Hourly values of X, Y and Z geomagnetic components recorded at Coimbra Station (IAGA code: COI) and local geomagnetic K indices for 2007 to 2014

This study examines the performance of series of two geomagnetic indices and series synthesized from a semi-empirical model of magnetospheric currents, in explaining the geomagnetic activity observed at Northern Hemipshere's mid-latitude ground-based stations. We analyse data, for the 2007 to 2014 period, from four magnetic observatories (Coimbra, Portugal; Panagyurishte, Bulgary; Novosibirsk, Russia and Boulder, USA), at geomagnetic latitudes between 40° and 50° N. The quiet daily (QD) variation is firstly removed from the time series of the geomagnetic horizontal component (H) using natural orthogonal components (NOC) tools. We compare the resulting series with series of storm-time disturbance (Dst) and ring current (RC) indices and with H series synthesized from the Tsyganenko and Sitnov (2005, doi:10.1029/2004JA010798) (TS05) semi-empirical model of storm-time geomagnetic field. In the analysis, we separate days with low and high local K-index values. Our results show that NOC models are as efficient as standard models of QD variation in preparing raw data to be compared with proxies, but with much less complexity. For the two stations in Europe, we obtain indication that NOC models could be able to separate ionospheric and magnetospheric contributions. Dst and RC series explain the four observatory H-series successfully, with values for the mean of significant correlation coefficients, from 0.5 to 0.6 during low geomagnetic activity (K less than 4) and from 0.6 to 0.7 for geomagnetic active days (K greater than or equal to 4). With regard to the performance of TS05, our results show that the four observatories separate into two groups: Coimbra and Panagyurishte, in one group, for which the magnetospheric/ionospheric ratio in QD variation is smaller, a dominantly QD ionospheric contribution can be removed and TS05 simulations are the best proxy; Boulder and Novosibirsk,in the other group, for which the ionospheric and magnetospheric contributions in QD variation can not be differentiated and correlations with TS05 series can not be made to improve. The main contributor to magnetospheric QD signal are Birkeland currents. The relatively good success of TS05 model in explaining ground-based irregular geomagnetic activity at mid-latitudes makes it an effective tool to classify storms according to their main sources. For Coimbra and Panagyurishte in particular, where ionospheric and magnetospheric daily contributions seem easier to separate, we can aspire to use the TS05 model for ensemble generation in space weather (SW) forecasting and interpretation of past SW events.

Automatic maintenance of category hierarchy

Category hierarchy is an abstraction mechanism for efficiently managing large-scale resources. In an open environment, a category hierarchy will inevitably become inappropriate for managing resources that constantly change with unpredictable pattern. An inappropriate category hierarchy will mislead the management of resources. The increasing dynamicity and scale of online resources increase the requirement of automatically maintaining category hierarchy. Previous studies about category hierarchy mainly focus on either the generation of category hierarchy or the classification of resources under a pre-defined category hierarchy. The automatic maintenance of category hierarchy has been neglected. Making abstraction among categories and measuring the similarity between categories are two basic behaviours to generate a category hierarchy. Humans are good at making abstraction but limited in ability to calculate the similarities between large-scale resources. Computing models are good at calculating the similarities between large-scale resources but limited in ability to make abstraction. To take both advantages of human view and computing ability, this paper proposes a two-phase approach to automatically maintaining category hierarchy within two scales by detecting the internal pattern change of categories. The global phase clusters resources to generate a reference category hierarchy and gets similarity between categories to detect inappropriate categories in the initial category hierarchy. The accuracy of the clustering approaches in generating category hierarchy determines the rationality of the global maintenance. The local phase detects topical changes and then adjusts inappropriate categories with three local operations. The global phase can quickly target inappropriate categories top-down and carry out cross-branch adjustment, which can also accelerate the local-phase adjustments. The local phase detects and adjusts the local-range inappropriate categories that are not adjusted in the global phase. By incorporating the two complementary phase adjustments, the approach can significantly improve the topical cohesion and accuracy of category hierarchy. A new measure is proposed for evaluating category hierarchy considering not only the balance of the hierarchical structure but also the accuracy of classification. Experiments show that the proposed approach is feasible and effective to adjust inappropriate category hierarchy. The proposed approach can be used to maintain the category hierarchy for managing various resources in dynamic application environment. It also provides an approach to specialize the current online category hierarchy to organize resources with more specific categories.

