Refine
Has Fulltext
- yes (130) (remove)
Year of publication
- 2019 (130) (remove)
Document Type
- Doctoral Thesis (130) (remove)
Is part of the Bibliography
- yes (130)
Keywords
- Klimawandel (4)
- Spektroskopie (4)
- climate change (4)
- Biodiversität (3)
- machine learning (3)
- spectroscopy (3)
- Anden (2)
- Andes (2)
- Himalaya (2)
- Klima (2)
Institute
- Institut für Biochemie und Biologie (22)
- Institut für Geowissenschaften (18)
- Institut für Physik und Astronomie (17)
- Institut für Chemie (15)
- Hasso-Plattner-Institut für Digital Engineering GmbH (10)
- Extern (9)
- Department Linguistik (8)
- Institut für Umweltwissenschaften und Geographie (6)
- Sozialwissenschaften (5)
- Institut für Mathematik (4)
Medical imaging plays an important role in disease diagnosis, treatment planning, and clinical monitoring. One of the major challenges in medical image analysis is imbalanced training data, in which the class of interest is much rarer than the other classes. Canonical machine learning algorithms suppose that the number of samples from different classes in the training dataset is roughly similar or balance. Training a machine learning model on an imbalanced dataset can introduce unique challenges to the learning problem.
A model learned from imbalanced training data is biased towards the high-frequency samples. The predicted results of such networks have low sensitivity and high precision. In medical applications, the cost of misclassification of the minority class could be more than the cost of misclassification of the majority class. For example, the risk of not detecting a tumor could be much higher than referring to a healthy subject to a doctor. The current Ph.D. thesis introduces several deep learning-based approaches for handling class imbalanced problems for learning multi-task such as disease classification and semantic segmentation.
At the data-level, the objective is to balance the data distribution through re-sampling the data space: we propose novel approaches to correct internal bias towards fewer frequency samples. These approaches include patient-wise batch sampling, complimentary labels, supervised and unsupervised minority oversampling using generative adversarial networks for all.
On the other hand, at algorithm-level, we modify the learning algorithm to alleviate the bias towards majority classes. In this regard, we propose different generative adversarial networks for cost-sensitive learning, ensemble learning, and mutual learning to deal with highly imbalanced imaging data.
We show evidence that the proposed approaches are applicable to different types of medical images of varied sizes on different applications of routine clinical tasks, such as disease classification and semantic segmentation. Our various implemented algorithms have shown outstanding results on different medical imaging challenges.
Simulating the impact of herbicide drift exposure on non-target terrestrial plant communities
(2019)
In Europe, almost half of the terrestrial landscape is used for agriculture. Thus, semi-natural habitats such as field margins are substantial for maintaining diversity in intensively managed farmlands. However, plants located at field margins are threatened by agricultural practices such as the application of pesticides within the fields. Pesticides are chemicals developed to control for undesired species within agricultural fields to enhance yields. The use of pesticides implies, however, effects on non-target organisms within and outside of the agricultural fields. Non-target organisms are organisms not intended to be sprayed or controlled for. For example, plants occurring in field margins are not intended to be sprayed, however, can be impaired due to herbicide drift exposure. The authorization of plant protection products such as herbicides requires risk assessments to ensure that the application of the product has no unacceptable effects on the environment. For non-target terrestrial plants (NTTPs), the risk assessment is based on standardized greenhouse studies on plant individual level. To account for the protection of plant populations and communities under realistic field conditions, i.e. extrapolating from greenhouse studies to field conditions and from individual-level to community-level, assessment factors are applied. However, recent studies question the current risk assessment scheme to meet the specific protection goals for non-target terrestrial plants as suggested by the European Food Safety Authority (EFSA). There is a need to clarify the gaps of the current risk assessment and to include suitable higher tier options in the upcoming guidance document for non-target terrestrial plants.
