Novel ecosystems emerge from alterations to historic abiotic regimes and contain new species combinations. Everglades canals offer an opportunity to understand the function of novel habitat for native and non-native fishes and how novel conditions in turn influence distribution, abundance and assembly patterns. I examined native and non-native fish assemblages collected across a gradient in novelty, defined by the loss of wetland connectivity and habitat complexity. As novelty increased, native species richness and abundance strongly declined, and the contribution of non-natives increased. Community structure vastly differed among canals and was strongly influenced by spatial factors and secondarily by hydrological factors. Natives and non-natives had opposing responses to key hydrologic and habitat parameters. This study represents the first comprehensive assessment of Everglades canal fishes, providing insight into the factors influencing native and non-native abundance and assembly patterns and contributing to our understanding of this novel but permanent habitat.

Dans cette thèse, nous nous intéressons essentiellement à trois sujets : les inégalités fonctionnelles à contenu probabiliste, les limites hydrodynamiques pour les systèmes de spins continus en interaction et la discrétisation des équations différentielles stochastiques. Ce document, outre l'introduction, comporte trois parties. La première s'intéresse aux inégalités fonctionnelles, et notamment aux inégalités de Sobolev logarithmiques, pour les mesures canoniques, ainsi qu'aux limites hydrodynamiques pour les systèmes des spins continus. La convergence vers la limite hydrodynamique pour plusieurs variantes du modèle de Ginzburg--Landau équipé de la dynamique de Kawasaki y est obtenue, avec notamment des bornes quantitatives en le nombre de spins. On y étudie également la convergence de l'entropie microscopique vers l'entropie hydrodynamique. La deuxième partie étudie les liens entre flots gradients dans les espaces de mesures de probabilités et grandes déviations pour les suites de lois de solutions d'équations différentielles stochastiques. On y obtient l'équivalence entre le principe de grandes déviations et la Gamma-Convergence d'une suite de fonctionnelles apparaissant dans la formulation en flots gradients du flot de marginales des lois des solutions des équations différentielles stochastiques. Comme application de ce principe, on obtient les grandes déviations par rapport à la limite hydrodynamique pour deux variantes du modèle de Ginzburg--Landau. La troisième partie concerne la discrétisation des équations différentielles stochastiques. On y prouve une inégalité transport-Entropie pour la loi du schéma d'Euler explicite. Cette inégalité implique des bornes sur les intervalles de confiance pour l'estimation de quantités de la forme $\mathbb{E}[f(X_T)]$. On y étudie également l'erreur de discrétisation pour l'évaluation des coefficients de transport avec l'algorithme MALA (qui est une combinaison du schéma d'Euler explicite et de l'algorithme de Metropolis--Hastings). / In this thesis, we are mainly interested in three topics : functional inequalities and their probabilistic aspects, hydrodynamic limits for interacting continuous spin systems and discretizations of stochastic differential equations. This document, in addition to a general introduction (written in French), contains three parts. The first part deals with functional inequalities, especially logarithmic Sobolev inequalities, for canonical ensembles, and with hydrodynamic limits for continuous spin systems. We prove convergence to the hydrodynamic limit for several variants of the Ginzburg--Landau model endowed with Kawasaki dynamics, with quantitative bounds in the number of spins. We also study convergence of the microscopic entropy to its hydrodynamic counterpart. In the second part, we study links between gradient flows in spaces of probability measures and large deviations for sequences of laws of solutions to stochastic differential equations. We show that the large deviations principle is equivalent to the Gamma--Convergence of a sequence of functionals that appear in the gradient flow formulation of the flow of marginals of the laws of the diffusion processes. As an application of this principle, we obtain large deviations from the hydrodynamic limit for two variants of the Ginzburg-Landau model. The third part deals with the discretization of stochastic differential equations. We prove a transport-Entropy inequality for the law of the explicit Euler scheme. This inequality implies bounds on the confidence intervals for quantities of the form $\mathbb{E}[f(X_T)]$. We also study the discretization error for the evaluation of transport coefficients with the Metropolis-Adjusted Langevin algorithm (which is a combination of the explicit Euler scheme and the Metropolis algorithm).