(Table 1) Dissolved organic carbon and dissolved sugar of interstitial water of DSDP Holes 96-618, 96-619 and 96-623

Preliminary data on dissolved organic carbon (DOC) and dissolved sugars in interstitial water samples collected at Sites 618, 619, and 623 of Deep Sea Drilling Project Leg 96 are presented. At Site 618 in Orca Basin, the DOC content of the interstitial water peaks in the hypersaline sulfate reduction zone. The sugar content reaches a maximum and the DOC content begins to decrease at the depth of methane gas generation. Below that depth, the sugar and DOC contents are about constant. At Site 619 in Pigmy Basin, the DOC content increases slightly with depth in the sulfate reduction and the methane fermentation zones. The sugar content is lower in the sulfate reduction zone than in the methane fermentation zone; sugar concentration increases and fluctuates with methane gas percentages within the methane fermentation zone. At Site 623 in the lower fan region of the Mississippi Fan, there is no sulfate reduction zone. The DOC and sugar contents of the interstitial water are almost constant with depth.

A probabilistic approach to the estimation of regional photovoltaic power generation using meteorological data

In Germany the upscaling algorithm is currently the standard approach for evaluating the PV power produced in a region. This method involves spatially interpolating the normalized power of a set of reference PV plants to estimate the power production by another set of unknown plants. As little information on the performances of this method could be found in the literature, the first goal of this thesis is to conduct an analysis of the uncertainty associated to this method. It was found that this method can lead to large errors when the set of reference plants has different characteristics or weather conditions than the set of unknown plants and when the set of reference plants is small. Based on these preliminary findings, an alternative method is proposed for calculating the aggregate power production of a set of PV plants. A probabilistic approach has been chosen by which a power production is calculated at each PV plant from corresponding weather data. The probabilistic approach consists of evaluating the power for each frequently occurring value of the parameters and estimating the most probable value by averaging these power values weighted by their frequency of occurrence. Most frequent parameter sets (e.g. module azimuth and tilt angle) and their frequency of occurrence have been assessed on the basis of a statistical analysis of parameters of approx. 35 000 PV plants. It has been found that the plant parameters are statistically dependent on the size and location of the PV plants. Accordingly, separate statistical values have been assessed for 14 classes of nominal capacity and 95 regions in Germany (two-digit zip-code areas). The performances of the upscaling and probabilistic approaches have been compared on the basis of 15 min power measurements from 715 PV plants provided by the German distribution system operator LEW Verteilnetz. It was found that the error of the probabilistic method is smaller than that of the upscaling method when the number of reference plants is sufficiently large (>100 reference plants in the case study considered in this chapter). When the number of reference plants is limited (<50 reference plants for the considered case study), it was found that the proposed approach provides a noticeable gain in accuracy with respect to the upscaling method.

Semi-automatic parallelisation of unstructured mesh codes

The availability of CFD software that can easily be used and produce high efficiency on a wide range of parallel computers is extremely limited. The investment and expertise required to parallelise a code can be enormous. In addition, the cost of supercomputers forces high utilisation to justify their purchase, requiring a wide range of software. To break this impasse, tools are urgently required to assist in the parallelisation process that dramatically reduce the parallelisation time but do not degrade the performance of the resulting parallel software. In this paper we discuss enhancements to the Computer Aided Parallelisation Tools (CAPTools) to assist in the parallelisation of complex unstructured mesh-based computational mechanics codes.

