Heat development collaboration

Heat is a generic framework that can be used seamlessly within the Python array ecosystem. It focuses on HPC-optimized linear algebra and machine learning, made accessible to expert and non-expert researchers via the library’s numpy-like API.

Core partners of the collaboration are Research Centre Jülich / Jülich Supercomputing Centre (JSC), German Aerospace Center (DLR) and Kalrsruhe Institute of Technology (KIT).

Topics

• Machine Learning
• Large-Scale Data Processing and Analysis
• GPU acceleration
• High-Performance Computing
• Parallel Computing
• Python

Projects

• Helmholtz Analytics Framework (HAF)
• ParFlowDiagnostics (FZJ IBG-3)
• ESAPCA (funded by the European Space Agency)
• Scalable Radio Frequency Interference Mitigation for the SKA-MPI radiotelescope (with MPIfR)
• Scaling and accelerating the Spectral Clustering for Interstellar Molecular Emission Segmentation module (SCIMES) for large datasets (with Argelander Institut, Uni Bonn)
• Anomaly detection in massive datasets within the RESIKOAST project (DLR)
• (PLANNED) Support parallel I/O for AtmoRep (FZJ JSC)

Contact

Dr. Claudia Comito

• Institute for Advanced Simulation (IAS)
• Jülich Supercomputing Centre (JSC)

Building 16.3 /
Room 224

+49 2461/61-9719

E-Mail