The stratification of bird species along elevational gradients is widely reported, with montane bird communities typically characterized by distinctive species occurring in relatively small and isolated populations; as such, these species are the subject of considerable interest to ecologists and conservationists. The stratification of species along elevation is largely attributed to compressed climatic zonation. Recent evidence that bird species are shifting up in elevation has fueled speculation that these species are tracking their climactic niches in response to climate change. However, there is also evidence plant communities are shifting in elevation, presenting a potential additional mechanism explaining changes observed in the bird community. Uncertainty as to the degree to which climate directly influences bird abundance versus the degree to which climate indirectly influences bird abundance via habitat composition and structure represents a key impediment to understanding the ecology of these species in montane environments. To address this question I measured species abundance, habitat characteristics, and temperature at 150 survey points located along 15 elevational transects in the Presidential Mountains of New Hampshire in the summers of 2014 and 2015. I used N-mixture models to correct for imperfect detection of species and structural equation models, incorporating abundance, habitat, temperature, and precipitation derived from a downscaled regional dataset to assign variation to the direct and indirect effects of climate upon birds. Analysis of 21 species revealed species-specific patterns on how climate exerts direct effects and indirect effects mediated by forest composition and structure on bird abundance. This work represents an important contribution to the ecological understanding of the pathways by which climate influences bird abundance. Finding that 62% of species experience both direct and indirect effects of climate, with 62% experiencing stronger direct than indirect effects, these results underscore the vulnerability of these species to climate change. With 81% of species found to experience indirect effects of climate via forests, these findings indicate great conservation value of maintaining forest habitat amidst climate change. Overall, this information will facilitate the refinement of predictive models of the abundance of montane bird species and represents an approach that will advance future investigations of climate effects in the Northern Appalachians and other systems.

Gegenstand der Arbeit ist die PFG NMR (nuclear magnetic resonance with pulsed field gradients), wobei speziell die apparativen und experimentellen Bedingungen untersucht werden, welche sich durch die Verwendung ultra-hoher gepulster magnetischer Feldgradienten von bis zu 35T/m ergeben. Motiv für die Arbeit ist die Untersuchung von Diffusionserscheinungen in mikroporösen Wirtssystemen mit inneren magnetischen Feldgradienten oder/und kurzen T2-Relaxationzeiten. Nach Zusammenstellung der notwendigen Werkzeuge zur mathematischen Beschreibung von PFG NMR-Experimenten werden die aus der Literatur bekannten Impulssequenzen kritisch untersucht und durch eigene Weiterentwicklungen ergänzt. Für wichtige PFG NMR-Impulssequenzen wird eine verallgemeinerte Schreibweise vorgestellt und auf beliebige Formen der gepulsten magnetischen Feldgradienten ausgedehnt. Weiterhin werden Störeinflüsse auf das PFG NMR-Experiment untersucht und zunächst in allgemeiner Form Möglichkeiten zu deren Beseitigung bzw. Unterdrückung dargestellt. Die so gewonnenen Erkenntnisse fanden konkrete Anwendung bei der Konzeption und dem Bau des PFG NMR-Spektrometers Fegris 400 NT. Dieses Gerät wird, soweit es den Gegenstand der Arbeit berührt, ebenfalls beschrieben und in der Anlage dokumentiert. Abschließend sind einige Untersuchungen, die mit dem Fegris 400 NT durchgeführt wurden und in der dargestellten Form erst mit diesem Gerät möglich waren, kurz skizziert, wobei für weitergehende Informationen auf die entsprechenden Veröffentlichungen verwiesen wird.

In Finance, the use of Automated Trading Systems (ATS) on markets is growing every year and the trades generated by an algorithm now account for most of orders that arrive at stock exchanges (Kissell, 2020). Historically, these systems were based on advanced statistical methods and signal processing designed to extract trading signals from financial data. The recent success of Machine Learning has attracted the interest of the financial community. Reinforcement Learning is a subcategory of machine learning and has been broadly applied by investors and researchers in building trading systems (Kissell, 2020). In this thesis, we address the issue that deep reinforcement learning may be susceptible to sampling errors and over-fitting and propose a robust deep reinforcement learning method that integrates techniques from reinforcement learning and robust optimization. We back-test and compare the performance of the developed algorithm, Robust DDPG, with UBAH (Uniform Buy and Hold) benchmark and other RL algorithms and show that the robust algorithm of this research can reduce the downside risk of an investment strategy significantly and can ensure a safer path for the investor’s portfolio value.

