ExecUtable RuntimE MegAmodels (EUREMA), an integrated model-driven engineering (MDE) solution that rigorously uses models for engineering feedback loops. EUREMA provides a domain-specific modeling language to specify and an interpreter to execute feedback loops. The language allows developers to freely design a feedback loop concerning the activities and runtime models (knowledge) as well as the number of feedback loops. It further supports structuring the feedback loops in the adaptation engine that follows a layered architectural style. Thus, EUREMA makes the feedback loops explicit in the design and enables developers to reason about design decisions.
To address the interplay of runtime models, we propose the concept of a runtime megamodel, which is a runtime model that contains other runtime models as well as activities (e.g., MAPE) working on the contained models. This concept is the underlying principle of EUREMA. The resulting EUREMA (mega)models are kept alive at runtime and they are directly executed by the EUREMA interpreter to run the feedback loops. Interpretation provides the flexibility to dynamically adapt a feedback loop. In this context, EUREMA supports engineering self-adaptive software in which feedback loops run independently or in a coordinated fashion within the same layer as well as on top of each other in different layers of the adaptation engine. Moreover, we consider preliminary means to evolve self-adaptive software by providing a maintenance interface to the adaptation engine.
This thesis discusses in detail EUREMA by applying it to different scenarios such as single, multiple, and stacked feedback loops for self-repairing and self-optimizing the mRUBiS application. Moreover, it investigates the design and expressiveness of EUREMA, reports on experiments with a running system (mRUBiS) and with alternative solutions, and assesses EUREMA with respect to quality attributes such as performance and scalability.
The conducted evaluation provides evidence that EUREMA as an integrated and open MDE approach for engineering self-adaptive software seamlessly integrates the development and runtime environments using the same formalism to specify and execute feedback loops, supports the dynamic adaptation of feedback loops in layered architectures, and achieves an efficient execution of feedback loops by leveraging incrementality.
The sequencing of the human genome in the early 2000s led to an increased interest in cheap and fast sequencing technologies. This interest culminated in the advent of next generation sequencing (NGS). A number of different NGS platforms have arisen since then all promising to do the same thing, i.e. produce large amounts of genetic information for relatively low costs compared to more traditional methods such as Sanger sequencing. The capabilities of NGS meant that researchers were no longer bound to species for which a lot of previous work had already been done (e.g. model organisms and humans) enabling a shift in research towards more novel and diverse species of interest. This capability has greatly benefitted many fields within the biological sciences, one of which being the field of evolutionary biology. Researchers have begun to move away from the study of laboratory model organisms to wild, natural populations and species which has greatly expanded our knowledge of evolution. NGS boasts a number of benefits over more traditional sequencing approaches. The main benefit comes from the capability to generate information for drastically more loci for a fraction of the cost. This is hugely beneficial to the study of wild animals as, even when large numbers of individuals are unobtainable, the amount of data produced still allows for accurate, reliable population and species level results from a small selection of individuals.
The use of NGS to study species for which little to no previous research has been carried out on and the production of novel evolutionary information and reference datasets for the greater scientific community were the focuses of this thesis. Two studies in this thesis focused on producing novel mitochondrial genomes from shotgun sequencing data through iterative mapping, bypassing the need for a close relative to serve as a reference sequence. These mitochondrial genomes were then used to infer species level relationships through phylogenetic analyses. The first of these studies involved reconstructing a complete mitochondrial genome of the bat eared fox (Otocyon megalotis). Phylogenetic analyses of the mitochondrial genome confidently placed the bat eared fox as sister to the clade consisting of the raccoon dog and true foxes within the canidae family. The next study also involved reconstructing a mitochondrial genome but in this case from the extinct Macrauchenia of South America. As this study utilised ancient DNA, it involved a lot of parameter testing, quality controls and strict thresholds to obtain a near complete mitochondrial genome devoid of contamination known to plague ancient DNA studies. Phylogenetic analyses confidently placed Macrauchenia as sister to all living representatives of Perissodactyla with a divergence time of ~66 million years ago. The third and final study of this thesis involved de novo assemblies of both nuclear and mitochondrial genomes from brown and striped hyena and focussed on demographic, genetic diversity and population genomic analyses within the brown hyena. Previous studies of the brown hyena hinted at very low levels of genomic diversity and, perhaps due to this, were unable to find any notable population structure across its range. By incorporating a large number of genetic loci, in the form of complete nuclear genomes, population structure within the brown hyena was uncovered. On top of this, genomic diversity levels were compared to a number of other species. Results showed the brown hyena to have the lowest genomic diversity out of all species included in the study which was perhaps caused by a continuous and ongoing decline in effective population size that started about one million years ago and dramatically accelerated towards the end of the Pleistocene.
