Program correctness is a central problem in computer science. Code inspection and testing can reveal many program bugs, but subtle errors need a rigorous analysis. A fully automated analysis is impossible: deciding whether a program terminates on a given input is undecidable. Thanks to unremitting developments in program verification and incredible advancements in satisfiability checking, program verification is nowadays supported by software tools in industrial practice. Meta and Amazon Web Services use program verification tools on a daily basis. In the advent of AI, probabilistic programming emerged as a popular paradigm combining programming with learning from (big) data. Since 2018, the UN uses such probabilistic programs to predict the location and classify seismological activities on the earth. Other application areas include security, planning in AI, cognitive science, and neural network training. Probabilistic programs are fundamentally different. Due to randomness, they sometimes terminate and sometimes not. Their outcome depends on coin flips. They may terminate with probability one, while having an infinite expected run time. Classical program verification techniques no longer apply. The ERC project FRAPPANT has resulted in proof calculi for probabilistic programs, equipped with powerful proof rules, and identified a relative complete syntax for quantitative properties. This has led to a prototypical deductive verifier for an “assembler” programming language. A software tool for which no equivalent exists. Successful analyses of intricate programs showed its potential. The proposed project aims to explore the commercial and innovative aspects of our deductive verifier. It takes the necessary innovative steps to enable a commercialisation by including invariant synthesis and program slicing and supporting the popular probabilistic programming language STAN. Its potential will be investigated engaging potential users, and a market analysis.

Chemical energy carriers will play an essential role for future energy systems, where harvesting and utilization of renewable energy occur not necessarily at the same time or place, hence long-time storage and long-range transport of energy are needed. For this, hydrogen-based energy carriers, such as hydrogen and ammonia, hold great promise. Their utilization by combustion-based energy conversion has many advantages, e.g., versatile use for heat and power, robust and flexible technologies, and its suitability for a continuous energy transition. However, combustion of both hydrogen and ammonia is very challenging. For technically relevant conditions, both form intrinsic, so-called thermo-diffusive instabilities (very different from the often-discussed thermo-acoustic instabilities), which can increase burn rates by a stunning factor of three to five! Without considering this, computational design is impossible. Yet, while linear theories exist, little is understood for the more relevant non-linear regime, and beyond some data and observations, virtually nothing is known about the interactions of intrinsic flame instabilities (IFI) with turbulence. Here, rigorous analysis of new data for neat H2 and NH3/H2-blends from simulations and experiments will lead to a quantitative understanding of the relevant aspects. From this, a novel modeling framework with uncertainty estimates will be developed. The key hypothesis then is that combustion processes of hydrogen-based fuels can be improved by targeted weakening or promotion of IFI, and that this kind of instability-controlled combustion can jointly improve efficiency, emissions, stability, and fuel flexibility in different combustion devices, such as spark-ignition engines, gas turbines, and industrial burners. Guided by the developed knowledge and tools, this intrinsic-flame-instability-controlled combustion concept will be demonstrated computationally and experimentally for two sample applications.

This project will develop a unifying framework of novel methods for sequence classification and thus make a major break-through in automatic speech recognition and machine translation, advancing these areas of human language technology (HLT) beyond state-of-the-art. Despite the huge progress made in the field, the specific aspect of sequence classification has not been addressed adequately in the past research in these disciplines and remains a big challenge. The proposed project will provide a novel framework under consistent consideration of the leading aspect of sequence classification. It will break the ground for a deeper, more comprehensive foundation for sequence classification and pave the way for a new generation of algorithms that will put human language technology on a more solid basis and that will accelerate progress in the field across several disciplines. The leading research objectives are: 1. A novel theoretical framework for sequence classification. 2. Consistent sequence modeling across training and testing, which is specifically lacking in machine translation. 3. Adequate sequence-level performance-aware training criteria to learn the free parameters of the models. 4. Investigation of (true) unsupervised training for HLT sequence classification: its principles, its prerequisites, its limitations and its practical usage. The study of these four problems will provide key enabling techniques for HLT sequence classification in general that will carry over to and create high impact on the areas of speech recognition, machine translation and handwritten text recognition. Using our top-ranking research prototype systems, we will verify the validity and effectiveness or our research on public international benchmarks.

Functionalized amines are high-value materials with numerous applications as therapeutic agents, agrochemicals and organic materials. The development of strategies for enabling the fast and divergent synthesis of these material can have a positive impact to our ability to discover, manufacture and evolve molecules of high-relevance to our society. In this proposal we present an innovative approach for the functionalization of homoallylic amines based on their coordination borane (BH3). This will provide access to novel boryl radical species that we will exploit in cyclization-functionalization cascade. This novel reactivity mode will convert the homoallylic amine into a cyclic borylated and functionalized building block that can be further diversified across the sp3 C–B bond by oxidation or Suzuki-Miyaura cross-coupling. Overall, this strategy will constitute a divergent platform for the diversification of high-value amines and also a novel retrosynthetic tactic for the preparation of high-value and structurally complex drug analogues. This research squarely fits within the expertise of the Leonori group in the generation and use of boryl radicals in synthesis and catalysis. The completion of such an innovative and ambitious project at RWTH Aachen University will be facilitated by generating, transferring, sharing and disseminating knowledge, and will enhance the Researcher future career following the training plan envisioned.