No description available.

No description available.

No description available.

Yes / In this paper we discuss techniques for 3D data modelling and processing where the data are usually provided as point clouds which arise from 3D scanning devices. The particular approaches we adopt in modelling 3D data involves the use of Partial Differential Equations (PDEs). In particular we show how the continuous and discrete versions of elliptic PDEs can be used for data modelling. We show that using PDEs it is intuitively possible to model data corresponding to complex scenes. Furthermore, we show that data can be stored in compact format in the form of PDE boundary conditions. In order to demonstrate the methodology we utlise several examples of practical nature.

The DataFrame is a powerful table-like data structure used frequently in Data Science, the in-demand and innovative field focused on the extraction of valuable insights from data. Typically, datasets are not perfect upon collection and need to be prepared so that the resulting dataset is useful for statistical analysis. A DataFrame API supports optimized methods such as, selecting, aggregating and filtering rows, columns, and cells as well as renaming row and column labels. It also supports methods for normalizing data, merging data, adding new columns and labelling missing data among numerous other features. An API to work with tabular data would be useful in any general purpose language, so DataFrames have been incorporated into libraries like Pandas for Python and provided as native libraries in the languages R and Scala. Due to their wide-ranging use it is not uncommon to find implementations in many other languages like Java and Julia \cite{BigData}. In this work, we introduce RacketFrames, a Racket V8.0+ DataFrame implementation. We show the benefits an implementation can have on existing and future Racket projects. To quantify the performance of major DataFrame operations, we measure speed against Python Pandas and compare functional and object oriented paradigms. We hope to continue the trend for further Data Science tool development for Racket and other programming languages.

Studien som utförs i detta examensarbete behandlar ämnet datalagring för applikationen Vklass. Problemet som studeras i detta arbete består i att det varje månad inkommer ett stort antal filer till applikationen som lagras på en server lokalt i Stockholm där regelbun-den säkerhetskopiering görs till en annan server. Den implementerade datalagringslösning-en innebär att servern som lagrar data växer ständigt och utan särskild struktur då filerna lagras i en och samma mapp, vilket i sin tur innebär en mer komplex och mindre överskåd-lig lagring av filerna i lagringsmedierna samtidigt som prestandabegränsningar uppstår vid återläsning av säkerhetskopian. Kunden efterfrågar en lösning som innebär en mer effektiv hantering för alla de filer som lagras i applikationen på ett kostnads- och prestandaeffektivt vis. Lösningen bör även möjliggöra lagring av data med en mer överskådlig och lätthanter-lig struktur. De lösningsmetoder som undersökts i arbetet avser datalagring lokalt samt datalagring i molntjänster från de mest framstående aktörerna. Undersökningen och analysen av de valda lösningsmetoderna visade att datalagring i molntjänster uppfyller kraven som den nya data-lagringslösningen ämnar uppnå till skillnad från datalagring lokalt som visade brister och som därmed inte uppfyllde dessa krav. Slutsatsen i arbetet identifierade Microsoft Azure Storage som den mest lämpliga molnleverantören för datalagring då data lagras på ett pre-standa- och struktureffektivt sätt. Lösningen är även kostnadseffektiv då det dyraste lag-ringsalternativet som Microsoft erbjuder sänker lagringskostnaderna för Vklass med 84 % under de kommande fem åren i jämförelse med aktuell lagringsmetod. / The study in this report examines data storage for the application Vklass. The problem that has been studied in this work is that the application receives and stores a big amount of files which are stored on a server locally in Stockholm where back up is made to another server on a regular basis. Today’s implemented solution for data storage endorse that the server that stores the files continues to grow each month without further structure since all the upload files is stored in the same folder which yields a more complex and less lucid storage and management for the data. This yields limitations for the performance of the application when the backup-copy needs to be restored. The owners of the application request a solu-tion that gives a more effective management of the stored data with a cost and performance effective technique. The solution that will be presented in this study should also enable stor-ing data with a more lucid and convenient structure. The studied methods for the solution include storing data locally and data storage in the biggest public cloud services. The investigation and analysis of the chosen methods proved that data storage in cloud services fulfilled the requirements for the identified solution as opposed to the methods for storing data locally that proved to yield deficits which would not fulfill these requirements. The deduction of this study identified Microsoft Azure Sto-rage to be the best public cloud solution for the given problem since data is stored in a per-formance and structure effective way. It was also proven that even the most expensive sto-rage solution provided by Azure Storage lowered the costs for data storage by 84 percent compared to today’s data storage solution during the first five years.

NAND Flash-based Solid State Disks have many attractive technical merits, such as low power consumption, light weight, shock resistance, sustainability of hotter operation regimes, and extraordinarily high performance for random read access, which makes SSDs immensely popular and be widely employed in different types of environments including portable devices, personal computers, large data centers, and distributed data systems. However, current SSDs still suffer from several critical inherent limitations, such as the inability of in-place-update, asymmetric read and write performance, slow garbage collection processes, limited endurance, and degraded write performance with the adoption of MLC and TLC techniques. To alleviate these limitations, we propose optimizations from both specific outside applications layer and SSDs' internal layer. Since SSDs are good compromise between the performance and price, so SSDs are widely deployed as second layer caches sitting between DRAMs and hard disks to boost the system performance. Due to the special properties of SSDs such as the internal garbage collection processes and limited lifetime, traditional cache devices like DRAM and SRAM based optimizations might not work consistently for SSD-based cache. Therefore, for the outside applications layer, our work focus on integrating the special properties of SSDs into the optimizations of SSD caches. Moreover, our work also involves the alleviation of the increased Flash write latency and ECC complexity due to the adoption of MLC and TLC technologies by analyzing the real work workloads.

