This thesis aims to develop a new model for stress-testing financial portfolios using Extreme Value Theory (EVT) and General Adversarial Networks (GANs). The current practice of risk management relies on mathematical or historical models, such as Value-at-Risk and expected shortfall. The problem with historical models is that the data which is available for very extreme events is limited, and therefore we need a method to interpolate and extrapolate beyond the available range. EVT is a statistical framework that analyzes extreme events in a distribution and allows such interpolation and extrapolation, and GANs are machine-learning techniques that generate synthetic data. The combination of these two areas can generate more realistic stress-testing scenarios to help financial institutions manage potential risks better. The goal of this thesis is to develop a new model that can handle complex dependencies and high-dimensional inputs with different kinds of assets such as stocks, indices, currencies, and commodities and can be used in parallel with traditional risk measurements. The evtGAN algorithm shows promising results and is able to mimic actual distributions, and is also able to extrapolate data outside the available data range. / Detta examensarbete handlar om att utveckla en ny modell för stresstestning av finansiella portföljer med hjälp av extremvärdesteori (EVT) och Generative Adversarial Networks (GAN). Dom modeller för riskhantering som används idag bygger på matematiska eller historiska modeller, som till exempel Value-at-Risk och Expected Shortfall. Problemet med historiska modeller är att det finns begränsat med data för mycket extrema händelser. EVT är däremot en del inom statistisk som analyserar extrema händelser i en fördelning, och GAN är maskininlärningsteknik som genererar syntetisk data. Genom att kombinera dessa två områden kan mer realistiska stresstestscenarier skapas för att hjälpa finansiella institutioner att bättre hantera potentiella risker. Målet med detta examensarbete är att utveckla en ny modell som kan hantera komplexa beroenden i högdimensionell data med olika typer av tillgångar, såsom aktier, index, valutor och råvaror, och som kan användas parallellt med traditionella riskmått. Algoritmen evtGAN visar lovande resultat och kan imitera verkliga fördelningar samt extrapolera data utanför tillgänglig datamängd.

This dissertation analyzes the impact of climate, and atmospheric carbon dioxide (CO2) on crop yields and grain transportation. The analysis of crop yields endeavors to advance the literature by statistically estimating the effects of atmospheric carbon dioxide (CO2) on observed crop yields. This is done using an econometric model estimated over pooled historical data for 1950-2009 and data from the free air CO2 enrichment experiments. The main findings are: 1) yields of soybeans, cotton, and wheat directly respond to the elevated CO2, while yields of corn and sorghum do not; 2) the effect of crop technological progress on mean yields is non-linear; 3) ignoring atmospheric CO2 in an econometric model of crop yield likely leads to overestimates of the pure effects of climate change and technological progress on crop yields; and 4) average climate conditions and climate variability contribute in a statistically significant way to average crop yields and their variability. To examine climate change impacts on grain transportation flows, this study employs two modeling systems, a U.S. agricultural sector model and an international grain transportation model, with linked inputs/outputs. The main findings are that under climate change: 1) the excess supply of corn and soybeans generally increases in Northern U.S. regions, while it declines in Central and Southern regions; 2) the Corn Belt, the largest producer of corn in the U.S., is anticipated to ship less corn; 3) the importance of lower Mississippi River ports, the largest current destination for U.S. grain exports, diminishes under the climate change cases, whereas the role of Pacific Northwest ports, Great Lakes ports, and Atlantic ports is projected to increase; 4) the demand for grain shipment via rail and truck rises, while demand for barge transport drops.

Ces travaux de thèse rapportent l’étude des instabilités non-linéaires et des évènements extrêmesse développant lors de la propagation guidée d’un champ électromagnétique au sein de fibresoptiques. Après un succinct rappel des divers processus linéaires et non-linéaires menant à lagénération de super continuum optique, nous montrons que le spectre de celui-ci peut présenterde larges fluctuations, incluant la formation d’événements extrêmes, dont les propriétés statistiqueset l’analogie avec les vagues scélérates hydrodynamiques sont abordées en détail. Nous présentonsune preuve de principe de l’application de ces fluctuations spectrales à la génération de nombres etde marches aléatoires et identifions le phénomène d’instabilité de modulation, ayant lieu lors de laphase initiale d’expansion spectrale du super continuum, comme principale contribution à la formationd’événements extrêmes. Ce mécanisme est étudié numériquement et analytiquement, en considérantune catégorie de solutions exactes de l’équation de Schrödinger non-linéaire présentant descaractéristiques de localisations singulières. Les résultats obtenus sont vérifiés expérimentalement,notamment grâce à un système de caractérisation spectrale en temps réel et à l’utilisation conjointede métriques statistiques innovantes (ex : cartographie de corrélations spectrales). L’excellent accordentre simulations et expériences a permis de valider les prédictions théoriques et d’accéder àune meilleure compréhension des dynamiques complexes inhérentes à la propagation non-linéaired’impulsions optiques. / This thesis reports the study of nonlinear instabilities and extreme events occurring during the guidedpropagation of an electromagnetic field into optical fibers. After a short overview of the various linearand nonlinear processes leading to optical supercontinuum generation, we show that its spectrumcan exhibit large fluctuations, including the formation of extreme events, whose statistical propertiesas well as hydrodynamic rogue waves analogy are studied in detail. We provide a proof of principle ofusing these spectral fluctuations for random number and random walk generation and identify modulationinstability, associated with the onset phase of supercontinuum spectral broadening, as themain phenomenon leading to extreme event formation. This mechanism is studied both numericallyand analytically, considering a class of exact solutions of nonlinear Schrödinger equation which exhibitsingular localization characteristics. The results are experimentally verified, especially througha real-time spectral characterization system along with the use of innovative statistical metrics (e.g.spectral correlation maps). The excellent agreement between simulations and experiments allowedus to validate the theoretical predictions and get further insight into the complex dynamics associatedto nonlinear optical pulse propagation.