Diploma thesis deals with usage of the diffusive gradients in thin films technique (DGT) for the determination of labile metal species in the surface water and sediments of Brno water reservoir. Sediment and water samples were collected on the selected sides of Brno water reservoir on September and October 2008. The DGT technique was used for determination of depth profiles of Fe, Mn, Pb, Cd, Zn, Cu, Ni and Al. The DGT probes with three different thicknesses of diffusive layer were applied into the sediment samples. The obtained results gave the information about release of metals from solid phase into the pore water of sediment. The concentrations of Fe, Mn, Pb and Cd in sediments were determined by atomic absorption spectrometry after microwave decomposition. The DGT technique was used also for determination of Fe, Mn, Pb and Cd in surface water from Brno water reservoir.

The goal of this thesis is the utilization of computer vision methods, in a way that will lead to detection and identification of traffic signs in an image. The final application is to analyze video feed from a video camcorder placed in a vehicle. With focus placed on effective utilization of computer resources in order to achieve real time identification of signs in a video stream.

This study aims to modify Vision Transformer (ViT) to achieve higher accuracy. ViT is a model used in computer vision to, among other things, classify images. By applying ViT to the MNIST data set, an accuracy of approximately 98% is achieved. ViT is modified by implementing a method called Histogram of Oriented Gradients (HOG) in two different ways. The results show that the first approach with HOG gives an accuracy of 98,74% (setup 1) and the second approach gives an accuracy of 96,87% (patch size 4x4 pixels). The study shows that when HOG is applied on the entire image, a better accuracy is obtained. However, no systematic optimization has taken place, which makes it difficult to draw conclusions with certainty.

The present study investigates experimentally the laminar mixing and heat transfer of a range of helical ribbon and anchor impellers for both Newtonian and inelastic non-Newtonian fluids. The work also correlates the experimental data empirically in the form of dimensionless groups. In order to estimate the relative importance and the effect of all the geometrical parameters on the mixing power and heat transfer, data from the published literature sources will be utilized and combined with the results from this study. Thus, reliable empirical correlations will be obtained which are applicable over the widest range of operating conditions. The study also investigates the ablity of the various impellers to level out temerature distributions. The measurement of these temperature gradients and the impeller power requirements gives a measure of the mixing efficiency of the impeller used. / Science Research Council

The variation in community composition among sites is often used to gain insight into the processes of plant community assembly. In this study, we looked for evidence of environmentally and spatially mediated community assembly mechanisms in a temperate forest in eastern North America. To test this, we measured, identified and mapped all woody stems ≥1 cm in diameter at breast height (DBH) in 12 ha of a 20 ha forest plot. We used principal coordinates of neighbor matrices (PCNM) to obtain variables that modelled spatial processes (eg. dispersal, drift) at the community level. Topographic variables (slope, elevation, convexity, aspect) were used to model environmental conditions. Variation partitioning was used to isolate the unique and shared effects of topographic and spatial variables on community composition. We were also interested in studying how associations with the environment change with tree size. For this we assessed the abundance of a subset of focal species in response to topography as well as human disturbance. Species abundance were divided into three stem size classes: small (< 5 cm), medium (≥5 cm and < 15), and big (≥15 cm). We found that topography and space jointly explained 63% of the variation in community composition. This variation was almost entirely spatially structured with the component of pure topography only contributing 1% to the total explained variation. A redundancy analysis showed that slope and elevation were the most important topographic variables structuring the distribution of trees. The focal species had largely independent distributions across the environmental gradients and three of the five species showed within-species differences associated with size class effects on the relationship with topography and human disturbance. The implications of these results are relevant to conservation efforts and suggest that large contiguous areas of heterogeneous environments are essential in maintaining biodiversity. / Thesis / Master of Science (MSc)