36 research outputs found
Accelerated test execution using GPUs
Get PDFAs product life-cycles become shorter and the scale and complexity of systems increase, accelerating the execution of large test suites gains importance. Existing research has primarily focussed on techniques that reduce the size of the test suite. By contrast, we propose a technique that accelerates test execution, allowing test suites to run in a fraction of the original time, by parallel execution with a Graphics Processing Unit (GPU).
Program testing, which is in essence execution of the same program withmmultiple sets of test data, naturally exhibits the kind of data parallelism that can be exploited with GPUs. Our approach simultaneously executes the program with one test case per GPU thread. GPUs have severe limitations, and we discuss these in the context of our approach and define the scope of our applications. We observe speed-ups up to a factor of 27 compared to single-core execution on conventional CPUs with embedded systems benchmark programs
Applications of light-induced hyperpolarization in EPR and NMR
No full textMagnetic resonance methods are widely used to provide atomic level information on the structure and dynamics of chemical and biochemical systems, but often suffer from poor sensitivity. This review examines how optical excitation can provide increased electron spin-polarization, and how this can be used to increase sensitivity and/or information content in both Nuclear Magnetic Resonance (NMR) and Electron Paramagnetic Resonance (EPR) spectroscopy
Domain-specific languages for the design, deployment and manipulation of heterogeneous databases
Get PDFThe need for levels of availability and scalability beyond those supported by relational databases has led to the emergence of a new generation of purpose-specific databases grouped under the term NoSQL. In general, NoSQL databases are designed with horizontal scalability as a primary concern and deliver increased availability and fault tolerance at a cost of temporary inconsistency and reduced durability of data. To balance the requirements for data consistency and availability, organisations increasingly migrate towards hybrid data persistence architectures comprising both relational and NoSQL databases. The consensus is that this trend will only become stronger in the future; critical data will continue to be stored in ACID (largely relational) databases while non-critical data will be progressively migrated to high-availability NoSQL databases. Designing and deploying a hybrid data persistence architecture that involves a combination of relational and NoSQL databases is a complex, technically challenging and error-prone task. In this paper we outline a model-based methodology developed in the context of the EC-funded H2020 TYPHON project for designing, developing, querying and evolving such scalable architectures for persistence, analytics and monitoring of large volumes of hybrid (relational, graph-based, document-based, natural language, etc.) data, in a systematic and disciplined manner
Molecules as qubits, qudits and quantum gates
No full textQuantum coherence is a fundamental property of electron spins that could be exploited in quantum computers for the processing of information. Within quantum calculation protocols, information is encoded in two-level quantum objects (Qubits) and processed by the operations of logical quantum gates (or Qugates). Examples of qubits include electronic spins of magnetic molecules, which present the advantage that they can be easily manipulated by external electromagnetic fields, via electron paramagnetic resonance (EPR). Creation of multi-level systems (so-called Qudits, where d is the dimension of the Hilbert space) is also possible in magnetic molecules carrying both electronic and magnetic nuclear spins due to cooperative hyperfine interactions. Such qudits can simultaneously access a multitude of states reducing the number of iterations in quantum-computation algorithms. The implementation of the Grover's algorithm in a single molecular unit was experimentally probed in a multi-level molecular nanomagnet thanks to hyperfine transitions combined with the shielded nature of nuclear states that limits decoherence. This chapter gives an overview of the latest advances in the design and testing of molecular electron spin systems with Quantum Information Processing (QIP) attributes, while highlighting the tremendous progress made in electron spin manipulations driven by pulse EPR spectroscopy.</p
Verifying Feedforward Neural Networks for Classification in Isabelle/HOL
Get PDFThis is the author accepted manuscript. The final version is available from Springer via the DOI in this recordFormal Methods: 25th International Symposium, FM 2023, Lübeck, Germany, 6 - 10 March 2023Data Availability Statement:
The formalisation and case studies are available to view on Zenodo: https://doi.org/10.5281/zenodo.7418170. The materials include both the Isabelle/HOL implementation and the detailed documentation generated by Isabelle.Neural networks are being used successfully to solve classification problems, e.g., for detecting objects in images. It is well known that neural networks are susceptible if small changes applied to their input result in misclassification. Situations in which such a slight input change, often hardly noticeable by a human expert, results in a misclassification are called adversarial examples. If such inputs are used for adversarial attacks, they can be life-threatening if, for example, they occur in image classification systems used in autonomous cars or medical diagnosis.
Systems employing neural networks, e.g., for safety or security-critical functionality, are a particular challenge for formal verification, which usually expects a formal specification (e.g., given as source code in a programming language for which a formal semantics exists). Such a formal specification does, per se, not exist for neural networks.
In this paper, we address this challenge by presenting a formal embedding of feedforward neural networks into Isabelle/HOL and discussing desirable properties for neural networks in critical applications. Our Isabelle-based prototype can import neural networks trained in TensorFlow, and we demonstrate our approach using a neural network trained for the classification of digits on a dot-matrix display.Engineering and Physical Sciences Research Council (EPSRC
Parenthood and Gender in Russian Academia
Get PDFBased on data from a survey conducted in November 2021 involving 2,395 economists working in the academic field, the impact of parenthood on both mothers and fathers on elements of academic careers, such as defending a doctoral thesis and publishing productivity, is being analyzed.
It has been observed that men successfully balance the arrival of their first child with the defense of their doctoral thesis, while for most women, these events occur sequentially with a time gap. Furthermore, the strategies for combining parenthood with building a professional career among young respondents differ from those of their older colleagues. The older the woman in our sample, the more likely she is to prioritize the "child first, then defense" strategy. Over time, priorities have shifted, and now women are more inclined to postpone childbirth.
Using data from the elibrary.ru library, indicators of respondents' publication productivity were obtained, indicating that, on average, men publish more works indexed in RINC and the RINC core. A similar publication gap has been identified in other countries, with researchers suggesting unequal burdens that women and men bear during parenthood as a possible explanation for this gap. The results of the study do not provide grounds to consider the presence or absence of children as a significant factor explaining this gap. The author considers self-selection among respondents due to barriers within the academic environment and differences in career goals as possible reasons for the publication gap. Both women with and without children are relatively more oriented towards a teaching career, and the presence of a child may strengthen this tendency
Transformations of Software Product Lines: A Generalizing Framework Based on Category Theory
No full textA Variability-Based Approach to Reusable and Efficient Model Transformations
No full textLarge model transformation systems often contain transformation rules that are substantially similar to each other, causing performance bottlenecks for systems in which rules are applied nondeterministically, as long as one of them is applicable. We tackle this problem by introducing variability-based graph trans-formations. We formally define variability-based rules and contribute a novel match-finding algorithm for applying them. We prove correctness of our approach by showing its equivalence to the classic one of applying the rules individually, and demonstrate the achieved performance speed-up on a realistic transformation scenario