Automatic implementation of dynamic load balancing strategies for structured computational mechanics codes

This paper presents a new dynamic load balancing technique for structured mesh computational mechanics codes in which the processor partition range limits of just one of the partitioned dimensions uses non-coincidental limits, as opposed to using coincidental limits in all of the partitioned dimensions. The partition range limits are 'staggered', allowing greater flexibility in obtaining a balanced load distribution in comparison to when the limits are changed 'globally'. as the load increase/decrease on one processor no longer restricts the load decrease/increase on a neighbouring processor. The automatic implementation of this 'staggered' load balancing strategy within an existing parallel code is presented in this paper, along with some preliminary results.

Automated Software Debugging Using Hybrid Static/Dynamic Analysis

With the increasing complexity of today's software, the software development process is becoming highly time and resource consuming. The increasing number of software configurations, input parameters, usage scenarios, supporting platforms, external dependencies, and versions plays an important role in expanding the costs of maintaining and repairing unforeseeable software faults. To repair software faults, developers spend considerable time in identifying the scenarios leading to those faults and root-causing the problems. While software debugging remains largely manual, it is not the case with software testing and verification. The goal of this research is to improve the software development process in general, and software debugging process in particular, by devising techniques and methods for automated software debugging, which leverage the advances in automatic test case generation and replay. In this research, novel algorithms are devised to discover faulty execution paths in programs by utilizing already existing software test cases, which can be either automatically or manually generated. The execution traces, or alternatively, the sequence covers of the failing test cases are extracted. Afterwards, commonalities between these test case sequence covers are extracted, processed, analyzed, and then presented to the developers in the form of subsequences that may be causing the fault. The hypothesis is that code sequences that are shared between a number of faulty test cases for the same reason resemble the faulty execution path, and hence, the search space for the faulty execution path can be narrowed down by using a large number of test cases. To achieve this goal, an efficient algorithm is implemented for finding common subsequences among a set of code sequence covers. Optimization techniques are devised to generate shorter and more logical sequence covers, and to select subsequences with high likelihood of containing the root cause among the set of all possible common subsequences. A hybrid static/dynamic analysis approach is designed to trace back the common subsequences from the end to the root cause. A debugging tool is created to enable developers to use the approach, and integrate it with an existing Integrated Development Environment. The tool is also integrated with the environment's program editors so that developers can benefit from both the tool suggestions, and their source code counterparts. Finally, a comparison between the developed approach and the state-of-the-art techniques shows that developers need only to inspect a small number of lines in order to find the root cause of the fault. Furthermore, experimental evaluation shows that the algorithm optimizations lead to better results in terms of both the algorithm running time and the output subsequence length.

Profile detection through source code static analysis

The present article reflects the progress of an ongoing master’s dissertation on language engineering. The main goal of the work here described, is to infer a programmer’s profile through the analysis of his source code. After such analysis the programmer shall be placed on a scale that characterizes him on his language abilities. There are several potential applications for such profiling, namely, the evaluation of a programmer’s skills and proficiency on a given language or the continuous evaluation of a student’s progress on a programming course. Throughout the course of this project and as a proof of concept, a tool that allows the automatic profiling of a Java programmer is under development. This tool is also introduced in the paper and its preliminary outcomes are discussed.

Détection de vulnérabilités appliquée à la vérification de code intermédiaire de Java Card

La vérification de la résistance aux attaques des implémentations embarquées des vérifieurs de code intermédiaire Java Card est une tâche complexe. Les méthodes actuelles n'étant pas suffisamment efficaces, seule la génération de tests manuelle est possible. Pour automatiser ce processus, nous proposons une méthode appelée VTG (Vulnerability Test Generation, génération de tests de vulnérabilité). En se basant sur une représentation formelle des comportements fonctionnels du système sous test, un ensemble de tests d'intrusions est généré. Cette méthode s'inspire des techniques de mutation et de test à base de modèle. Dans un premier temps, le modèle est muté selon des règles que nous avons définies afin de représenter les potentielles attaques. Les tests sont ensuite extraits à partir des modèles mutants. Deux modèles Event-B ont été proposés. Le premier représente les contraintes structurelles des fichiers d'application Java Card. Le VTG permet en quelques secondes de générer des centaines de tests abstraits. Le second modèle est composé de 66 événements permettant de représenter 61 instructions Java Card. La mutation est effectuée en quelques secondes. L'extraction des tests permet de générer 223 tests en 45 min. Chaque test permet de vérifier une précondition ou une combinaison de préconditions d'une instruction. Cette méthode nous a permis de tester différents mécanismes d'implémentations de vérifieur de code intermédiaire Java Card. Bien que développée pour notre cas d'étude, la méthode proposée est générique et a été appliquée à d'autres cas d'études.

