Numerical and analytical approaches to strongly correlated electron systems

In this thesis we consider three different models for strongly correlated electrons, namely a multi-band Hubbard model as well as the spinless Falicov-Kimball model, both with a semi-elliptical density of states in the limit of infinite dimensions d, and the attractive Hubbard model on a square lattice in d=2. In the first part, we study a two-band Hubbard model with unequal bandwidths and anisotropic Hund's rule coupling (J_z-model) in the limit of infinite dimensions within the dynamical mean-field theory (DMFT). Here, the DMFT impurity problem is solved with the use of quantum Monte Carlo (QMC) simulations. Our main result is that the J_z-model describes the occurrence of an orbital-selective Mott transition (OSMT), in contrast to earlier findings. We investigate the model with a high-precision DMFT algorithm, which was developed as part of this thesis and which supplements QMC with a high-frequency expansion of the self-energy. The main advantage of this scheme is the extraordinary accuracy of the numerical solutions, which can be obtained already with moderate computational effort, so that studies of multi-orbital systems within the DMFT+QMC are strongly improved. We also found that a suitably defined Falicov-Kimball (FK) model exhibits an OSMT, revealing the close connection of the Falicov-Kimball physics to the J_z-model in the OSM phase. In the second part of this thesis we study the attractive Hubbard model in two spatial dimensions within second-order self-consistent perturbation theory. This model is considered on a square lattice at finite doping and at low temperatures. Our main result is that the predictions of first-order perturbation theory (Hartree-Fock approximation) are renormalized by a factor of the order of unity even at arbitrarily weak interaction (U->0). The renormalization factor q can be evaluated as a function of the filling n for 00, the q-factor vanishes, signaling the divergence of self-consistent perturbation theory in this limit. Thus we present the first asymptotically exact results at weak-coupling for the negative-U Hubbard model in d=2 at finite doping.

