Submitted by Lucienne Costa (lucienneferreira@ufcg.edu.br) on 2018-09-13T20:44:41Z No. of bitstreams: 1 REUDISMAM ROLIM DE SOUSA – TESE (PPGCC) 2018.pdf: 4395945 bytes, checksum: 2241c8bad2cdc8eda86eb53c2e64c227 (MD5) / Made available in DSpace on 2018-09-13T20:44:41Z (GMT). No. of bitstreams: 1 REUDISMAM ROLIM DE SOUSA – TESE (PPGCC) 2018.pdf: 4395945 bytes, checksum: 2241c8bad2cdc8eda86eb53c2e64c227 (MD5) Previous issue date: 2018-08-02 / Capes / Ferramentas como ErrorProne, ReSharper e PMD ajudam os programadores a detectar e/ou remover automaticamente vários padrões de códigos suspeitos, possíveis bugs ou estilo de código incorreto. Essas regras podem ser expressas como quick fixes que detectam e reescrevem padrões de código indesejados. No entanto, estender seus catálogos de regras é complexo e demorado. Nesse contexto, os programadores podem querer executar uma edição repetitiva automaticamente para melhorar sua produtividade, mas as ferramentas disponíveis não a suportam. Além disso, os projetistas de ferramentas podem querer identificar regras úteis para automatizarem. Fenômeno semelhante ocorre em sistemas de tutoria inteligente, onde os instrutores escrevem transformações complicadas que descrevem "falhas comuns" para consertar submissões semelhantes de estudantes a tarefas de programação. Nesta tese, apresentamos duas técnicas. REFAZER, uma técnica para gerar automaticamente transformações de programa. Também propomos REVISAR, nossa técnica para aprender quick fixes em repositórios. Nós instanciamos e avaliamos REFAZER em dois domínios. Primeiro, dados exemplos de edições de código dos alunos para corrigir submissões de tarefas incorretas, aprendemos transformações para corrigir envios de outros alunos com falhas semelhantes. Em nossa avaliação em quatro tarefas de programação de setecentos e vinte alunos, nossa técnica ajudou a corrigir submissões incorretas para 87% dos alunos. No segundo domínio, usamos edições de código repetitivas aplicadas por desenvolvedores ao mesmo projeto para sintetizar a transformação de programa que aplica essas edições a outros locais no código. Em nossa avaliação em 56 cenários de edições repetitivas de três grandes projetos de código aberto em C#, REFAZER aprendeu a transformação pretendida em 84% dos casos e usou apenas 2.9 exemplos em média. Para avaliar REVISAR, selecionamos 9 projetos e REVISAR aprendeu 920 transformações entre projetos. Atuamos como projetistas de ferramentas, inspecionamos as 381 transformações mais comuns e classificamos 32 como quick fixes. Para avaliar a qualidade das quick fixes, realizamos uma survey com 164 programadores de 124 projetos, com os 10 quick fixes que apareceram em mais projetos. Os programadores suportaram 9 (90%) quick fixes. Enviamos 20 pull requests aplicando quick fixes em 9 projetos e, no momento da escrita, os programadores apoiaram 17 (85%) e aceitaram 10 delas. / Tools such as ErrorProne, ReSharper, and PMD help programmers by automatically detecting and/or removing several suspicious code patterns, potential bugs, or instances of bad code style. These rules could be expressed as quick fixes that detect and rewrite unwanted code patterns. However, extending their catalogs of rules is complex and time-consuming. In this context, programmers may want to perform a repetitive edit into their code automatically to improve their productivity, but available tools do not support it. In addition, tool designers may want to identify rules helpful to be automated. A similar phenomenon appears in intelligent tutoring systems where instructors have to write cumbersome code transformations that describe “common faults” to fix similar student submissions to programming assignments. In this thesis, we present two techniques. REFAZER, a technique for automatically generating program transformations. We also propose REVISAR, our technique for learning quick fixes from code repositories. We instantiate and evaluate REFAZER in two domains. First, given examples of code edits used by students to fix incorrect programming assignment submissions, we learn program transformations that can fix other students’ submissions with similar faults. In our evaluation conducted on four programming tasks performed by seven hundred and twenty students, our technique helped to fix incorrect submissions for 87% of the students. In the second domain, we use repetitive code edits applied by developers to the same project to synthesize a program transformation that applies these edits to other locations in the code. In our evaluation conducted on 56 scenarios of repetitive edits taken from three large C# open-source projects, REFAZER learns the intended program transformation in 84% of the cases and using only 2.9 examples on average. To evaluate REVISAR, we select 9 projects, and REVISAR learns 920 transformations across projects. We acted as tool designers, inspected the most common 381 transformations and classified 32 as quick fixes. To assess the quality of the quick fixes, we performed a survey with 164 programmers from 124 projects, showing the 10 quick fixes that appeared in most projects. Programmers supported 9 (90%) quick fixes. We submitted 20 pull requests applying our quick fixes to 9 projects and, at the time of the writing, programmers supported 17 (85%) and accepted 10 of them.

