A team led by Professor Stefaan Poedts used VSC’s supercomputing power to model the May 2024 solar eruptions from the Sun to Earth. Through advanced MHD simulations, they revealed how magnetic energy buildup triggers solar flares and coronal mass ejections, improving space weather forecasts and contributing to more accurate predictions of geomagnetic storms.

Plasma catalysis offers a sustainable route to convert CO₂ into valuable chemicals using renewable electricity. Researchers used VSC’s computational resources to model plasma–catalyst interactions and reaction kinetics on glass and transition metal surfaces. Their findings reveal how catalyst choice and reaction conditions influence CO₂ splitting efficiency and product formation, advancing understanding of plasma-driven CO₂ conversion.

Researchers from KU Leuven and UHasselt demonstrated that personalized federated learning (PFL) can significantly improve predictions of disability progression in Multiple Sclerosis (MS). Using data from over 26,000 patients, the team developed novel approaches like AdaptiveDualBranchNet, balancing global collaboration with local adaptation. The work was made possible by the VSC’s high-performance infrastructure, enabling large-scale federated experiments.

Reading is shaped not only by word length and frequency but also by how predictable a word is in context. Using Dutch language models, researchers show that smaller models often predict reading times better and that predictability follows a logarithmic effect. With VSC’s computing power, large-scale analyses became possible, revealing new insights into how humans and machines process language.

Researchers from Ghent University have developed a new computational method that accurately predicts the exotic properties of strongly correlated materials, such as high-temperature superconductors. By combining downfolding techniques with tensor networks, and using the power of VSC supercomputing, the team achieved highly precise simulations of electronic structures—opening new paths for quantum devices and next-generation electronics.

Functional dependencies describe strong relationships in relational databases, helping ensure data consistency and integrity. When data is incomplete or contains errors, approximate functional dependencies (AFDs) become essential to detect patterns that “almost hold.” This study evaluates multiple AFD measures, identifying effective methods for discovering hidden dependencies in real-world large-scale datasets.

Why do some offenders repeatedly team up with the same partners—and what does this mean for what they learn? This study uses agent-based modeling to explore how homophily, or the preference for similar partners, shapes co-offending networks. The findings show that while homophily fosters trust and stable ties, it also slows the flow of criminal skills across the network, revealing a trade-off between stability and broader learning.

This study explores how side branches in acute Type-B aortic dissections affect pressure and wall stabilization using fluid-solid interaction (FSI) simulations. By modeling blood flow and tissue dynamics in an idealized aorta, the research reveals that side branches—especially from the false lumen—can increase systolic pressure and influence membrane displacement. High-performance computing at VSC enabled this complex analysis.

Topology optimization unlocks lightweight, high-performance designs, but poor connectivity often limits manufacturability and reliability. This review explores methods to promote structural continuity, comparing physics- and geometry-based approaches, and offering practical guidelines to bridge the gap between theory and real-world engineering.

Millions of today’s children face unprecedented lifetime exposure to heatwaves, floods, droughts, and storms if global warming reaches 3.5 °C. Deep emission cuts to meet the Paris 1.5 °C goal could protect millions. Urgent climate action is needed to safeguard children’s futures worldwide.