In my thesis, I studied the impact of herbicide drift exposure on NTTP communities using a mechanistic modelling approach. I addressed main gaps and uncertainties of the current risk assessment and finally suggested this modelling approach as a novel higher tier option in future risk assessments. Specifically, I extended the plant community model IBC-grass (Individual-based community model for grasslands) to reflect herbicide impacts on plant individuals. In the first study, I compared model predictions of short-term herbicide impacts on artificial plant communities with empirical data. I demonstrated the capability of the model to realistically reflect herbicide impacts. In the second study, I addressed the research question whether or not reproductive endpoints need to be included in future risk assessments to protect plant populations and communities. I compared the consequences of theoretical herbicide impacts on different plant attributes for long-term plant population dynamics in the community context. I concluded that reproductive endpoints only need to be considered if the herbicide effect is assumed to be very high. The endpoints measured in the current vegetative vigour and seedling emergence studies had high impacts for the dynamic of plant populations and communities already at lower effect intensities. Finally, the third study analysed long-term impacts of herbicide application for three different plant communities. This study highlighted the suitability of the modelling approach to simulate different communities and thus detecting sensitive environmental conditions.
Overall, my thesis demonstrates the suitability of mechanistic modelling approaches to be used as higher tier options for risk assessments. Specifically, IBC-grass can incorporate available individual-level effect data of standardized greenhouse experiments to extrapolate to community-level under various environmental conditions. Thus, future risk assessments can be improved by detecting sensitive scenarios and including worst-case impacts on non-target plant communities.
Introduction: Cystic fibrosis (CF) is a genetic disease which disrupts the function of an epithelial surface anion channel, CFTR (cystic fibrosis transmembrane conductance regulator). Impairment to this channel leads to inflammation and infection in the lung causing the majority of morbidity and mortality. However, CF is a multiorgan disease affecting many tissues, including vascular smooth muscle. Studies have revealed young people with cystic fibrosis lacking inflammation and infection still demonstrate vascular endothelial dysfunction, measured per flow-mediated dilation (FMD). In other disease cohorts, i.e. diabetic and obese, endurance exercise interventions have been shown improve or taper this impairment. However, long-term exercise interventions are risky, as well as costly in terms of time and resources. Nevertheless, emerging research has correlated the acute effects of exercise with its long-term benefits and advocates the study of acute exercise effects on FMD prior to longitudinal studies. The acute effects of exercise on FMD have previously not been examined in young people with CF, but could yield insights on the potential benefits of long-term exercise interventions.
The aims of these studies were to 1) develop and test the reliability of the FMD method and its applicability to study acute exercise effects; 2) compare baseline FMD and the acute exercise effect on FMD between young people with and without CF; and 3) explore associations between the acute effects of exercise on FMD and demographic characteristics, physical activity levels, lung function, maximal exercise capacity or inflammatory hsCRP levels.
Methods: Thirty young volunteers (10 people with CF, 10 non-CF and 10 non-CF active matched controls) between the ages of 10 and 30 years old completed blood draws, pulmonary function tests, maximal exercise capacity tests and baseline FMD measurements, before returning approximately 1 week later and performing a 30-min constant load training at 75% HRmax. FMD measurements were taken prior, immediately after, 30 minutes after and 1 hour after constant load training. ANOVAs and repeated measures ANOVAs were employed to explore differences between groups and timepoints, respectively. Linear regression was implemented and evaluated to assess correlations between FMD and demographic characteristics, physical activity levels, lung function, maximal exercise capacity or inflammatory hsCRP levels. For all comparisons, statistical significance was set at a p-value of α < 0.05.
Results: Young people with CF presented with decreased lung function and maximal exercise capacity compared to matched controls. Baseline FMD was also significantly decreased in the CF group (CF: 5.23% v non-CF: 8.27% v non-CF active: 9.12%). Immediately post-training, FMD was significantly attenuated (approximately 40%) in all groups with CF still demonstrating the most minimal FMD. Follow-up measurements of FMD revealed a slow recovery towards baseline values 30 min post-training and improvements in the CF and non-CF active groups 60 min post-training. Linear regression exposed significant correlations between maximal exercise capacity (VO2 peak), BMI and FMD immediately post-training.