[pt] Garantir que programas são implementados de forma a cumprir uma especificação é uma questão fundamental em computação, por isso, têm sido propostos vários métodos que almejam provar a correção dos programas. Este trabalho apresenta um método, baseado no isomorfismo de Curry-Howard, que extrai conteúdos computacionais de provas intuicionistas, conhecido como síntese construtiva ou proofs-as-programs. É proposto um processo de síntese construtiva de programas, onde a extração do conteúdo computacional gera um programa em linguagem imperativa a partir de uma prova em lógica intuicionista poli-sortida, cujos axiomas definem os tipos abstratos de dados, sendo utilizado como sistema dedutivo a Dedução Natural. Também é apresentada uma prova de correção, bem como uma prova de completude do método atráves do uso de um sistema com regra ômega (computacional) para a aritmética de Heyting, concluindo com uma demonstração da relação entre o uso da indução finita no lugar da regra ômega computacional no processo de síntese. / [en] One of the main problems in computer science is to assure that programs are implemented in such a way that they satisfy a given specification. There are many studies about methods to prove correctness of programs. This work presents a method, belonging to the constructive synthesis or proofs-as-programs paradigm, that comes from the Curry- Howard isomorphism and extracts the computational contents of intuitionist proofs. The synthesis process proposed produces a program in an imperative language from a proof in many-sorted intuitionist logic, where the axioms define the abstract data types using Natural Deduction as deductive system. It is proved the correctness, as well as the completeness of the method regarding the Heyting arithmetic with ômega-rule(in its computational version). A discussion about the use of the finitary induction instead of computational ômega-rule concludes the work.

The Hybrid Electric Vehicle (HEV) is a complex electromechanical system with complex interactions among various components. Due to the large number of design variables involved, the design flexibility in the HEV makes performance studies difficult. As the system complexity and sophistication increases, it becomes much more difficult to predict these interactions and design the system accordingly. Also, different variations in the design and manufacture of various components and systems involve a large amount of work and cost to keep updated of all these variations. While the above issues ask for a flexible design environment suitable for vehicle design, most of the existing powertrain design tools are based on experiential models, such as look-up tables, which use idealized assumptions and limited experimental data. The accuracy of the results produced by these tools is not good enough for designing these new generation vehicles. Also, sometimes the designs may lead to components or systems beyond physical limitations. To make the powertrain design more efficient, the models developed must be closely related to the underlying physics of the components. Only such physics-based models can facilitate high fidelity simulations for dynamics at different time scales. This results in the quest for a design tool that manages the vehicle?s development process while maintaining tight integration between the software and physical artifacts. The thesis addresses the above issues and focuses on the modeling of HEV using model integrated computing and employing physics-based resistive companion form modeling method. For this purpose, Generic Modeling Environment (GME), software developed by Institute of Software and Integrated Systems (ISIS), Vanderbilt University is used as the platform for developing the models. A modeling environment for hybrid vehicle design is prepared and a Battery Electric Vehicle (BEV) is developed as an application of the developed environment. Resistive companion form models of various BEV components are prepared and a model interpreter is prepared for integrating the developed component models and simulating the design.