Oxidation photochemistry in the remote marine boundary layer

Die bedeutendste Folge der Luftverschmutzung ist eine erhöhte Konzentration an Ozon (O3) in der Troposphäre innerhalb der letzten 150 Jahre. Ozon ist ein photochemisches Oxidationsmittel und ein Treibhausgas, das als wichtigste Vorstufe des Hydroxyradikals OH die Oxidationskraft der Atmosphäre stark beeinflusst. Um die Oxidationskraft der Atmosphäre und ihren Einfluss auf das Klima verstehen zu können, ist es von großer Bedeutung ein detailliertes Wissen über die Photochemie des Ozons und seiner Vorläufer, den Stickoxiden (NOx), in der Troposphäre zu besitzen. Dies erfordert das Verstehen der Bildungs- und Abbaumechanismen von Ozon und seiner Vorläufer. Als eine für den chemischen Ozonabbau wichtige Region kann die vom Menschen weitgehend unberührte marine Grenzschicht (Marine boundary layer (MBL)) angesehen werden. Bisher wurden für diese Region jedoch kaum Spurengasmessungen durchgeführt, und so sind die dort ablaufenden photochemischen Prozesse wenig untersucht. Da etwa 70 % der Erdoberfläche mit Ozeanen bedeckt sind, können die in der marinen Granzschicht ablaufenden Prozesse als signifikant für die gesamte Atmosphäre angesehen werden. Dies macht eine genaue Untersuchung dieser Region interessant. Um die photochemische Produktion und den Abbau von Ozon abschätzen zu können und den Einfluss antrophogener Emissionen auf troposphärisches Ozon zu quantifizieren, sind aktuelle Messergebnisse von NOx im pptv-Bereich für diese Region erforderlich. Die notwendigen Messungen von NO, NO2, O3, JNO2, J(O1D), HO2, OH, ROx sowie einiger meteorologischer Parameter wurden während der Fahrt des französischen Forschungsschiffes Marion-Dufresne auf dem südlichen Atlantik (28°S-57°S, 46°W-34°E) im März 2007 durchgeführt. Dabei sind für NO und NO2 die bisher niedrigsten gemessenen Werte zu verzeichnen. Die während der Messcampagne gewonnen Daten wurden hinsichtlich Ihrer Übereinstimmung mit den Bedingungen des photochemischen stationären Gleichgewichts (photochemical steady state (PSS)) überprüft. Dabei konnte eine Abweichung vom PSS festgestellt werden, welche unter Bedingungen niedriger NOx-Konzentrationen (5 bis 25pptv) einen unerwarteten Trend im Leighton-Verhältnis bewirkt, der abhängig vom NOx Mischungsverhältnis und der JNO2 Intensität ist. Signifikante Abweichungen vom Verhältnis liegen bei einer Zunahme der JNO2 Intensität vor. Diese Ergebnisse zeigen, dass die Abweichung vom PSS nicht beim Minimum der NOx-Konzentrationen und der JNO2 Werte liegt, so wie es in bisherigen theoretischen Studien dargelegt wurde und können als Hinweis auf weitere photochemische Prozesse bei höheren JNO2-Werten in einem System mit niedrigem NOx verstanden werden. Das wichtigste Ergebnis dieser Untersuchung, ist die Verifizierung des Leighton-Verhältnisses, das zur Charakterisierung des PSS dient, bei sehr geringen NOx-Konzentrationen in der MBL. Die bei dieser Doktorarbeit gewonnenen Erkenntnisse beweisen, dass unter den Bedingungen der marinen Granzschicht rein photochemischer Abbau von Ozon stattfindet und als Hauptursache hierfür während des Tages die Photolyse gilt. Mit Hilfe der gemessenen Parameter wurde der kritische NO-Level auf Werte zwischen 5 und 9 pptv abgeschätzt, wobei diese Werte im Vergleich zu bisherigen Studien vergleichsweise niedrig sind. Möglicherweise bedeutet dies, dass das Ozon Produktion/ Abbau-Potential des südlichen Atlantiks deutlich stärker auf die Verfügbarkeit von NO reagiert, als es in anderen Regionen der Fall ist. Im Rahmen der Doktorarbeit wurde desweiteren ein direkter Vergleich der gemessenen Spezies mit dem Modelergebnis eines 3-dimensionalen Zirkulationsmodel zur Simulation atmosphären chemischer Prozesse (EMAC) entlang der exakten Schiffsstrecke durchgeführt. Um die Übereinstimmung der Messergebnisse mit dem bisherigen Verständnis der atmosphärischen Radikalchemie zu überprüfen, wurde ein Gleichgewichtspunktmodel entwickelt, das die während der Überfahrt erhaltenen Daten für Berechungen verwendet. Ein Vergleich zwischen der gemessenen und der modellierten ROx Konzentrationen in einer Umgebung mit niedrigem NOx zeigt, dass die herkömmliche Theorie zur Reproduktion der Beobachtungen unzureichend ist. Die möglichen Gründe hierfür und die Folgen werden in dieser Doktorarbeit diskutiert.