Conclusion: These new findings confirm that CF vascular endothelial dysfunction can be acutely modified by exercise and will aid in underlining the importance of exercise in CF populations. The potential benefits of long-term exercise interventions on vascular endothelial dysfunction in young people with CF warrants further investigation.
The goal of this thesis is to broaden the empirical basis for a better, comprehensive understanding
of massive star evolution, star formation and feedback at low metallicity. Low metallicity massive stars are a key to understand the early universe. Quantitative information on metal-poor massive stars was sparse before. The quantitative spectroscopic studies of massive star populations associated with large-scale ISM structures were not performed at low metallicity before, but are important to investigate star-formation histories and feedback in detail. Much of this work relies on spectroscopic observations with VLT-FLAMES of ~500 OB stars in the Magellanic Clouds. When available, the optical spectroscopy was complemented by UV spectra from the HST, IUE, and FUSE archives. The two representative young stellar populations that have been studied are associated with the superbubble N 206 in the Large Magellanic Cloud (LMC) and with the supergiant shell SMC-SGS 1 in the Wing of the Small Magellanic Cloud (SMC), respectively. We performed spectroscopic analyses of the massive stars using the nonLTE Potsdam Wolf-Rayet (PoWR) model atmosphere code. We estimated the stellar, wind, and feedback parameters of the individual massive stars and established their statistical distributions.
The mass-loss rates of N206 OB stars are consistent with theoretical expectations for LMC metallicity. The most massive and youngest stars show nitrogen enrichment at their surface and are found to be slower rotators than the rest of the sample. The N 206 complex has undergone star formation episodes since more than 30 Myr, with a current star formation rate higher than average in the LMC. The spatial age distribution of stars across the complex possibly indicates triggered star formation due to the expansion of the superbubble. Three very massive, young Of stars in the region dominate the ionizing and mechanical feedback among hundreds of other OB stars in the sample. The current stellar wind feedback rate from the two WR stars in the complex is comparable to that released by the whole OB sample. We see only a minor fraction of this stellar wind feedback converted into X-ray emission. In this LMC complex, stellar winds and supernovae equally contribute to the total energy feedback, which eventually powered the central superbubble. However, the total energy input accumulated over the time scale of the superbubble significantly exceeds the observed energy content of the complex. The lack of energy along with the morphology of the complex suggests a leakage of hot gas from the superbubble.
With a detailed spectroscopic study of massive stars in SMC-SGS 1, we provide the stellar and wind parameters of a large sample of OB stars at low metallicity, including those in the lower mass-range. The stellar rotation velocities show a broad, tentatively bimodal distribution, with Be stars being among the fastest. A few very luminous O stars are found close to the main sequence, while all other, slightly evolved stars obey a strict luminosity limit. Considering additional massive stars in evolved stages, with published parameters and located all over the SMC, essentially confirms this picture. The comparison with single-star evolutionary tracks suggests a dichotomy in the fate of massive stars in the SMC. Only stars with an initial mass below 30 solar masses seem to evolve from the main sequence to the cool side of the HRD to become a red supergiant and to explode as type II-P supernova. In contrast, more massive stars appear to stay always hot and might evolve quasi chemically homogeneously, finally collapsing to relatively massive black holes. However, we find no indication that chemical mixing is correlated with rapid rotation. We measured the key parameters of stellar feedback and established the links between the rates of star formation and supernovae. Our study demonstrates that in metal-poor environments stellar feedback is dominated by core-collapse supernovae in combination with winds and ionizing radiation supplied by a few of the most massive stars. We found indications of the stochastic mode of star formation, where the resulting stellar population is fully capable of producing large-scale structures such as the supergiant shell SMC-SGS 1 in the Wing. The low level of feedback in metal-poor stellar populations allows star formation episodes to persist over long timescales.