The studies within this thesis show the power NGS sequencing has and its utility within evolutionary biology. The most notable capabilities outlined in this thesis involve the study of species for which no reference data is available and in the production of large amounts of data, providing evolutionary answers at the species and population level that data produced using more traditional techniques simply could not.
Business process management is an acknowledged asset for running an organization in a productive and sustainable way. One of the most important aspects of business process management, occurring on a daily basis at all levels, is decision making. In recent years, a number of decision management frameworks have appeared in addition to existing business process management systems. More recently, Decision Model and Notation (DMN) was developed by the OMG consortium with the aim of complementing the widely used Business Process Model and Notation (BPMN). One of the reasons for the emergence of DMN is the increasing interest in the evolving paradigm known as the separation of concerns. This paradigm states that modeling decisions complementary to processes reduces process complexity by externalizing decision logic from process models and importing it into a dedicated decision model. Such an approach increases the agility of model design and execution. This provides organizations with the flexibility to adapt to the ever increasing rapid and dynamic changes in the business ecosystem. The research gap, identified by us, is that the separation of concerns, recommended by DMN, prescribes the externalization of the decision logic of process models in one or more separate decision models, but it does not specify this can be achieved.
The goal of this thesis is to overcome the presented gap by developing a framework for discovering decision models in a semi-automated way from information about existing process decision making. Thus, in this thesis we develop methodologies to extract decision models from: (1) control flow and data of process models that exist in enterprises; and (2) from event logs recorded by enterprise information systems, encapsulating day-to-day operations. Furthermore, we provide an extension of the methodologies to discover decision models from event logs enriched with fuzziness, a tool dealing with partial knowledge of the process execution information. All the proposed techniques are implemented and evaluated in case studies using real-life and synthetic process models and event logs. The evaluation of these case studies shows that the proposed methodologies provide valid and accurate output decision models that can serve as blueprints for executing decisions complementary to process models. Thus, these methodologies have applicability in the real world and they can be used, for example, for compliance checks, among other uses, which could improve the organization's decision making and hence it's overall performance.
Active and passive source data from two seismic experiments within the interdisciplinary project TIPTEQ (from The Incoming Plate to mega Thrust EarthQuake processes) were used to image and identify the structural and petrophysical properties (such as P- and S-velocities, Poisson's ratios, pore pressure, density and amount of fluids) within the Chilean seismogenic coupling zone at 38.25°S, where in 1960 the largest earthquake ever recorded (Mw 9.5) occurred. Two S-wave velocity models calculated using traveltime and noise tomography techniques were merged with an existing velocity model to obtain a 2D S-wave velocity model, which gathered the advantages of each individual model. In a following step, P- and S-reflectivity images of the subduction zone were obtained using different pre stack and post-stack depth migration techniques. Among them, the recent prestack line-drawing depth migration scheme yielded revealing results. Next, synthetic seismograms modelled using the reflectivity method allowed, through their input 1D synthetic P- and S-velocities, to infer the composition and rocks within the subduction zone. Finally, an image of the subduction zone is given, jointly interpreting the results from this work with results from other studies. The Chilean seismogenic coupling zone at 38.25°S shows a continental crust with highly reflective horizontal, as well as (steep) dipping events. Among them, the Lanalhue Fault Zone (LFZ), which is interpreted to be east-dipping, is imaged to very shallow depths. Some steep reflectors are observed for the first time, for example one near the coast, related to high seismicity and another one near the LFZ. Steep shallow reflectivity towards the volcanic arc could be related to a steep west-dipping reflector interpreted as fluids and/or melts, migrating upwards due to material recycling in the continental mantle wedge. The high resolution of the S-velocity model in the first kilometres allowed to identify several sedimentary basins, characterized by very low P- and S-velocities, high Poisson's ratios and possible steep reflectivity. Such high Poisson's ratios are also observed within the oceanic crust, which reaches the seismogenic zone hydrated due to bending-related faulting. It is interpreted to release water until reaching the coast and under the continental mantle wedge. In terms of seismic velocities, the inferred composition and rocks in the continental crust is in agreement with field geology observations at the surface along the proflle. Furthermore, there is no requirement to call on the existence of measurable amounts of present-day fluids above the plate interface in the continental crust of the Coastal Cordillera and the Central Valley in this part of the Chilean convergent margin. A large-scale anisotropy in the continental crust and upper mantle, previously proposed from magnetotelluric studies, is proposed from seismic velocities. However, quantitative studies on this topic in the continental crust of the Chilean seismogenic zone at 38.25°S do not exist to date.