Simulations of charge transport in organic compounds

To aid the design of organic semiconductors, we study the charge transport properties of organic liquid crystals, i.e. hexabenzocoronene and carbazole macrocycle, and single crystals, i.e. rubrene, indolocarbazole and benzothiophene derivatives (BTBT, BBBT). The aim is to find structure-property relationships linking the chemical structure as well as the morphology with the bulk charge carrier mobility of the compounds. To this end, molecular dynamics (MD) simulations are performed yielding realistic equilibrated morphologies. Partial charges and molecular orbitals are calculated based on single molecules in vacuum using quantum chemical methods. The molecular orbitals are then mapped onto the molecular positions and orientations, which allows calculation of the transfer integrals between nearest neighbors using the molecular orbital overlap method. Thus we obtain realistic transfer integral distributions and their autocorrelations. In case of organic crystals the differences between two descriptions of charge transport, namely semi-classical dynamics (SCD) in the small polaron limit and kinetic Monte Carlo (KMC) based on Marcus rates, are studied. The liquid crystals are investigated solely in the hopping limit. To simulate the charge dynamics using KMC, the centers of mass of the molecules are mapped onto lattice sites and the transfer integrals are used to compute the hopping rates. In the small polaron limit, where the electronic wave function is spread over a limited number of neighboring molecules, the Schroedinger equation is solved numerically using a semi-classical approach. The results are compared for the different compounds and methods and, where available, with experimental data. The carbazole macrocycles form columnar structures arranged on a hexagonal lattice with side chains facing inwards, so columns can closely approach each other allowing inter-columnar and thus three-dimensional transport. When taking only intra-columnar transport into account, the mobility is orders of magnitude lower than in the three-dimensional case. BTBT is a promising material for solution-processed organic field-effect transistors. We are able to show that, on the time-scales of charge transport, static disorder due to slow side chain motions is the main factor determining the mobility. The resulting broad transfer integral distributions modify the connectivity of the system but sufficiently many fast percolation paths remain for the charges. Rubrene, indolocarbazole and BBBT are examples of crystals without significant static disorder. The high mobility of rubrene is explained by two main features: first, the shifted cofacial alignment of its molecules, and second, the high center of mass vibrational frequency. In comparsion to SCD, only KMC based on Marcus rates is capable of describing neighbors with low coupling and of taking static disorder into account three-dimensionally. Thus it is the method of choice for crystalline systems dominated by static disorder. However, it is inappropriate for the case of strong coupling and underestimates the mobility of well-ordered crystals. SCD, despite its one-dimensionality, is valuable for crystals with strong coupling and little disorder. It also allows correct treatment of dynamical effects, such as intermolecular vibrations of the molecules. Rate equations are incapable of this, because simulations are performed on static snapshots. We have thus shown strengths and weaknesses of two state of the art models used to study charge transport in organic compounds, partially developed a program to compute and visualize transfer integral distributions and other charge transport properties, and found structure-mobility relations for several promising organic semiconductors.

Numerische und analytische Untersuchungen stark korrelierter fermionischer Mehrbandsysteme

In dieser Arbeit werden vier unterschiedliche, stark korrelierte, fermionische Mehrbandsysteme untersucht. Es handelt sich dabei um ein Mehrstörstellen-Anderson-Modell, zwei Hubbard-Modelle sowie ein Mehrbandsystem, wie es sich aus einer ab initio-Beschreibung für ein korreliertes Halbmetall ergibt.rnrnDie Betrachtung des Mehrstörstellen-Anderson-Modells konzentriert sich auf die Untersuchung des Einflusses der Austauschwechselwirkung und der nicht-lokalen Korrelationen zwischen zwei Störstellen in einem einfach-kubischen Gitter. Das zentrale Resultat ist die Abstandsabhängigkeit der Korrelationen der Störstellenelektronen, welche stark von der Gitterdimension und der relativen Position der Störstellen abhängen. Bemerkenswert ist hier die lange Reichweite der Korrelationen in der Diagonalrichtung des Gitters. Außerdem ergibt sich, dass eine antiferromagnetische Austauschwechselwirkung ein Singulett zwischen den Störstellenelektronen gegenüber den Kondo-Singuletts der einzelnen Störstellen favorisiert und so den Kondo-Effekt der einzelnen Störstellen behindert.rnrnEin Zweiband-Hubbard-Modell, das Jz-Modell, wird im Hinblick auf seine Mott-Phasen in Abhängigkeit von Dotierung und Kristallfeldaufspaltung auf dem Bethe-Gitter untersucht. Die Entartung der Bänder ist durch eine unterschiedliche Bandbreite aufgehoben. Wichtigstes Ergebnis sind die Phasendiagramme in Bezug auf Wechselwirkung, Gesamtfüllung und Kristallfeldparameter. Im Vergleich zu Einbandmodellen kommen im Jz-Modell sogenannte orbital-selektive Mott-Phasen hinzu, die, abhängig von Wechselwirkung, Gesamtfüllung und Kristallfeldparameter, einerseits metallischen und andererseits isolierenden Charakter haben. Ein neuer Aspekt ergibt sich durch den Kristallfeldparameter, der die ionischen Einteilchenniveaus relativ zueinander verschiebt, und für bestimmte Werte eine orbital-selektive Mott-Phase des breiten Bands ermöglicht. Im Vergleich mit analytischen Näherungslösungen und Einbandmodellen lassen sich generische Vielteilchen- und Korrelationseffekte von typischen Mehrband- und Einteilcheneffekten differenzieren.rnrnDas zweite untersuchte Hubbard-Modell beschreibt eine magneto-optische Falle mit einer endlichen Anzahl Gitterplätze, in welcher fermionische Atome platziert sind. Es wird eine z-antiferromagnetische Phase unter Berücksichtigung nicht-lokaler Vielteilchenkorrelationen erhalten, und dabei werden bekannte Ergebnisse einer effektiven Einteilchenbeschreibung verbessert.rnrnDas korrelierte Halbmetall wird im Rahmen einer Mehrbandrechnung im Hinblick auf Korrelationseffekte untersucht. Ausgangspunkt ist eine ab initio-Beschreibung durch die Dichtefunktionaltheorie (DFT), welche dann durch die Hinzunahme lokaler Korrelationen ergänzt wird. Die Vielteilcheneffekte werden an Hand einer einfachen Wechselwirkungsnäherung verdeutlicht, und für ein Wechselwirkungsmodell in sphärischer Symmetrie präzisiert. Es ergibt sich nur eine schwache Quasiteilchenrenormierung. Besonders für röntgenspektroskopische Experimente wird eine gute Übereinstimmung erzielt.rnrnDie numerischen Ergebnisse für das Jz-Modell basieren auf Quanten-Monte-Carlo-Simulationen im Rahmen der dynamischen Molekularfeldtheorie (DMFT). Für alle anderen Systeme wird ein Mehrband-Algorithmus entwickelt und implementiert, welcher explizit nicht-diagonale Mehrbandprozesse berücksichtigt.rnrn