Our study showcases the importance of quantitative spectroscopy of massive stars with adequate stellar-atmosphere models in order to understand star-formation, evolution, and feedback. The stellar population analyses in the LMC and SMC make us understand that massive stars and their impact can be very different depending on their environment. Obviously, due to their different metallicity, the massive stars in the LMC and the SMC follow different evolutionary paths. Their winds differ significantly, and the key feedback agents are different. As a consequence, the star formation can proceed in different modes.
Einleitung
Die Implantation einer Knie- oder Hüft-Totalendoprothese (TEP) ist eine der häufigsten operativen Eingriffe. Im Anschluss an die Operation und die postoperative Rehabilitation stellt die Bewegungstherapie einen wesentlichen Bestandteil der Behandlung zur Verbesserung der Gelenkfunktion und der Lebensqualität dar. In strukturschwachen Gebieten werden entsprechende Angebote nur in unzureichender Dichte vorgehalten. Zudem zeichnet sich ein flächendeckender Fachkräftemangel im Bereich der Physiotherapie ab. Die Tele-Nachsorge bietet daher einen innovativen Ansatz für die postrehabilitative Versorgung der Patienten. Das Ziel der vorliegenden Untersuchung war die Überprüfung der Wirksamkeit einer interaktiven Tele-Nachsorgeintervention für Patienten mit Knie- oder Hüft-TEP im Vergleich zur herkömmlichen Versorgung (usual care). Dazu wurden die Funktionalität und die berufliche Wiedereingliederung untersucht.
Methode
Zwischen August 2016 und August 2017 wurden 111 Patienten (54,9 ± 6,8 Jahre, 54,3 % weiblich) zu Beginn ihrer stationären Anschlussheilbehandlung nach Implantation einer Knie- oder Hüft-TEP in diese randomisiert, kontrolliert, multizentrische Studie eingeschlossen. Nach Entlassung aus der orthopädischen Anschlussrehabilitation (Baseline) führte die Interventionsgruppe (IG) ein dreimonatiges interaktives Training über ein Telerehabilitationssystem durch. Hierfür erstellte ein betreuender Physiotherapeut einen individuellen Trainingsplan aus 38 Übungen zur Verbesserung der Kraft sowie der posturalen Kontrolle. Zur Anpassung des Trainingsplans übermittelte das System dem Physiotherapeuten Daten zur Quantität sowie zur Qualität des Trainings. Die Kontrollgruppe (KG) konnte die herkömmlichen Versorgungsangebote nutzen. Zur Beurteilung der Wirksamkeit der Intervention wurde die Differenz der Verbesserung im 6MWT zwischen der IG und der KG nach drei Monaten als primärer Endpunkt definiert. Als sekundäre Endpunkte wurden die Return-to-Work-Rate sowie die funktionelle Mobilität mittels des Stair Ascend Tests, des Five-Times-Sit-to-Stand Test und des Timed Up and Go Tests untersucht. Weiterhin wurden die gesundheitsbezogene Lebensqualität mit dem Short-Form 36 (SF-36) und die gelenkbezogenen Einschränkungen mit dem Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) evaluiert. Der primäre und die sekundären Endpunkte wurden anhand von baseline-adjustierten Kovarianzanalysen im intention-to-treat-Ansatz ausgewertet. Zusätzlich wurde die Teilnahme an Nachsorgeangeboten und die Adhärenz der Interventionsgruppe an der Tele-Nachsorge erfasst und evaluiert.
Ergebnisse
Zum Ende der Intervention wiesen beide Gruppen einen statistisch signifikanten Anstieg ihrer 6MWT Strecke auf (p < 0,001). Zu diesem Zeitpunkt legten die Teilnehmer der IG im Mittel 530,8 ± 79,7 m, die der KG 514,2 ± 71,2 m zurück. Dabei betrug die Differenz der Verbesserung der Gehstrecke in der IG 88,3 ± 57,7 m und in der KG 79,6 ± 48,7 m. Damit zeigt der primäre Endpunkt keine signifikanten Gruppenunterschiede (p = 0,951). Bezüglich der beruflichen Wiedereingliederung konnte jedoch eine signifikant höhere Rate in der IG (64,6 % versus 46,2 %; p = 0,014) festgestellt werden. Für die sekundären Endpunkte der funktionellen Mobilität, der Lebensqualität und der gelenkbezogenen Beschwerden belegen die Ergebnisse eine Gleichwertigkeit beider Gruppen zum Ende der Intervention.
