Computer models in the geosciences have become essential tools to predict the behavior of environmental systems and address societal concerns, such as the impact of climate change on river flooding risks or the contamination of groundwater used for drinking water supply. As a smart but coarse abstraction of real-world systems, geomodels allow for identifying dominant processes, making future predictions, and investigating system reactions to changing boundary conditions.
Overcoming Limitations of Traditional Geomodels
Typically, such models are built with a specific purpose in mind, which drastically limits their applicability and transferability. “The ‘GeoMod4Future’ project aims to overcome the limitations of current geomodelling practices by developing a fundamentally new, open-purpose model type: neural stars”, explains Dr. Anneli Guthke, Junior Research Group Leader for Statistical Model-Data Integration at the Stuttgart Center for Simulation Science (SC SimTech). By leveraging innovative machine learning methods, these new models are designed to answer unforeseen questions, saving resources in model development and modification, and enabling efficient knowledge management and extension.
Pioneering with Vector Foundation Support
The funding of this research idea in a very early phase by the Vector Foundation underscores the importance and potential of unconventional and high-risk research approaches in the STEM field. "The development of neural stars represents a significant leap forward in our ability to create versatile and efficient models. This innovation has the potential to revolutionize how we approach and solve complex environmental problems," underlines Dr. Anneli Guthke. The foundation supports bold and high-risk research projects at universities and research institutions in Baden-Württemberg with up to 100,000 euros for a maximum duration of 24 months. With this funding, a first proof-of-concept shall pave the ground for follow-up PhD projects and create visibility for this new modelling philosophy in an internationally active research field.
Cluster of Excellence: Advancing SimTech's Vision for Future Simulations
The GeoMod4Future project at the University of Stuttgart is embedded in the Cluster of Excellence Data-Integrated Simulation Science (SimTech). GeoMod4Future aligns perfectly with SimTech's mission to advance simulation technologies through the integration of data and cutting-edge computational methods. The project embodies this mission by pioneering the use of neural stars to create adaptable, data-driven models. This approach not only enhances the flexibility and scope of geomodelling but also contributes to SimTech's overarching goal of improving predictive accuracy and efficiency in simulations across various scientific and engineering domains. By pushing the boundaries of traditional modelling, GeoMod4Future supports SimTech's vision of transforming how simulations are conducted, ultimately leading to more resilient and informed decision-making processes in the face of complex global challenges.
Expert Contact:
Anneli Guthke, University of Stuttgart, Stuttgart Center for Simulation Science (SC SimTech), Tel.: +49 711 685-60157, E-Mail: anneli.guthke@simtech.uni-stuttgart.de
About the Cluster of Excellence SimTech
SimTech, the Cluster of Excellence “Data-Integrated Simulation Science”, is a leading research institution dedicated to advancing the field of simulation science through interdisciplinary collaboration and innovation. It provides a vibrant research environment that fosters the development of new simulation methods and technologies, integrating expertise from various scientific and engineering disciplines.
MINT-Innovations 2024: University of Stuttgart successful with six projects
The Stuttgart-based Vector Foundation regularly awards start-up funding of up to 100,000 euros for pioneering research projects in the disciplines of mathematics, computer science, natural sciences and technology as part of the "MINT Innovations" call. The funding is intended to enable preliminary work for particularly unconventional and risky research ideas, which, if successful, will facilitate subsequent integration into the current funding landscape. In May 2024, the Vector Foundation selected a total of six projects from the University of Stuttgart for funding:
Ammonia2H2 Institut für Photovoltaik / Mirja Mannigel
MULTIVIEW Künstliche Intelligenz für ausgewogene Nachrichtenempfehlung Institut für maschinelle Sprachverarbeitung / Amelie Wührl, Tanise Ceron
CarbEx Solid Carbon Extraction from Energized Atmospheric CO2 Institut für Raumfahrtsysteme / Hendrik Burghaus, Dr. Georg Herdrich
Nanophotonik im ultravioletten Spektralbereich Physikalisches Institut / Prof. Dr. Harald Gießen, Dr. Mario Hentschel
Sensor-Schäume Einfacher Nachweis von Kontaminationen Institut für Physikalische Chemie / Dr. Tamara Schad, Dr. Cosima Stubenrauch
GeoMod4Future Zukunftsfähige Modellierung für die Geowissenschaften Stuttgart Center for Simulation Science (SC SimTech) / Dr. Anneli Guthke, PD Dr. Uwe Ehret, KIT