Heterogeneous Multi-core Architectures for High Performance Computing

This thesis deals with heterogeneous architectures in standard workstations. Heterogeneous architectures represent an appealing alternative to traditional supercomputers because they are based on commodity components fabricated in large quantities. Hence their price-performance ratio is unparalleled in the world of high performance computing (HPC). In particular, different aspects related to the performance and consumption of heterogeneous architectures have been explored. The thesis initially focuses on an efficient implementation of a parallel application, where the execution time is dominated by an high number of floating point instructions. Then the thesis touches the central problem of efficient management of power peaks in heterogeneous computing systems. Finally it discusses a memory-bounded problem, where the execution time is dominated by the memory latency. Specifically, the following main contributions have been carried out: A novel framework for the design and analysis of solar field for Central Receiver Systems (CRS) has been developed. The implementation based on desktop workstation equipped with multiple Graphics Processing Units (GPUs) is motivated by the need to have an accurate and fast simulation environment for studying mirror imperfection and non-planar geometries. Secondly, a power-aware scheduling algorithm on heterogeneous CPU-GPU architectures, based on an efficient distribution of the computing workload to the resources, has been realized. The scheduler manages the resources of several computing nodes with a view to reducing the peak power. The two main contributions of this work follow: the approach reduces the supply cost due to high peak power whilst having negligible impact on the parallelism of computational nodes. from another point of view the developed model allows designer to increase the number of cores without increasing the capacity of the power supply unit. Finally, an implementation for efficient graph exploration on reconfigurable architectures is presented. The purpose is to accelerate graph exploration, reducing the number of random memory accesses.

Numerical study of homogeneous and heterogeneous crystal nucleation in colloidal systems

Im Rahmen dieser Arbeit wurden Computersimulationen von Keimbildungs- und Kris\-tallisationsprozessen in rnkolloidalen Systemen durchgef\"uhrt. rnEine Kombination von Monte-Carlo-Simulationsmethoden und der Forward-Flux-Sampling-Technik wurde rnimplementiert, um die homogene und heterogene Nukleation von Kristallen monodisperser Hart\-kugeln zu untersuchen. rnIm m\"a\ss{ig} unterk\"uhlten Bulk-Hartkugelsystem sagen wir die homogenen Nukleationsraten voraus und rnvergleichen die Resultate mit anderen theoretischen Ergebnissen und experimentellen Daten. rnWeiterhin analysieren wir die kristallinen Cluster in den Keimbildungs- und Wachstumszonen, rnwobei sich herausstellt, dass kristalline Cluster sich in unterschiedlichen Formen im System bilden. rnKleine Cluster sind eher l\"anglich in eine beliebige Richtung ausgedehnt, w\"ahrend gr\"o\ss{ere} rnCluster kompakter und von ellipsoidaler Gestalt sind. rn rnIm n\"achsten Teil untersuchen wir die heterogene Keimbildung an strukturierten bcc (100)-W\"anden. rnDie 2d-Analyse der kristallinen Schichten an der Wand zeigt, dass die Struktur der rnWand eine entscheidende Rolle in der Kristallisation von Hartkugelkolloiden spielt. rnWir sagen zudem die heterogenen Kristallbildungsraten bei verschiedenen \"Ubers\"attigungsgraden voraus. rnDurch Analyse der gr\"o\ss{ten} Cluster an der Wand sch\"atzen wir zus\"atzlich den Kontaktwinkel rnzwischen Kristallcluster und Wand ab. rnEs stellt sich heraus, dass wir in solchen Systemen weit von der Benetzungsregion rnentfernt sind und der Kristallisationsprozess durch heterogene Nukleation stattfindet. rn rnIm letzten Teil der Arbeit betrachten wir die Kristallisation von Lennard-Jones-Kolloidsystemen rnzwischen zwei ebenen W\"anden. rnUm die Erstarrungsprozesse f\"ur ein solches System zu untersuchen, haben wir eine Analyse des rnOrdnungsparameters f\"ur die Bindung-Ausrichtung in den Schichten durchgef\"urt. rnDie Ergebnisse zeigen, dass innerhalb einer Schicht keine hexatische Ordnung besteht, rnwelche auf einen Kosterlitz-Thouless-Schmelzvorgang hinweisen w\"urde. rnDie Hysterese in den Erhitzungs-Gefrier\-kurven zeigt dar\"uber hinaus, dass der Kristallisationsprozess rneinen aktivierten Prozess darstellt.