Schlussfolgerung
Die telemedizinisch assistierte Bewegungstherapie für Knie- oder Hüft-TEP Patienten ist der herkömmlichen Versorgung zur Nachsorge hinsichtlich der erzielten Verbesserungen der funktionellen Mobilität, der gesundheitsbezogenen Lebensqualität und der gelenkbezogenen Beschwerden gleichwertig. In dieser Patientenpopulation ließen sich klinisch relevante Verbesserungen unabhängig von der Form der Bewegungstherapie erzielen. Im Hinblick auf die berufliche Wiedereingliederung zeigte sich eine signifikant höhere Rate in der Interventionsgruppe. Die telemedizinisch assistierte Bewegungstherapie scheint eine geeignete Versorgungsform der Nachsorge zu sein, die orts- und zeitunabhängig durchgeführt werden kann und somit den Bedürfnissen berufstätiger Patienten entgegenkommt und in den Alltag der Patienten integriert werden kann. Die Tele-Nachsorge sollte daher als optionale und komplementäre Form der postrehabilitativen Nachsorge angeboten werden. Auch im Hinblick auf den zunehmenden Fachkräftemangel im Bereich der Physiotherapie und bestehende Versorgungslücken in strukturschwachen Gebieten kann der Einsatz der Tele-Nachsorge innovative und bedarfsgerechte Lösungsansätze bieten.
Predator-prey interactions provide central links in food webs. These interaction are directly or indirectly impacted by a number of factors. These factors range from physiological characteristics of individual organisms, over specifics of their interaction to impacts of the environment. They may generate the potential for the application of different strategies by predators and prey. Within this thesis, I modelled predator-prey interactions and investigated a broad range of different factors driving the application of certain strategies, that affect the individuals or their populations. In doing so, I focused on phytoplankton-zooplankton systems as established model systems of predator-prey interactions.
At the level of predator physiology I proposed, and partly confirmed, adaptations to fluctuating availability of co-limiting nutrients as beneficial strategies. These may allow to store ingested nutrients or to regulate the effort put into nutrient assimilation. We found that these two strategies are beneficial at different fluctuation frequencies of the nutrients, but may positively interact at intermediate frequencies. The corresponding experiments supported our model results. We found that the temporal structure of nutrient fluctuations indeed has strong effects on the juvenile somatic growth rate of {\itshape Daphnia}.
Predator colimitation by energy and essential biochemical nutrients gave rise to another physiological strategy. High-quality prey species may render themselves indispensable in a scenario of predator-mediated coexistence by being the only source of essential biochemical nutrients, such as cholesterol. Thereby, the high-quality prey may even compensate for a lacking defense and ensure its persistence in competition with other more defended prey species.
We found a similar effect in a model where algae and bacteria compete for nutrients. Now, being the only source of a compound that is required by the competitor (bacteria) prevented the competitive exclusion of the algae. In this case, the essential compounds were the organic carbon provided by the algae. Here again, being indispensable served as a prey strategy that ensured its coexistence.
The latter scenario also gave rise to the application of the two metabolic strategies of autotrophy and heterotrophy by algae and bacteria, respectively. We found that their coexistence allowed the recycling of resources in a microbial loop that would otherwise be lost. Instead, these resources were made available to higher trophic levels, increasing the trophic transfer efficiency in food webs.
The predation process comprises the next higher level of factors shaping the predator-prey interaction, besides these factors that originated from the functioning or composition of individuals. Here, I focused on defensive mechanisms and investigated multiple scenarios of static or adaptive combinations of prey defense and predator offense. I confirmed and extended earlier reports on the coexistence-promoting effects of partially lower palatability of the prey community. When bacteria and algae are coexisting, a higher palatability of bacteria may increase the average predator biomass, with the side effect of making the population dynamics more regular. This may facilitate experimental investigations and interpretations. If defense and offense are adaptive, this allows organisms to maximize their growth rate. Besides this fitness-enhancing effect, I found that co-adaptation may provide the predator-prey system with the flexibility to buffer external perturbations.