Volume Holographic Data Storage Systems (HDSS) has been of interest for almost seven decades, and are now considered as a viable option for Write Once Read Many (WORM) cold data storage applications. Thanks to the Bragg selectivity of thick volume holograms, HDSS stores several hundreds of holograms on top of each other, called multiplexed data pages, by which data recording density can be substantially increased compared to surface recordings. On the other hand, signal intensity upon reconstruction of such multiplexed data pages inversely scales with number of multiplexing squared. Therefore, longer detection time and/or a high power laser along with a large dynamic range material is needed to make HDSS a truly viable "fast and high density" option for WORM applications. Historically, the trade-off between data density and data rate is well recognized. The challenge has been partially solved by continuous efforts such as improvement of materials, optical architectures, opto-mechanical systems and signal processing [1,2]. In this dissertation, we provide an additional pathway for HDSS to further increase both data density and transfer rates which is Cavities Enhancement Techniques for HDSS, to overcome the fundamental tradeoff. Key ideas are: recycling light with cavity to enhance data rate, and increasing number of multiplexing by combining cavity-eigenmode multiplexing, a subset of orthogonal phasecode multiplexing, with angular multiplexing. Based on this idea, we design and demonstrate Cavity-enhanced HDSS in such a way that we increase data rate and/or data density by at least factor of 2 while taking advantage of previous improvements as they are, or only with the minimum amount of modifications. In Section 1, we review history of HDSS and summarize the latest research results of HDSS and requirements on modern optical data storage systems as they relate to our solutions. In Section 2, theory of volume holography is reviewed by emphasizing understanding of angular and orthogonal phase code multiplexing. In Section 3 the theory of cavity enhanced reference arms is presented. We discuss how cavities provide a coherent boost to the beam power, which can be used in recording to alleviate source power requirements and/or increase the data recording rate and demonstrate the enhancement experimentally. Beyond basic enhancement, cavities also enable orthogonal phase code multiplexing via cavity eigenmodes. In Section 4, we experimentally demonstrate angular and orthogonal phase code hybrid multiplexing to overcome the limitation of the maximum number of multiplexing imposed by the geometrical constraints of angular multiplexing. In Section 5, novel aspects of the research are discussed in conjunction with the application of the technology for commercial use. Conclusions and future research direction are addressed in Section 6.

The research that stems from my doctoral dissertation focuses on addressing essential challenges in developing techniques that utilize solid-state memory technologies (with emphasis on NAND flash memory) from device, circuit, architecture, and system perspectives in order to exploit their true potential for improving I/O performance in high-performance computing systems. These challenges include not only the performance quirks arising from the physical nature of NAND flash memory, e.g., the inability to modify data in-place, read/write performance asymmetry, and slow and constrained erase functionality, but also the reliability drawbacks that limits solid state drives (SSDs) from widely deployed. To address these challenges, I have proposed, analyzed, and evaluated the I/O scheduling schemes, strategies for storage space virtualization, and data protection methods, to boost the performance and reliability of SSDs.

L’argent associé aux colloïdes de titane est connu depuis longtemps pour produire un effet photochrome réversible. Le changement de couleur est dû à la réduction des sels d’argent en nanoparticules métalliques sous lumière UV et à l’oxydation de ces dernières sous lumière visible. Récemment, une nouvelle fonctionnalité induisant du photochromisme multicolore a été rapportée dans des matériaux nanocomposites constitués de nanoparticules d’argent introduites dans des films nanoporeux d’oxyde de titane. Dans cette thèse, nous étudions l’influence d’une matrice mésoporeuse d’oxyde de titane avec des tailles de pores contrôlées sur le comportement photochrome de tels films. Nous montrons que la porosité des films permet de contrôler la taille des particules formées sous insolations UV et qu’il est possible de blanchir les motifs photo-induits avec une lumière monochromatique visible bien qu’habituellement ce type d’insolation colore le film. L’utilisation de ces matériaux permet aussi d’améliorer sensiblement la stabilité temporelle des inscriptions photo-induites, ce qui permet de les utiliser comme support de données reconfigurables. Nous démontrons aussi qu’au-delà d’un seuil en intensité, il est possible d’inscrire des motifs permanents avec une lumière UV ou visible, ce qui permet l’utilisation de ces films comme support de données permanentes. Enfin, dans une dernière partie, nous montrons qu’il est possible de photo-inscrire en lumière visible des couleurs dichroïques fortement réfléchissantes sans aucune réduction préalable des sels d’argent / Silver species adsorbed on colloidal titania have been known for a long time to exhibit photochromism. The color change is due to the reduction of silver salts from metallic nanoparticles under UV illumination and oxidation of these latters under visible illumination. Recently, a new functionality inducing multicolor photochromism has been reported in nanocomposite materials constituted by silver nanoparticles introduced in nanoporous titania film. In this dissertation, we study the influence of mesoporous titania matrix with controlled pores sizes on the photochromisrn behavior of such films. We show that the films porosity allows to control the formed particles under UV illumination and that is possible to bleach the photo-induced patterns with a monochromatic visible light although usually this type of illumination color the film. The utilization of these materials allows also to sensibly improve the temporal stability of photo-induced inscriptions, which allows to use them as rewritable data carriers. We demonstrate also that above an intensity threshold, it is possible to inscribe permanent pattern with an UV or visible illumination, which allows to use these films like permanent data carriers. Finally, in the last part, we show that it is possible to photo-induce in visible light dichroic color highly reflective without prior reduction of silver salts