Some aspects of grading Java code submissions in MOOCs

Recently, massive open online courses (MOOCs) have been offering a new online approach in the field of distance learning and online education. A typical MOOC course consists of video lectures, reading material and easily accessible tests for students. For a computer programming course, it is important to provide interactive, dynamic, online coding exercises and more complex programming assignments for learners. It is expedient for the students to receive prompt feedback on their coding submissions. Although MOOC automated programme evaluation subsystem is capable of assessing source programme files that are in learning management systems, in MOOC systems there is a grader that is responsible for evaluating students’ assignments with the result that course staff would be required to assess thousands of programmes submitted by the participants of the course without the benefit of an automatic grader. This paper presents a new concept for grading programming submissions of students and improved techniques based on the Java unit testing framework that enables automatic grading of code chunks. Some examples are also given such as the creation of unique exercises by dynamically generating the parameters of the assignment in a MOOC programming course combined with the kind of coding style recognition to teach coding standards.

Wireless Network Analytics for Next Generation Spectrum Awareness

The importance of networks, in their broad sense, is rapidly and massively growing in modern-day society thanks to unprecedented communication capabilities offered by technology. In this context, the radio spectrum will be a primary resource to be preserved and not wasted. Therefore, the need for intelligent and automatic systems for in-depth spectrum analysis and monitoring will pave the way for a new set of opportunities and potential challenges. This thesis proposes a novel framework for automatic spectrum patrolling and the extraction of wireless network analytics. It aims to enhance the physical layer security of next generation wireless networks through the extraction and the analysis of dedicated analytical features. The framework consists of a spectrum sensing phase, carried out by a patrol composed of numerous radio-frequency (RF) sensing devices, followed by the extraction of a set of wireless network analytics. The methodology developed is blind, allowing spectrum sensing and analytics extraction of a network whose key features (i.e., number of nodes, physical layer signals, medium access protocol (MAC) and routing protocols) are unknown. Because of the wireless medium, over-the-air signals captured by the sensors are mixed; therefore, blind source separation (BSS) and measurement association are used to estimate the number of sources and separate the traffic patterns. After the separation, we put together a set of methodologies for extracting useful features of the wireless network, i.e., its logical topology, the application-level traffic patterns generated by the nodes, and their position. The whole framework is validated on an ad-hoc wireless network accounting for MAC protocol, packet collisions, nodes mobility, the spatial density of sensors, and channel impairments, such as path-loss, shadowing, and noise. The numerical results obtained by extensive and exhaustive simulations show that the proposed framework is consistent and can achieve the required performance.

Big Code Applications and Approaches

The availability of a huge amount of source code from code archives and open-source projects opens up the possibility to merge machine learning, programming languages, and software engineering research fields. This area is often referred to as Big Code where programming languages are treated instead of natural languages while different features and patterns of code can be exploited to perform many useful tasks and build supportive tools. Among all the possible applications which can be developed within the area of Big Code, the work presented in this research thesis mainly focuses on two particular tasks: the Programming Language Identification (PLI) and the Software Defect Prediction (SDP) for source codes. Programming language identification is commonly needed in program comprehension and it is usually performed directly by developers. However, when it comes at big scales, such as in widely used archives (GitHub, Software Heritage), automation of this task is desirable. To accomplish this aim, the problem is analyzed from different points of view (text and image-based learning approaches) and different models are created paying particular attention to their scalability. Software defect prediction is a fundamental step in software development for improving quality and assuring the reliability of software products. In the past, defects were searched by manual inspection or using automatic static and dynamic analyzers. Now, the automation of this task can be tackled using learning approaches that can speed up and improve related procedures. Here, two models have been built and analyzed to detect some of the commonest bugs and errors at different code granularity levels (file and method levels). Exploited data and models’ architectures are analyzed and described in detail. Quantitative and qualitative results are reported for both PLI and SDP tasks while differences and similarities concerning other related works are discussed.

