BACHELOR THESIS - AUTOMATED DESIGN-SPACE EXPLORATION USING QUANTIFIER-FREE BIT-VECTOR LOGIC (M/F/D)
fortiss is the research institute of the Free State of Bavaria for the development of software-intensive systems with headquarters in Munich. The scientists at the institute cooperate in research, development and transfer projects with universities and technology companies in Bavaria, Germany and Europe. The focus is on research into state-of-the-art methods, techniques and tools for the development of software- and AI-based technologies for dependable, secure cyber-physical systems such as the Internet of Things (IoT). fortiss is organized in the legal form of a non-profit limited liability company. Shareholders are the Free State of Bavaria (majority shareholder) and the Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. www.fortiss.org
With cyber-physical systems becoming increasingly complex w.r.t. both hardware and software, model-based systems engineering has proven successful in coping with the intricacy of such software-intensive systems. However, with the separation of hardware and software comes the problem of system integration, which due to the growing scale has become increasingly time-consuming and error-prone when performed manually. While efforts to automate this step have successfully applied design-space exploration to synthesize optimal deployments of tasks onto execution units, finding a scalable formulation of the NP-hard deployment problem using satisfiability modulo theories has proven itself to be challenging.
The so-called deployment problem, consisting of the allocation of tasks onto execution units based on resource and safety constraints, is currently solved in AutoFOCUS 3 (https://af3.fortiss.org/main-features/design-space-exploration/) by formulating it as a satisfiability modulo theory (SMT) problem using free (uninterpreted) functions and quantifiers. This approach results in human-readable representations of the constraint-satisfaction problem and has been successfully applied to industry-size deployment problems. However, there is reason to expect considerable performance improvements by formulating SMT problems using solely quantifier-free bit-vector logic.
Unlike propositional logic, first-order logic is undecidable, i.e. there is no decision procedure that can correctly determine whether an arbitrary decision problem is satisfiable or not. Hence, while SMT solvers offer such decision procedures for a combination of quantifier-free theories, once a problem includes quantifiers (and thereby first-order logic) the solver might not be able to determine its satisfiability. Moreover, the definition of uninterpreted functions over infinite sets leads to unbounded search spaces – again leading to undecidability and requiring sophisticated automated reasoning techniques instead of the powerful, yet efficient methods available for bounded model checking.
In this project, we search for a new team member:
Bachelor Thesis – Automated Design-Space Exploration using Quantifier-free Bit-Vector Logic (m/f/d)
- Introduction to the concepts of DSE (Design-Space Exploration), SMT (Satisfiability Modulo Theory
- Implementation of the quantifier-free bit-vector-based DSEML-to-SMT translator in AutoFOCUS 3
- Evaluation of the approach by comparison with the existing translations based on uninterpreted functions
- You are a computer science (or similar) bachelor’s student
- You enjoy solving challenging problems
- You have practical experience with Python and Java development, software versioning (Git in particular)
- You have basic knowledge of logic in the context of computer science (such as propositional and first-order logic)
- You have excellent communication skills in English
- Varied and future-oriented field of activity
- Dynamic and team-oriented working atmosphere
- Close collaborations with leading research groups and industrial technology leaders for a practical development work
- Please note that this Bachelor thesis is not remunerated