Techniques for Lagrangian modelling of dispersion in geophysical flows

Basic concepts and definitions relative to Lagrangian Particle Dispersion Models (LPDMs)for the description of turbulent dispersion are introduced. The study focusses on LPDMs that use as input, for the large scale motion, fields produced by Eulerian models, with the small scale motions described by Lagrangian Stochastic Models (LSMs). The data of two different dynamical model have been used: a Large Eddy Simulation (LES) and a General Circulation Model (GCM). After reviewing the small scale closure adopted by the Eulerian model, the development and implementation of appropriate LSMs is outlined. The basic requirement of every LPDM used in this work is its fullfillment of the Well Mixed Condition (WMC). For the dispersion description in the GCM domain, a stochastic model of Markov order 0, consistent with the eddy-viscosity closure of the dynamical model, is implemented. A LSM of Markov order 1, more suitable for shorter timescales, has been implemented for the description of the unresolved motion of the LES fields. Different assumptions on the small scale correlation time are made. Tests of the LSM on GCM fields suggest that the use of an interpolation algorithm able to maintain an analytical consistency between the diffusion coefficient and its derivative is mandatory if the model has to satisfy the WMC. Also a dynamical time step selection scheme based on the diffusion coefficient shape is introduced, and the criteria for the integration step selection are discussed. Absolute and relative dispersion experiments are made with various unresolved motion settings for the LSM on LES data, and the results are compared with laboratory data. The study shows that the unresolved turbulence parameterization has a negligible influence on the absolute dispersion, while it affects the contribution of the relative dispersion and meandering to absolute dispersion, as well as the Lagrangian correlation.