Virtual commissioning of automatic machines for mesh panels manufacturing

The purpose of this thesis is to present the concept of simulation for automatic machines and how it might be used to test and debug software implemented for an automatic machine. The simulation is used to detect errors and allow corrections of the code before the machine has been built. Simulation permits testing different solutions and improving the software to get an optimized one. Additionally, simulation can be used to keep track of a machine after the installation in order to improve the production process during the machine’s life cycle. The central argument of this project is discussing the advantage of using virtual commissioning to test the implemented software in a virtual environment. Such an environment is getting benefit in avoiding potential damages as well as reduction of time to have the machine ready to work. Also, the use of virtual commissioning allows testing different solutions without high losses of time and money. Subsequently, an optimized solution could be found after testing different proposed solutions. The software implemented is based on the Object-Oriented Programming paradigm which implies different features such as encapsulation, modularity, and reusability of the code. Therefore, this way of programming helps to get simplified code that is easier to be understood and debugged as well as its high efficiency. Finally, different communication protocols are implemented in order to allow communication between the real plant and the simulation model. By the outcome that this communication provides, we might be able to gather all the necessary data for the simulation and the analysis, in real-time, of the production process in a way to improve it during the machine life cycle.

Quality and Aspect based Argument Generation

Natural Language Processing has always been one of the most popular topics in Artificial Intelligence. Argument-related research in NLP, such as argument detection, argument mining and argument generation, has been popular, especially in recent years. In our daily lives, we use arguments to express ourselves. The quality of arguments heavily impacts the effectiveness of our communications with others. In professional fields, such as legislation and academic areas, arguments of good quality play an even more critical role. Therefore, argument generation with good quality is a challenging research task that is also of great importance in NLP. The aim of this work is to investigate the automatic generation of arguments with good quality, according to the given topic, stance and aspect (control codes). To achieve this goal, a module based on BERT [17] which could judge an argument's quality is constructed. This module is used to assess the quality of the generated arguments. Another module based on GPT-2 [19] is implemented to generate arguments. Stances and aspects are also used as guidance when generating arguments. After combining all these models and techniques, the ranks of the generated arguments could be acquired to evaluate the final performance. This dissertation describes the architecture and experimental setup, analyzes the results of our experimentation, and discusses future directions.

NLG-Metricverse: An End-to-End Library for Evaluating Natural Language Generation

Driven by recent deep learning breakthroughs, natural language generation (NLG) models have been at the center of steady progress in the last few years. However, since our ability to generate human-indistinguishable artificial text lags behind our capacity to assess it, it is paramount to develop and apply even better automatic evaluation metrics. To facilitate researchers to judge the effectiveness of their models broadly, we suggest NLG-Metricverse—an end-to-end open-source library for NLG evaluation based on Python. This framework provides a living collection of NLG metrics in a unified and easy- to-use environment, supplying tools to efficiently apply, analyze, compare, and visualize them. This includes (i) the extensive support of heterogeneous automatic metrics with n-arity management, (ii) the meta-evaluation upon individual performance, metric-metric and metric-human correlations, (iii) graphical interpretations for helping humans better gain score intuitions, (iv) formal categorization and convenient documentation to accelerate metrics understanding. NLG-Metricverse aims to increase the comparability and replicability of NLG research, hopefully stimulating new contributions in the area.