Earth's climate varies continuously across space and time, but humankind has witnessed only a small snapshot of its entire history, and instrumentally documented it for a mere 200 years. Our knowledge of past climate changes is therefore almost exclusively based on indirect proxy data, i.e. on indicators which are sensitive to changes in climatic variables and stored in environmental archives. Extracting the data from these archives allows retrieval of the information from earlier times. Obtaining accurate proxy information is a key means to test model predictions of the past climate, and only after such validation can the models be used to reliably forecast future changes in our warming world. The polar ice sheets of Greenland and Antarctica are one major climate archive, which record information about local air temperatures by means of the isotopic composition of the water molecules embedded in the ice. However, this temperature proxy is, as any indirect climate data, not a perfect recorder of past climatic variations. Apart from local air temperatures, a multitude of other processes affect the mean and variability of the isotopic data, which hinders their direct interpretation in terms of climate variations. This applies especially to regions with little annual accumulation of snow, such as the Antarctic Plateau. While these areas in principle allow for the extraction of isotope records reaching far back in time, a strong corruption of the temperature signal originally encoded in the isotopic data of the snow is expected. This dissertation uses observational isotope data from Antarctica, focussing especially on the East Antarctic low-accumulation area around the Kohnen Station ice-core drilling site, together with statistical and physical methods, to improve our understanding of the spatial and temporal isotope variability across different scales, and thus to enhance the applicability of the proxy for estimating past temperature variability. The presented results lead to a quantitative explanation of the local-scale (1–500 m) spatial variability in the form of a statistical noise model, and reveal the main source of the temporal variability to be the mixture of a climatic seasonal cycle in temperature and the effect of diffusional smoothing acting on temporally uncorrelated noise. These findings put significant limits on the representativity of single isotope records in terms of local air temperature, and impact the interpretation of apparent cyclicalities in the records. Furthermore, to extend the analyses to larger scales, the timescale-dependency of observed Holocene isotope variability is studied. This offers a deeper understanding of the nature of the variations, and is crucial for unravelling the embedded true temperature variability over a wide range of timescales.
Microswimmers, i.e. swimmers of micron size experiencing low Reynolds numbers, have received a great deal of attention in the last years, since many applications are envisioned in medicine and bioremediation. A promising field is the one of magnetic swimmers, since magnetism is biocom-patible and could be used to direct or actuate the swimmers. This thesis studies two examples of magnetic microswimmers from a physics point of view.
The first system to be studied are magnetic cells, which can be magnetic biohybrids (a swimming cell coupled with a magnetic synthetic component) or magnetotactic bacteria (naturally occurring bacteria that produce an intracellular chain of magnetic crystals). A magnetic cell can passively interact with external magnetic fields, which can be used for direction. The aim of the thesis is to understand how magnetic cells couple this magnetic interaction to their swimming strategies, mainly how they combine it with chemotaxis (the ability to sense external gradient of chemical species and to bias their walk on these gradients). In particular, one open question addresses the advantage given by these magnetic interactions for the magnetotactic bacteria in a natural environment, such as porous sediments. In the thesis, a modified Active Brownian Particle model is used to perform simulations and to reproduce experimental data for different systems such as bacteria swimming in the bulk, in a capillary or in confined geometries. I will show that magnetic fields speed up chemotaxis under special conditions, depending on parameters such as their swimming strategy (run-and-tumble or run-and-reverse), aerotactic strategy (axial or polar), and magnetic fields (intensities and orientations), but it can also hinder bacterial chemotaxis depending on the system.
The second example of magnetic microswimmer are rigid magnetic propellers such as helices or random-shaped propellers. These propellers are actuated and directed by an external rotating magnetic field. One open question is how shape and magnetic properties influence the propeller behavior; the goal of this research field is to design the best propeller for a given situation. The aim of the thesis is to propose a simulation method to reproduce the behavior of experimentally-realized propellers and to determine their magnetic properties. The hydrodynamic simulations are based on the use of the mobility matrix. As main result, I propose a method to match the experimental data, while showing that not only shape but also the magnetic properties influence the propellers swimming characteristics.