Coarse-graining and quantum-classical adaptive coupling in soft matter

Die vorliegende Arbeit untersucht den Zusammenhang zwischen Skalen in Systemen weicher Materie, der für Multiskalen-Simulationen eine wichtige Rolle spielt. Zu diesem Zweck wurde eine Methode entwickelt, die die Approximation der Separierbarkeit von Variablen für die Molekulardynamik und ähnliche Anwendungen bewertet. Der zweite und größere Teil dieser Arbeit beschäftigt sich mit der konzeptionellen und technischen Erweiterung des Adaptive Resolution Scheme'' (AdResS), einer Methode zur gleichzeitigen Simulation von Systemen mit mehreren Auflösungsebenen. Diese Methode wurde auf Systeme erweitert, in denen klassische und quantenmechanische Effekte eine Rolle spielen.rnrnDie oben genannte erste Methode benötigt nur die analytische Form der Potentiale, wie sie die meisten Molekulardynamik-Programme zur Verfügung stellen. Die Anwendung der Methode auf ein spezielles Problem gibt bei erfolgreichem Ausgang einen numerischen Hinweis auf die Gültigkeit der Variablenseparation. Bei nicht erfolgreichem Ausgang garantiert sie, dass keine Separation der Variablen möglich ist. Die Methode wird exemplarisch auf ein zweiatomiges Molekül auf einer Oberfläche und für die zweidimensionale Version des Rotational Isomer State (RIS) Modells einer Polymerkette angewandt.rnrnDer zweite Teil der Arbeit behandelt die Entwicklung eines Algorithmus zur adaptiven Simulation von Systemen, in denen Quanteneffekte berücksichtigt werden. Die Quantennatur von Atomen wird dabei in der Pfadintegral-Methode durch einen klassischen Polymerring repräsentiert. Die adaptive Pfadintegral-Methode wird zunächst für einatomige Flüssigkeiten und tetraedrische Moleküle unter normalen thermodynamischen Bedingungen getestet. Schließlich wird die Stabilität der Methode durch ihre Anwendung auf flüssigen para-Wasserstoff bei niedrigen Temperaturen geprüft.

Modeling and simulation of fibrinogen and its adsorption behavior

In this thesis different approaches for the modeling and simulation of the blood protein fibrinogen are presented. The approaches are meant to systematically connect the multiple time and length scales involved in the dynamics of fibrinogen in solution and at inorganic surfaces. The first part of the thesis will cover simulations of fibrinogen on an all atom level. Simulations of the fibrinogen protomer and dimer are performed in explicit solvent to characterize the dynamics of fibrinogen in solution. These simulations reveal an unexpectedly large and fast bending motion that is facilitated by molecular hinges located in the coiled-coil region of fibrinogen. This behavior is characterized by a bending and a dihedral angle and the distribution of these angles is measured. As a consequence of the atomistic detail of the simulations it is possible to illuminate small scale behavior in the binding pockets of fibrinogen that hints at a previously unknown allosteric effect. In a second step atomistic simulations of the fibrinogen protomer are performed at graphite and mica surfaces to investigate initial adsorption stages. These simulations highlight the different adsorption mechanisms at the hydrophobic graphite surface and the charged, hydrophilic mica surface. It is found that the initial adsorption happens in a preferred orientation on mica. Many effects of practical interest involve aggregates of many fibrinogen molecules. To investigate such systems, time and length scales need to be simulated that are not attainable in atomistic simulations. It is therefore necessary to develop lower resolution models of fibrinogen. This is done in the second part of the thesis. First a systematically coarse grained model is derived and parametrized based on the atomistic simulations of the first part. In this model the fibrinogen molecule is represented by 45 beads instead of nearly 31,000 atoms. The intra-molecular interactions of the beads are modeled as a heterogeneous elastic network while inter-molecular interactions are assumed to be a combination of electrostatic and van der Waals interaction. A method is presented that determines the charges assigned to beads by matching the electrostatic potential in the atomistic simulation. Lastly a phenomenological model is developed that represents fibrinogen by five beads connected by rigid rods with two hinges. This model only captures the large scale dynamics in the atomistic simulations but can shed light on experimental observations of fibrinogen conformations at inorganic surfaces.