On top of these rather internal factors, environmental drivers also affect predator-prey interactions. I showed that environmental nutrient fluctuations may create a spatio-temporal resource heterogeneity that selects for different predator strategies. I hypothesized that this might favour either storage or acclimation specialists, depending on the frequency of the environmental fluctuations.
We found that many of these factors promote the coexistence of different strategies and may therefore support and sustain biodiversity. Thus, they might be relevant for the maintenance of crucial ecosystem functions that also affect us humans. Besides this, the richness of factors that impact predator-prey interactions might explain why so many species, especially in the planktonic regime, are able to coexist.
The central motivation of the thesis was to provide possible solutions and concepts to improve the performance (e.g. activity and selectivity) of electrochemical N2 reduction reaction (NRR). Given that porous carbon-based materials usually exhibit a broad range of structural properties, they could be promising NRR catalysts. Therefore, the advanced design of novel porous carbon-based materials and the investigation of their application in electrocatalytic NRR including the particular reaction mechanisms are the most crucial points to be addressed. In this regard, three main topics were investigated. All of them are related to the functionalization of porous carbon for electrochemical NRR or other electrocatalytic reactions.
In chapter 3, a novel C-TixOy/C nanocomposite has been described that has been obtained via simple pyrolysis of MIL-125(Ti). A novel mode for N2 activation is achieved by doping carbon atoms from nearby porous carbon into the anion lattice of TixOy. By comparing the NRR performance of M-Ts and by carrying out DFT calculations, it is found that the existence of (O-)Ti-C bonds in C-doped TixOy can largely improve the ability to activate and reduce N2 as compared to unoccupied OVs in TiO2. The strategy of rationally doping heteroatoms into the anion lattice of transition metal oxides to create active centers may open many new opportunities beyond the use of noble metal-based catalysts also for other reactions that require the activation of small molecules as well.
In chapter 4, a novel catalyst construction composed of Au single atoms decorated on the surface of NDPCs was reported. The introduction of Au single atoms leads to active reaction sites, which are stabilized by the N species present in NDPCs. Thus, the interaction within as-prepared AuSAs-NDPCs catalysts enabled promising performance for electrochemical NRR. For the reaction mechanism, Au single sites and N or C species can act as Frustrated Lewis pairs (FLPs) to enhance the electron donation and back-donation process to activate N2 molecules. This work provides new opportunities for catalyst design in order to achieve efficient N2 fixation at ambient conditions by utilizing recycled electric energy.
The last topic described in chapter 5 mainly focused on the synthesis of dual heteroatom-doped porous carbon from simple precursors. The introduction of N and B heteroatoms leads to the construction of N-B motives and Frustrated Lewis pairs in a microporous architecture which is also rich in point defects. This can improve the strength of adsorption of different reactants (N2 and HMF) and thus their activation. As a result, BNC-2 exhibits a desirable electrochemical NRR and HMF oxidation performance. Gas adsorption experiments have been used as a simple tool to elucidate the relationship between the structure and catalytic activity. This work provides novel and deep insights into the rational design and the origin of activity in metal-free electrocatalysts and enables a physically viable discussion of the active motives, as well as the search for their further applications.
Throughout this thesis, the ubiquitous problems of low selectivity and activity of electrochemical NRR are tackled by designing porous carbon-based catalysts with high efficiency and exploring their catalytic mechanisms. The structure-performance relationships and mechanisms of activation of the relatively inert N2 molecules are revealed by either experimental results or DFT calculations. These fundamental understandings pave way for a future optimal design and targeted promotion of NRR catalysts with porous carbon-based structure, as well as study of new N2 activation modes.
