Two out of three awards at ICCS 2026 went to JSC colleagues

Two papers by JSC authors received awards at the 26th International Conference on Computational Science (ICCS), which took place in Hamburg from 29 June until 1 July 2026: Adel Dabah and Andreas Herten received the “Best Main Track Paper Award” for their paper about “Efficient Accelerated Graph Edit Distance Computation on GPU”. Chelsea John, Andreas Herten, Stefan Kesselheim and their collaborators from Technical University of Hamburg received the “Best Poster Award” for their work about “Fourier Neural Operators for Rayleigh-Bénard Convection”.

Adel Dabah and Andreas Herten present a pioneering GPU-accelerated approach for computing Graph Edit Distance (GED), a fundamental similarity measure that quantifies the minimal editing path to transform one labeled graph into another. This is a critical operation in domains such as bioinformatics, machine learning, and pattern recognition. However, the high computational complexity of optimal and near-optimal methods limits their applicability to large-scale graphs, making parallel GED computation on HPC systems essential. To address this, the authors propose FAST-GED, a fast and scalable open-source framework for GED computation on GPUs. FAST-GED overcomes existing limitations by combining high accuracy with fast execution through GPU-friendly algorithmic design and efficient mapping to GPU hardware, minimizing host-device communication. The implementation is optimized and tested across multiple GPU architectures. Adel Dabah and Andreas Herten validate FAST-GED on real and synthetic datasets with diverse graph sizes and densities. It achieves speedups of several orders of magnitude over the Python NetworkX library while reaching optimal solutions in most cases. Moreover, it outperforms state-of-the-art approximate methods in both accuracy and scalability. The paper shows that FAST-GED enables broader adoption of GED-based solutions in real-world applications.

Modeling turbulent convection is challenging and has applications from atmospheric flows to industrial processes such as silicon wafer production. A standard benchmark is the Rayleigh–Bénard Convection (RBC), where a fluid heated from below develops an overturning circulation with the strength of turbulence governed by the Rayleigh number (Ra). At high Ra the flow becomes strongly turbulent, making high-resolution numerical simulations computationally expensive. Recent machine learning approaches, particularly Fourier Neural Operators (FNOs), provide an alternative to mesh-based solvers. In their paper, Chelsea John and her co-authors describe how to apply FNOs to two-dimensional RBC, focusing on stable small-step predictions that preserve turbulent statistics and align with time-stepping solvers. The resulting model is compact and fast, while maintaining similar accuracy as demonstrated in previous benchmarks. The authors show show that although FNOs generalize to finer meshes, accuracy remains limited by the resolution of the training data.

Both papers are published in the conference proceedings:

Dabah, A., Herten, A. (2026). Efficient Accelerated Graph Edit Distance Computation on GPU. In: Neumann, P., Puma, M.J., Lees, M.H., Groen, D., Dongarra, J.J., Sloot, P.M.A. (eds) Computational Science – ICCS 2026. ICCS 2026. Lecture Notes in Computer Science, vol 16783. Springer, Cham. https://doi.org/10.1007/978-3-032-29921-5_2

John, C.M., Lunet, T., Götschel, S., Herten, A., Kesselheim, S., Ruprecht, D. (2026). Fourier Neural Operators for Rayleigh–Bénard Convection. In: Neumann, P., Puma, M.J., Lees, M.H., Groen, D., Dongarra, J.J., Sloot, P.M.A. (eds) Computational Science – ICCS 2026. ICCS 2026. Lecture Notes in Computer Science, vol 16784. Springer, Cham. https://doi.org/10.1007/978-3-032-29924-6_40

Contact: Adel Dabah & Chelsea John