Evaluation of a bolus/infusion protocol for 11C-ABP688, a PET tracer for mGluR5

(11)C-ABP-688 is a selective tracer for the mGluR5 receptor. Its kinetics is fast and thus favourable for an equilibrium approach to determine receptor-related parameters. The purpose of this study was to test the hypothesis that the pattern of the (11)C-ABP688 uptake using a bolus-plus-infusion (B/I) protocol at early time points corresponds to the perfusion and at a later time point to the total distribution volume. METHODS: A bolus and a B/I study (1 h each) was performed in five healthy male volunteers. With the B/I protocol, early and late scans were normalized to gray matter, cerebellum and white matter. The same normalization was done on the maps of the total distribution volume (Vt) and K(1) which were calculated in the study with bolus only injection and the Logan method (Vt) and a two-tissue compartment model (K(1)). RESULTS: There was an excellent correlation close to the identity line between the pattern of the late uptake in the B/I study and Vt of the bolus-only study for all three normalizations. The pattern of the early uptake in the B/I study correlated well with the K(1) maps, but only when normalized to gray matter and cerebellum, not to white matter. CONCLUSION: It is demonstrated that with a B/I protocol the (11)C-ABP688 distribution in late scans reflects the pattern of the total distribution volume and is therefore a measure for the density pattern of mGluR5. The early scans following injection are related to blood flow, although not in a fully quantitative manner. The advantage of the B/I protocol is that no arterial blood sampling is required, which is advantageous in clinical studies.

Investigating the Effects of Various Crowd Characteristics on the Dynamic Properties of an Occupied Structure

One of the challenges for structural engineers during design is considering how the structure will respond to crowd-induced dynamic loading. It has been shown that human occupants of a structure do not simply add mass to the system when considering the overall dynamic response of the system, but interact with it and may induce changes of the dynamic properties from those of the empty structure. This study presents an investigation into the human-structure interaction based on several crowd characteristics and their effect on the dynamic properties of an empty structure. The dynamic properties including frequency, damping, and mode shapes were estimated for a single test structure by means of experimental modal analysis techniques. The same techniques were utilized to estimate the dynamic properties when the test structure was occupied by a crowd with different combinations of size, posture, and distribution. The goal of this study is to isolate the occupant characteristics in order to determine the significance of each to be considered when designing new structures to avoid crowd serviceability issues. The results are presented and summarized based on the level of influence of each characteristic. The posture that produces the most significant effects based on the scope of this research is standing with bent knees with a maximum decrease in frequency of the first mode of the empty structure by 32 percent atthe highest mass ratio. The associated damping also increased 36 times the damping of the empty structure. In addition to the analysis of the experimental data, finite element models and a two degree-of-freedom model were created. These models were used to gain an understanding of the test structure, model a crowd as an equivalent mass, and also to develop a single degree-of-freedom (SDOF) model to best represent a crowd of occupants based on the experimental results. The SDOF models created had an averagefrequency of 5.0 Hz, within the range presented in existing biomechanics research, and combined SDOF systems of the test structure and crowd were able to reproduce the frequency and damping ratios associated with experimental tests. Results of this study confirmed the existence of human-structure interaction andthe inability to simply model a crowd as only additional mass. The two degree-offreedom model determined was able to predict the change in natural frequency and damping ratio for a structure occupied by multiple group sizes in a single posture. These results and model are the preliminary steps in the development of an appropriate methodfor modeling a crowd in combination with a more complex FE model of the empty structure.

The Optimal Confidence Region for a Random Parameter

Under a two-level hierarchical model, suppose that the distribution of the random parameter is known or can be estimated well. Data are generated via a fixed, but unobservable realization of this parameter. In this paper, we derive the smallest confidence region of the random parameter under a joint Bayesian/frequentist paradigm. On average this optimal region can be much smaller than the corresponding Bayesian highest posterior density region. The new estimation procedure is appealing when one deals with data generated under a highly parallel structure, for example, data from a trial with a large number of clinical centers involved or genome-wide gene-expession data for estimating individual gene- or center-specific parameters simultaneously. The new proposal is illustrated with a typical microarray data set and its performance is examined via a small simulation study.