In modern software development, maintenance accounts for the majority of the total cost and effort in a software project. Especially burdensome are those tasks which require applying a new technology in order to adapt an application to changed requirements or a different environment. This research explores methodologies, techniques, and approaches for automating such adaptive maintenance tasks. By combining high-level specifications and generative techniques, a new methodology shapes the design of approaches to automating adaptive maintenance tasks in the application domains of high performance computing (HPC) and enterprise software. Despite the vast differences of these domains and their respective requirements, each approach is shown to be effective at alleviating their adaptive maintenance burden. This thesis proves that it is possible to effectively automate tedious and error-prone adaptive maintenance tasks in a diverse set of domains by exploiting high-level specifications to synthesize specialized low-level code. The specific contributions of this thesis are as follows: (1) a common methodology for designing automated approaches to adaptive maintenance, (2) a novel approach to automating the generation of efficient marshaling logic for HPC applications from a high-level visual model, and (3) a novel approach to automatically upgrading legacy enterprise applications to use annotation-based frameworks. The technical contributions of this thesis have been realized in two software tools for automated adaptive maintenance: MPI Serializer, a marshaling logic generator for MPI applications, and Rosemari, an inference and transformation engine for upgrading enterprise applications. This thesis is based on research papers accepted to IPDPS '08 and OOPSLA '08. / Master of Science

In the world that we live in today, we greatly rely on software in nearly every aspect of our lives. In many critical applications, such as in transportation and medical systems, catastrophic consequences could occur in case of buggy software. As the computational power and storage capacity of computer hardware keep increasing, so are the size and complexity of the software. This makes testing and verification increasingly challenging in practice, and consequentially creates a chance for software with critical bugs to find their way into the consumer market. In this dissertation, I present a set of innovative new methods for automatically verifying, as well as synthesizing, critical software and hardware in embedded computing applications. Based on a set of rigorous formal analysis techniques, my methods can guarantee that the resulting software are efficient and secure as well as provably correct. / Ph. D.

No description available.

<p>Program synthesis aims to generate programs automatically from user-provided specifications and has the potential to aid users in real-world programming tasks from different domains. Although there have been great achievements of synthesis techniques in specific domains such as spreadsheet programming, computer-aided education and software engineering, there still exist huge barriers that keep us from achieving scalable and practical synthesis tools.</p> <p><br></p> <p>This dissertation presents several techniques towards more scalable and practical program synthesis from three perspectives: 1) intention: Writing formal specification for synthesis is a major barrier for average programmers. In particular, in some quantitative synthesis scenarios (such as network design), the first challenge faced by users is expressing their optimization targets. To address this problem, we present comparative synthesis, an interactive synthesis framework that learns near optimal programs through comparative queries, without explicitly specified optimization targets. 2) invention: Synthesis algorithms are key to pushing the performance limit of program synthesis. Aiming to solve syntax-guided synthesis problems efficiently, we introduce a cooperative synthesis technique that combines the merits of enumerative and deductive synthesis. 3) adaptation: Besides functional correctness, quality of generated code is another important aspect. Towards automated provably-correct optimization over tree traversals, we propose a stack-based representation for iterations in tree traversals and an encoding to Monadic Second-Order logic over trees, which enables reasoning about tree traversal transformations which were not possible before.</p>

Automatically generating a program from its specification eliminates a large source of errors that is often unavoidable in a manual approach. While a general purpose code generator is impossible to build, it is possible to build a practical code generator for a specific domain. This thesis investigates the theory behind Booster — a domain specific, object based specification language and automatic code generator. The domain of Booster is information systems — systems that consist of a rich object model in which the objects refer to each other to form a complicated network of associations. The operations of such systems are conceptually simple (changing the attributes of objects, adding or removing new objects and creating or destroying associations) but they are tricky to implement correctly. The thesis focuses on the theoretical foundation of the Booster approach, in particular on three contributions: semantics, model completion, and code generation. The semantics of a Booster model is a single abstract data type (ADT) where the invariants and the methods of all the classes in the model are promoted to the level of the ADT. This is different from the traditional view that considers each class as a separate ADT. The thesis argues that the Booster semantics is a better model of object oriented systems. The second important contribution is the idea of model completion — a process that augments the postconditions of methods with additional predicates that follow from the system’s invariant and the method’s original intention. The third contribution describes a simple but effective code generation technique that is based on interpreting postconditions as executable statements and uses weakest preconditions to ensure that the generated code refines its specification.