Ficción herética
(2019)
La metáfora de la «isla» en la narrativa cubana contemporánea engloba toda una serie de complejidades simbólicas dependientes de la vivencia del espacio y el tiempo. Su potencial visual se manifiesta u oculta las propias vivencias insulares de los escritores cubanos. En los últimos 30 años en Cuba, los fenómenos políticos económicos y sociales han modificado categóricamente la percepción y la configuración del plano social e individual frente a las exigencias globales (Fornet 2006, Rojas 1999, 2002, 2006). Se ha confirmado una sensación de acinesia e ingravidez (Casamayor 2013) y se ha presentado una actitud «herética» por parte de los narradores cubanos, quienes se confrontan con las ideas de la postmodernidad, lo postsoviético o postutópico, reafirmando así una sensibilidad presentista (Guerrero 2016). Estos autores presentan resonancias y reivindicaciones de los imaginarios insulares de autores y de tradiciones estéticas dentro y fuera de la isla como José Lezama Lima, Virgilio Piñera, Guillermo Cabrera Infante, Reinaldo Arenas y Severo Sarduy). El análisis de las écfrasis insulares permite examinar las dinámicas de representación y de sentido: disimulación: la disimulación, la anamorfosis y la trompe l’oeil (Sarduy 1981). La novela Tuyo es el reino (1998) de Abilio Estévez es un modelo desde el que se localizará las relaciones de sentido entre canon literario y los referentes socioculturales de las variaciones somatopológicas de la isla en la narrativa cubana actual: Ena Lucía Portela, Atilio Caballero, Antonio José Ponte, Daniel Díaz Mantilla, Emerio Medina, Orlando Luis Pardo, Anisley Negrin y Ahmel Echeverría, entre otros.
Carbonate-rich silicate and carbonate melts play a crucial role in deep Earth magmatic processes and their melt structure is a key parameter, as it controls physical and chemical properties. Carbonate-rich melts can be strongly enriched in geochemically important trace elements. The structural incorporation mechanisms of these elements are difficult to study because such melts generally cannot be quenched to glasses, which are usually employed for structural investigations. This thesis investigates the influence of CO2 on the local environments of trace elements contained in silicate glasses with variable CO2 concentrations as well as in silicate and carbonate melts. The compositions studied include sodium-rich peralkaline silicate melts and glasses and carbonate melts similar to those occurring naturally at Oldoinyo Lengai volcano, Tanzania.
The local environments of the three elements yttrium (Y), lanthanum (La) and strontium (Sr) were investigated in synthesized glasses and melts using X-ray absorption fine structure (XAFS) spectroscopy. Especially extended X-ray absorption fine structure spectroscopy (EXAFS) provides element specific information on local structure, such as bond lengths, coordination numbers and the degree of disorder. To cope with the enhanced structural disorder present in glasses and melts, EXAFS analysis was based on fitting approaches using an asymmetric distribution function as well as a correlation model according to bond valence theory. Firstly, silicate glasses quenched from high pressure/temperature melts with up to 7.6 wt % CO2 were investigated. In strongly and extremely peralkaline glasses the local structure of Y is unaffected by the CO2 content (with oxygen bond lengths of ~ 2.29 Å). Contrary, the bond lengths for Sr-O and La-O increase with increasing CO2 content in the strongly peralkaline glasses from ~ 2.53 to ~ 2.57 Å and from ~ 2.52 to ~ 2.54 Å, respectively, while they remain constant in extremely peralkaline glasses (at ~ 2.55 Å and 2.54 Å, respectively). Furthermore, silicate and unquenchable carbonate melts were investigated in-situ at high pressure/temperature conditions (2.2 to 2.6 GPa, 1200 to 1500 °C) using a Paris-Edinburgh press. A novel design of the pressure medium assembly for this press was developed, which features increased mechanical stability as well as enhanced transmittance at relevant energies to allow for low content element EXAFS in transmission. Compared to glasses the bond lengths of Y-O, La-O and Sr-O are elongated by up to + 3 % in the melt and exhibit higher asymmetric pair distributions. For all investigated silicate melt compositions Y-O bond lengths were found constant at ~ 2.37 Å, while in the carbonate melt the Y-O length increases slightly to 2.41 Å. The La-O bond lengths in turn, increase systematically over the whole silicate – carbonate melt joint from 2.55 to 2.60 Å. Sr-O bond lengths in melts increase from ~ 2.60 to 2.64 Å from pure silicate to silicate-bearing carbonate composition with constant elevated bond length within the carbonate region.
