Leibniz ScienceCampus SCALES

Classical methods often fail to fully capture the interactions across these scales: feedback mechanisms are neglected, leading to inaccurate models and flawed predictions.

Our approach combines several strategies: large-scale simulations using novel computational methods, hybrid models in which local simulations complement larger systems, and physics-informed neural networks to describe local processes. A central element is the integration of data-driven methods that leverage observations and high-resolution simulations to continuously refine models and identify new phenomena. The applications considered range from astrophysics (e.g., galaxy formation, neutron star mergers, exoplanets) to Earth system science, including turbulence, climate, and the water cycle.

Achieving breakthroughs in the description of such multi-scale physical systems requires close collaboration across different scientific disciplines. At the same time, the project makes an important contribution to addressing global challenges and strengthens interdisciplinary collaboration in the Potsdam-Berlin region.