Atmospheric heterogeneous reactions of N 2 O 5 and NO 3 radicals with mineral dust particles

Die heterogenen Reaktionen von N2O5 bzw. NO3 auf mineralischen Staubpartikeln wurden untersucht, um deren Einfluss auf den Abbau atmosphärischer Stickoxide (NOx) sowie auf die chemische Veränderung der Staubpartikel während ihres Transportes durch die Atmosphäre besser verstehen zu können. Die experimentellen Studien wurden bei Atmosphärendruck, Raumtemperatur und unterschiedlichen relativen Luftfeuchten durchgeführt. Der Aufnahmekoeffizient γ(N2O5) von N2O5 auf dispergiertem Staub aus der Sahara wurde zu 0,020 ± 0,002 (1σ) bestimmt, unabhängig von der relativen Feuchte (0 - 67 %) sowie der N2O5-Konzentration (5x1011 - 3x1013 Moleküle cm-3).rnDie Analyse der Reaktionsprodukte in der Gasphase sowie auf der Partikeloberfläche führt zu der Annahme, dass N2O5 auf der Staubpartikeloberfläche zu Nitrat hydrolysiert wird. Es konnte kein Einfluss der relativen Feuchte auf den Aufnahmekoeffizienten ermittelt werden, was durch das vorhandene interlamellare Wasser, welches bis zu 10 % der Partikelmasse betragen kann, erklärbar ist. Der gemessene Wert des Aufnahmekoeffizienten ist unabhängig von der Eingangs-N2O5-Konzentration, was sich über die sehr große innere Oberfläche der Partikel erklären lässt. Dennoch ließ sich durch eine vorherige Konditionierung der Partikel mit gasförmigem HNO3, was eine Nitratanreicherung an der Oberfläche bewirkt, die Effizienz der N2O5-Aufnahme auf die Staubpartikel reduzieren. Zusätzliche Studien befassten sich mit der Bestimmung des Aufnahmekoeffizienten von N2O5 auf Illit-Partikeln und auf Teststaub aus Arizona. Bei einer relativen Luftfeuchte von 0 % wurden für γ(N2O5) Werte von 0,084 ± 0,019 (1σ) für Illit und von 0,010 ± 0,001 (1σ) für Arizona Teststaub ermittelt.rnUnter Anwendung einer neuartigen Messmethode, die auf der zeitgleichen Messung der Konzentrationsabnahme von NO3 und N2O5 relativ zueinander beruht, wurde das Verhältnis γ(NO3)/γ(N2O5) der Aufnahmekoeffizienten von NO3 und N2O5 auf Saharastaub zu 0,9 ± 0,4 (1σ) bestimmt. Dieser Wert war unabhängig von der relativen Feuchte, den NO3- und N2O5-Konzentrationen sowie der Reaktionszeit, obwohl eine Oberflächendeaktivierung für beide Spurenstoffe beobachtet wurde.

Diffusion in heterogeneous systems studied by laser scanning confocal microscopy and fluorescence correlation spectroscopy

Understanding and controlling the mechanism of the diffusion of small molecules, macromolecules and nanoparticles in heterogeneous environments is of paramount fundamental and technological importance. The aim of the thesis is to show, how by studying the tracer diffusion in complex systems, one can obtain information about the tracer itself, and the system where the tracer is diffusing. rnIn the first part of my thesis I will introduce the Fluorescence Correlation Spectroscopy (FCS) which is a powerful tool to investigate the diffusion of fluorescent species in various environments. By using the main advantage of FCS namely the very small probing volume (<1µm3) I was able to track the kinetics of phase separation in polymer blends at late stages by looking on the molecular tracer diffusion in individual domains of the heterogeneous structure of the blend. The phase separation process at intermediate stages was monitored with laser scanning confocal microscopy (LSCM) in real time providing images of droplet coalescence and growth. rnIn a further project described in my thesis I will show that even when the length scale of the heterogeneities becomes smaller than the FCS probing volume one can still obtain important microscopic information by studying small tracer diffusion. To do so, I will introduce a system of star shaped polymer solutions and will demonstrate that the mobility of small molecular tracers on microscopic level is nearly not affected by the transition of the polymer system to a “glassy” macroscopic state. rnIn the last part of the thesis I will introduce and describe a new stimuli responsive system which I have developed, that combines two levels of nanoporosity. The system is based on poly-N-isopropylacrylamide (PNIPAM) and silica inverse opals (iOpals), and allows controlling the diffusion of tracer molecules. rn

The wind profile in the atmospheric stable boundary layer over complex terrain and heterogeneous surfaces: limitations of local similarity theory