For comparison and deeper insight, glass and melt structures of Y and Sr bearing sodium-rich silicate to carbonate compositions were simulated in an explorative ab initio molecular dynamics (MD) study. The simulations confirm observed patterns of CO2-dependent local changes around Y and Sr and additionally provide further insights into detailed incorporation mechanisms of the trace elements and CO2. Principle findings include that in sodium-rich silicate compositions carbon either is mainly incorporated as a free carbonate-group or shares one oxygen with a network former (Si or [4]Al) to form a non-bridging carbonate. Of minor importance are bridging carbonates between two network formers. Here, a clear preference for two [4]Al as adjacent network formers occurs, compared to what a statistical distribution would suggest. In C-bearing silicate melts minor amounts of molecular CO2 are present, which is almost totally dissolved as carbonate in the quenched glasses.
The combination of experiment and simulation provides extraordinary insights into glass and melt structures. The new data is interpreted on the basis of bond valence theory and is used to deduce potential mechanisms for structural incorporation of investigated elements, which allow for prediction on their partitioning behavior in natural melts. Furthermore, it provides unique insights into the dissolution mechanisms of CO2 in silicate melts and into the carbonate melt structure. For the latter, a structural model is suggested, which is based on planar CO3-groups linking 7- to 9-fold cation polyhedra, in accordance to structural units as found in the Na-Ca carbonate nyerereite. Ultimately, the outcome of this study contributes to rationalize the unique physical properties and geological phenomena related to carbonated silicate-carbonate melts.
Synchronization – the adjustment of rhythms among coupled self-oscillatory systems – is a fascinating dynamical phenomenon found in many biological, social, and technical systems.
The present thesis deals with synchronization in finite ensembles of weakly coupled self-sustained oscillators with distributed frequencies.
The standard model for the description of this collective phenomenon is the Kuramoto model – partly due to its analytical tractability in the thermodynamic limit of infinitely many oscillators. Similar to a phase transition in the thermodynamic limit, an order parameter indicates the transition from incoherence to a partially synchronized state. In the latter, a part of the oscillators rotates at a common frequency. In the finite case, fluctuations occur, originating from the quenched noise of the finite natural frequency sample.
We study intermediate ensembles of a few hundred oscillators in which fluctuations are comparably strong but which also allow for a comparison to frequency distributions in the infinite limit.
First, we define an alternative order parameter for the indication of a collective mode in the finite case. Then we test the dependence of the degree of synchronization and the mean rotation frequency of the collective mode on different characteristics for different coupling strengths.
We find, first numerically, that the degree of synchronization depends strongly on the form (quantified by kurtosis) of the natural frequency sample and the rotation frequency of the collective mode depends on the asymmetry (quantified by skewness) of the sample. Both findings are verified in the infinite limit.
With these findings, we better understand and generalize observations of other authors. A bit aside of the general line of thoughts, we find an analytical expression for the volume contraction in phase space.
The second part of this thesis concentrates on an ordering effect of the finite-size fluctuations. In the infinite limit, the oscillators are separated into coherent and incoherent thus ordered and disordered oscillators. In finite ensembles, finite-size fluctuations can generate additional order among the asynchronous oscillators. The basic principle – noise-induced synchronization – is known from several recent papers. Among coupled oscillators, phases are pushed together by the order parameter fluctuations, as we on the one hand show directly and on the other hand quantify with a synchronization measure from directed statistics between pairs of passive oscillators.
We determine the dependence of this synchronization measure from the ratio of pairwise natural frequency difference and variance of the order parameter fluctuations. We find a good agreement with a simple analytical model, in which we replace the deterministic fluctuations of the order parameter by white noise.