MODELING DIFFERENTIATED TREATMENT EFFECTS FOR MULTIPLE OUTCOMES DATA

Multiple outcomes data are commonly used to characterize treatment effects in medical research, for instance, multiple symptoms to characterize potential remission of a psychiatric disorder. Often either a global, i.e. symptom-invariant, treatment effect is evaluated. Such a treatment effect may over generalize the effect across the outcomes. On the other hand individual treatment effects, varying across all outcomes, are complicated to interpret, and their estimation may lose precision relative to a global summary. An effective compromise to summarize the treatment effect may be through patterns of the treatment effects, i.e. "differentiated effects." In this paper we propose a two-category model to differentiate treatment effects into two groups. A model fitting algorithm and simulation study are presented, and several methods are developed to analyze heterogeneity presenting in the treatment effects. The method is illustrated using an analysis of schizophrenia symptom data.

Water transport in complex, non-wetting porous layers with applications to water management in low temperature fuel cells

An experimental setup was designed to visualize water percolation inside the porous transport layer, PTL, of proton exchange membrane, PEM, fuel cells and identify the relevant characterization parameters. In parallel with the observation of the water movement, the injection pressure (pressure required to transport water through the PTL) was measured. A new scaling for the drainage in porous media has been proposed based on the ratio between the input and the dissipated energies during percolation. A proportional dependency was obtained between the energy ratio and a non-dimensional time and this relationship is not dependent on the flow regime; stable displacement or capillary fingering. Experimental results show that for different PTL samples (from different manufacturers) the proportionality is different. The identification of this proportionality allows a unique characterization of PTLs with respect to water transport. This scaling has relevance in porous media flows ranging far beyond fuel cells. In parallel with the experimental analysis, a two-dimensional numerical model was developed in order to simulate the phenomena observed in the experiments. The stochastic nature of the pore size distribution, the role of the PTL wettability and morphology properties on the water transport were analyzed. The effect of a second porous layer placed between the porous transport layer and the catalyst layer called microporous layer, MPL, was also studied. It was found that the presence of the MPL significantly reduced the water content on the PTL by enhancing fingering formation. Moreover, the presence of small defects (cracks) within the MPL was shown to enhance water management. Finally, a corroboration of the numerical simulation was carried out. A threedimensional version of the network model was developed mimicking the experimental conditions. The morphology and wettability of the PTL are tuned to the experiment data by using the new energy scaling of drainage in porous media. Once the fit between numerical and experimental data is obtained, the computational PTL structure can be used in different types of simulations where the conditions are representative of the fuel cell operating conditions.

MODULARITY BASED FUZZY COMMUNITY DETECTION IN SOCIAL NETWORKS

Fuzzy community detection is to identify fuzzy communities in a network, which are groups of vertices in the network such that the membership of a vertex in one community is in [0,1] and that the sum of memberships of vertices in all communities equals to 1. Fuzzy communities are pervasive in social networks, but only a few works have been done for fuzzy community detection. Recently, a one-step forward extension of Newman’s Modularity, the most popular quality function for disjoint community detection, results into the Generalized Modularity (GM) that demonstrates good performance in finding well-known fuzzy communities. Thus, GMis chosen as the quality function in our research. We first propose a generalized fuzzy t-norm modularity to investigate the effect of different fuzzy intersection operators on fuzzy community detection, since the introduction of a fuzzy intersection operation is made feasible by GM. The experimental results show that the Yager operator with a proper parameter value performs better than the product operator in revealing community structure. Then, we focus on how to find optimal fuzzy communities in a network by directly maximizing GM, which we call it Fuzzy Modularity Maximization (FMM) problem. The effort on FMM problem results into the major contribution of this thesis, an efficient and effective GM-based fuzzy community detection method that could automatically discover a fuzzy partition of a network when it is appropriate, which is much better than fuzzy partitions found by existing fuzzy community detection methods, and a crisp partition of a network when appropriate, which is competitive with partitions resulted from the best disjoint community detections up to now. We address FMM problem by iteratively solving a sub-problem called One-Step Modularity Maximization (OSMM). We present two approaches for solving this iterative procedure: a tree-based global optimizer called Find Best Leaf Node (FBLN) and a heuristic-based local optimizer. The OSMM problem is based on a simplified quadratic knapsack problem that can be solved in linear time; thus, a solution of OSMM can be found in linear time. Since the OSMM algorithm is called within FBLN recursively and the structure of the search tree is non-deterministic, we can see that the FMM/FBLN algorithm runs in a time complexity of at least O (n2). So, we also propose several highly efficient and very effective heuristic algorithms namely FMM/H algorithms. We compared our proposed FMM/H algorithms with two state-of-the-art community detection methods, modified MULTICUT Spectral Fuzzy c-Means (MSFCM) and Genetic Algorithm with a Local Search strategy (GALS), on 10 real-world data sets. The experimental results suggest that the H2 variant of FMM/H is the best performing version. The H2 algorithm is very competitive with GALS in producing maximum modularity partitions and performs much better than MSFCM. On all the 10 data sets, H2 is also 2-3 orders of magnitude faster than GALS. Furthermore, by adopting a simply modified version of the H2 algorithm as a mutation operator, we designed a genetic algorithm for fuzzy community detection, namely GAFCD, where elite selection and early termination are applied. The crossover operator is designed to make GAFCD converge fast and to enhance GAFCD’s ability of jumping out of local minimums. Experimental results on all the data sets show that GAFCD uncovers better community structure than GALS.