L'indagine ha riguardato il profilo del vento nei primi 30 metri dello strato limite atmosferico stabile nell'ambito della teoria di similarità locale. Ad oggi, diversi esperimenti hanno confermato la validità della teoria per strati-limite su terreni livellati e superfici omogenee. Tali condizioni ideali sono però infrequenti nella realtà ed è perciò importante capire quali siano i limiti della similarità locale per strati-limite su terreni complessi e superfici disomogenee. Entrambe le condizioni sono presenti a Ny-Alesund (Svalbard, Norvegia) dove il Consiglio Nazionale delle Ricerche (CNR), nel 2009, ha installato una torre di 30 m, la Amudsen-Nobile Climate Change Tower (CCT), per lo studio dello strato-limite artico. Il lavoro di tesi ha riguardato misure di vento e turbolenza acquisite sulla CCT da maggio 2012 a maggio 2014. Il confronto tra le velocità del vento misurate dagli anemometri installati sulla CCT, ha rivelato criticità nel dato sonico manifestatesi con sovrastime sistematiche e maggiore erraticità rispetto alle misure provenienti dagli anemometri a elica. Un test condotto fra diversi metodi per il calcolo dei gradienti verticali della velocità del vento ha rivelato scarsa sensibilità dei risultati ottenuti al particolare metodo utilizzato. Lo studio ha riguardato i gradienti verticali adimensionali della velocità del vento nei primi 30-m dello strato limite stabile. Deviazioni significative tra i tra le osservazioni e i valori predetti dalla similarità locale sono state osservate in particolare per i livelli più distanti dal suolo e per valori crescenti del parametro di stabilità z/L (L, lunghezza di Obukhov locale). In particolare, si sono osservati gradienti adimensionali inferiori a quelli predetti dalle più usate relazioni di flusso-gradiente. Tali deviazioni, presenti perlopiù per z/L>0.1, sono state associate ad un effetto di accentuazione della turbolenza da parte delle irregolarità del terreno. Per condizioni meno stabili, z/L<0.1, scarti positivi tra i gradienti osservati e quelli attesi sono stati attribuiti alla formazione di strati limite interni in condizioni di vento dal mare verso la costa. Sono stati proposti diversi metodi per la stima dell'effetto della self-correlazione nella derivazione delle relazioni di flusso-gradiente, dovuta alla condivisione della variabile u*. La formula per il coefficiente lineare di self correlazione e le sue distribuzioni di probabilità empiriche sono state derivate e hanno permesso di stimare il livello di self-correlazione presente nel dataset considerato.

Intermittent spike-wave dynamics in a heterogeneous, spatially extended neural mass model

Generalised epileptic seizures are frequently accompanied by sudden, reversible transitions from low amplitude, irregular background activity to high amplitude, regular spike-wave discharges (SWD) in the EEG. The underlying mechanisms responsible for SWD generation and for the apparently spontaneous transitions to SWD and back again are still not fully understood. Specifically, the role of spatial cortico-cortical interactions in ictogenesis is not well studied. We present a macroscopic, neural mass model of a cortical column which includes two distinct time scales of inhibition. This model can produce both an oscillatory background and a pathological SWD rhythm. We demonstrate that coupling two of these cortical columns can lead to a bistability between out-of-phase, low amplitude background dynamics and in-phase, high amplitude SWD activity. Stimuli can cause state-dependent transitions from background into SWD. In an extended local area of cortex, spatial heterogeneities in a model parameter can lead to spontaneous reversible transitions from a desynchronised background to synchronous SWD due to intermittency. The deterministic model is therefore capable of producing absence seizure-like events without any time dependent adjustment of model parameters. The emergence of such mechanisms due to spatial coupling demonstrates the importance of spatial interactions in modelling ictal dynamics, and in the study of ictogenesis.

Heterogeneous response of J-wave syndromes to beta-adrenergic stimulation

Inferolateral early repolarization (ER) and Brugada syndrome manifest with J waves. Isoproterenol suppresses recurrent ventricular arrhythmias while reducing J waves in both disorders.

High Bee and Wasp Diversity in a Heterogeneous Tropical Farming System Compared to Protected Forest

It is a globally important challenge to meet increasing demands for resources and, at the same time, protect biodiversity and ecosystem services. Farming is usually regarded as a major threat to biodiversity due to its expansion into natural areas. We compared biodiversity of bees and wasps between heterogeneous small-scale farming areas and protected forest in northern coastal Belize, Central America. Malaise traps operated for three months during the transition from wet to dry season. Farming areas consisted of a mosaic of mixed crop types, open habitat, secondary forest, and agroforestry. Mean species richness per site (alpha diversity), as well as spatial and temporal community variation (beta diversity) of bees and wasps were equal or higher in farming areas compared to protected forest. The higher species richness and community variation in farmland was due to additional species that did not occur in the forest, whereas most species trapped in forest were also found in farming areas. The overall regional species richness (gamma diversity) increased by 70% with the inclusion of farming areas. Our results suggest that small-scale farming systems adjacent to protected forest may not only conserve, but even favour, biodiversity of some taxonomic groups. We can, however, not exclude possible declines of bee and wasp diversity in more intensified farmland or in landscapes completely covered by heterogeneous farming systems.