[pt] Esta dissertação investiga o uso de programação genética linear na síntese automática de programas em linguagem de montagem para microcontroladores, que implementem estratégias de controle de tempo ótimo ou sub-ótimo, do sistema a ser controlado, a partir da modelagem matemática por equações dinâmicas. Uma das dificuldades encontradas no projeto convencional de um sistema de controle ótimo é que soluções para este tipo de problema normalmente implicam em uma função altamente não-linear das variáveis de estado do sistema. Como resultado, várias vezes não é possível encontrar uma solução matemática exata. Já na implementação, surge a dificuldade de se ter que programar manualmente o microcontrolador para executar o controle desejado. O objetivo deste trabalho foi, portanto, contornar tais dificuldades através de uma metodologia que, a partir da modelagem matemática de uma planta, fornece como resultado um programa em linguagem de montagem. O trabalho consistiu no estudo sobre os possíveis tipos de representações para a manipulação genética de programas em linguagem de montagem, tendo sido concluído que a linear é a mais adequada, e na implementação de uma ferramenta para realizar os três estudos de caso: water bath, cart centering e pêndulo invertido. O desempenho de controle dos programas sintetizados foi comparado com o dos sistemas obtidos por outros métodos (redes neurais, lógica fuzzy, sistemas neurofuzzy e programação genética). Os programas sintetizados demonstraram, no mínimo, o mesmo desempenho, mas com a vantagem adicional de fornecerem a solução já no formato final da plataforma de implementação escolhida: um microcontrolador. / [en] This dissertation investigates the use of genetic programming in automatic synthesis of assembly language programs for microcontrollers, which implement time-optimal or sub-optimal control strategies of the system to be controlled, from the mathematical modeling by dynamic equations. One of the issues faced in conventional design of an optimal control system is that solutions for this kind of problem commonly involve a highly nonlinear function of the state variables of the system. As a result, frequently it is not possible to find an exact mathematical solution. On the implementation side, the difficulty comes when one has to manually program the microcontroller to run the desired control. Thus, the objective of this work was to overcome these difficulties applying a methodology that, starting from the mathematical modeling of a plant, provides as result an assembly language microcontroller program. The work included a study of the possible types of genetic representation for the manipulation of assembly language programs. In this regard, it has been concluded that the linear is the most suitable representation. The work also included the implementation of a tool to accomplish three study cases: water bath, cart centering and inverted pendulum. The performance of control of the synthesized programs was compared with the one obtained by other methods (neural networks, fuzzy logic, neurofuzzy systems and genetic programming). The synthesized programs achieved at least the same performance of the other systems, with the additional advantage of already providing the solution in the final format of the chosen implementation platform: a microcontroller.

Interoperability development between Autonomous Mobile Mobots, Auto- mated Guided Vehicles and their fleet managers provided by different brands became a highly important topic in the factory and warehouse management industry. The compatibility between the providers would provide endless opportunities for the customers to build scalable, flexible, and dynamic systems. The thesis develops a program synthesis method that is capable of making conversions between data structures. The results provide the programmers with a tool that can be used to create function compositions by providing input-output pairs and can save a significant amount of time. The study solves the search problem using Context-free grammar and a beam search guided by a neural network. The work compares the performance of three different artificial intelligence structures recurrent neural networks, long short-term memory networks, and gated recurrent unit networks, and validates the solution on a test set that consists of lines from data structures that are used in the Autonomous Mobile Robot industry combined with some generated data to measure performance in regards of the generalization power. / Interoperabilitetsutveckling mellan Autonoma Mobila Robotar, Automatiserade Guidade Fordon och deras flottförvaltare från olika varumärken har blivit ett mycket viktigt ämne inom fabriks- och lagerhanteringsbranschen. Kompatibiliteten mellan leverantörerna skulle ge oändliga möjligheter för kunderna att bygga skalbara, flexibla och dynamiska system. Avhandlingen utvecklar en programssyntesmetod som kan göra omvandlingar mellan datastrukturer. Resultaten ger programmerare ett verktyg som kan användas för att skapa funktionskompositioner genom att tillhandahålla in- och utdata- par och kan spara betydande tid. Studien löser sökproblemet med hjälp av kontextfri grammatik och en strålsökning som styrs av ett neuralt nätverk. Arbetet jämför prestandan hos tre olika konstgjorda intelligensstrukturer: ”recurrent neural networks", ”long short-term memory networks", och ”gated recurrent unit networks", och validerar lösningen på en testuppsättning som består av rader från datastrukturer som används inom den autonoma mobila robotindustrin kombinerat med några genererade data för att mäta prestanda med avseende på generaliseringskraften.