IMPROVING STARTABILITY AND REDUCING EMISSIONS IN FLEXFUEL SPARK IGNITION DIRECT INJECTION VARIABLE CAM TIMING ENGINE

Experimental work and analysis was done to investigate engine startup robustness and emissions of a flex-fuel spark ignition (SI) direct injection (DI) engine. The vaporization and other characteristics of ethanol fuel blends present a challenge at engine startup. Strategies to reduce the enrichment requirements for the first engine startup cycle and emissions for the second and third fired cycle at 25°C ± 1°C engine and intake air temperature were investigated. Research work was conducted on a single cylinder SIDI engine with gasoline and E85 fuels, to study the effect on first fired cycle of engine startup. Piston configurations that included a compression ratio change (11 vs 15.5) and piston geometry change (flattop vs bowl) were tested, along with changes in intake cam timing (95,110,125) and fuel pressure (0.4 MPa vs 3 MPa). The goal was to replicate the engine speed, manifold pressure, fuel pressure and testing temperature from an engine startup trace for investigating the first fired cycle for the engine. Results showed bowl piston was able to enable lower equivalence ratio engine starts with gasoline fuel, while also showing lower IMEP at the same equivalence ratio compared to flat top piston. With E85, bowl piston showed reduced IMEP as compression ratio increased at the same equivalence ratio. A preference for constant intake valve timing across fuels seemed to indicate that flattop piston might be a good flex-fuel piston. Significant improvements were seen with higher CR bowl piston with high fuel pressure starts, but showed no improvement with low fuel pressures. Simulation work was conducted to analyze initial three cycles of engine startup in GT-POWER for the same set of hardware used in the experimentations. A steady state validated model was modified for startup conditions. The results of which allowed an understanding of the relative residual levels and IMEP at the test points in the cam phasing space. This allowed selecting additional test points that enable use of higher residual levels, eliminating those with smaller trapped mass incapable of producing required IMEP for proper engine turnover. The second phase of experimental testing results for 2nd and 3rd startup cycle revealed both E10 and E85 prefer the same SOI of 240°bTDC at second and third startup cycle for the flat top piston and high injection pressures. E85 fuel optimal cam timing for startup showed that it tolerates more residuals compared to E10 fuel. Higher internal residuals drives down the Ø requirement for both fuels up to their combustion stability limit, this is thought to be direct benefit to vaporization due to increased cycle start temperature. Benefits are shown for an advance IMOP and retarded EMOP strategy at engine startup. Overall the amount of residuals preferred by an engine for E10 fuel at startup is thought to be constant across engine speed, thus could enable easier selection of optimized cam positions across the startup speeds.